Product Updates - Jelou AI Docs

Jelou Agent 2: unified agent, live workspace panel, and approvals

The separate Build and Insights modes are gone. Instead, the Brain Studio home view now has a single unified composer: you write what you need and, if your request needs to act on a workflow, the agent asks you which one — or which project — you want to work on.

No pre-flight selectors: you no longer pick a project or a workflow before typing; the agent works across your whole company and only asks for the target when the request requires one.
Attachments, history, and memories in the composer itself: attach files with the paperclip, open past chats from the clock icon, and manage what the agent remembers from the settings icon.
Full-screen chat: when you send a prompt, the view turns into a chat with the agent that you can collapse to see the workflow directly on the Brain Studio canvas.
Explicit approvals scoped by conversation, project, or company before running sensitive operations.
Templates and Recent workflows tabs below the composer to start from a curated template or return to your recent projects.

Read the docs

The composer, the workspace panel, decisions, approvals, and templates.

New version of Metrics: dashboards, insights, and custom metrics

Metrics has a new version. It stays in the sidebar and brings the project’s analytics into a single place, with the same global filter bar for every chart: date, channel, team, provider, currency, environment, biometric type, criterion, workflow, node, and workflow name.The General Summary dashboard ships preloaded with KPIs for active users, workflows, conversations, and HSM, plus the volume over time, the user flow, workflow performance, and top words by workflow. You can create your own dashboards (up to 10 metrics each) from templates — Operators and Tickets, Biometrics, Ecommerce, Brain Studio, Payments, AI consumption, and Outbound Messaging — or picking from the catalog by category. You can also register custom metrics based on events (up to 5 events compared on the same chart, with breakdowns and filters) or on your databases (count, aggregate, and funnel templates, no SQL).Every chart can be shared with a public read-only link or exported as a cURL ready to consume from your backend. Insights, generated through the Jelou Agent, capture the project’s evolution as an HTML document with text, tables, and embedded charts you can print or download.

Metrics

Explore dashboards, insights, and how to register custom metrics.

New URL Message node for Facebook and Instagram

You can now send a message with up to three buttons that open a link on Facebook and Instagram channels. Each button carries its own text (20 characters) and URL, which can be fixed or a variable resolved at runtime.The buttons do not branch the flow: the node has a single output. To branch based on what the user picks, keep using Buttons or List.

See the node documentation

Fields, limits, and validations.

Meta Business Agent available on the Free plan

Meta Business Agent is now available for projects on the Free plan. You can activate it on your own from Project settings → Meta Business Agent, using the same Self Service flow as every other plan: pick the WhatsApp Cloud channel, accept Meta’s AI Terms, and set up the instructions and audience.Jelou adds no extra charge for the agent on any plan. Meta bills usage directly to the credit line of your WhatsApp Business account.

Requirements and activation

Review the requirements and the step-by-step guide to activate Meta Business Agent.

Voice: agents that answer and place phone calls

You can now create voice agents that answer and place phone calls. Voice, in the side menu, brings together the agents, the call log, outbound campaigns, and event webhooks.A voice operation is built from two pieces created separately: the agent, which defines what gets said and how it sounds, and the phone number, provisioned by activating an ElevenLabs channel. The last step is connecting them with the Answering agent field in the channel configuration. An agent can exist without an assigned number, so you can design and test it before provisioning telephony.The agent editor has four tabs. In Behavior you define the system prompt — with {{variable}} dynamic variables, including {{system__time}} and {{system__time_utc}} — the first message, and whether the user can interrupt the greeting. The side panels cover the voice (voice, language, voice model filtered by language, speed, stability, similarity, and style) and the model and call (language model, maximum duration, and the time zone the agent interprets and expresses times in).Tools offers seven native telephony functions you enable with a toggle: end call, language detection, skip turn, voicemail detection, DTMF dialing, transfer to another agent, and transfer to a number. Below them you can define your own HTTP tools so the agent consumes your APIs during the conversation, with reusable authentication, headers, and body parameters. The Knowledge base tab accepts documents and URLs, and Analysis lets you define evaluation criteria and structured data fields extracted from every conversation.When provisioning the number you choose between SIP trunk — with the option to route outbound calls through your own trunk — or Twilio, either on the Jelou account searching by country and digits, or with your own credentials. Base rates are 1.15 USD/month rental, 0.0085 USD/min inbound, and 0.014 USD/min outbound.If you already have a WhatsApp number (Cloud API or Gupshup), you can enable voice calling on it without provisioning a new number, decide whether the call button shows in the chat, and connect a voice agent to it.Every call is logged with its direction, status, duration, and credits, and keeps its recording and its turn-by-turn transcript.

Go to the Voice documentation

Check the architecture and the step-by-step for creating the agent, giving it tools, and provisioning the number.

Meta Business Agent available on the free plan

You can now activate Meta Business Agent on your own from Project settings → Meta Business Agent, available on the free plan. Activation previously required our team’s involvement. Meta Business Agent is the AI that Meta runs natively inside WhatsApp; by connecting your Jelou workflows and tools, it stops merely replying and starts resolving.

Highlights

Per-channel activation: pick the WhatsApp Cloud number and activate it from Project settings, without depending on our team.
Agent instructions: up to 20,000 characters to define tone, rules and use cases.
Controlled audience: allowlist only to test without risk, or everyone once you have validated the behavior.
Connected tools: expose your Jelou workflows and tools so the agent can check an order, verify an identity or generate a payment request.
Operator handoff: when the customer asks to talk to a person, the conversation passes to Jelou with full context.

How to use it

Confirm the channel is WhatsApp Cloud. Meta Business Agent only works with WhatsApp Cloud: if the project uses another provider (for example Gupshup), that number cannot run it and you need to add a WhatsApp Cloud channel.
Go to Project settings → Meta Business Agent and choose the channel.
Accept Meta’s AI Terms. This is a one-time step that Jelou opens for you.
Address the warning about the standby webhook subscription. Without it the agent replies, but Jelou does not store the conversation: you are left with no history and no context for the operator.
Write the instructions, leave the audience on allowlist only and test with controlled numbers before opening it to everyone.

Meta Business Agent carries no additional charge from Jelou. Meta bills the usage directly to the credit line of the WhatsApp Business account, according to its rate. Meta does not expose the per-conversation breakdown yet, so the dashboard cannot show that cost today; once Meta publishes it, it will appear in the panel.

What it is and when to use it

What Meta Business Agent does and how it differs from Jelou’s AI Agent.

Requirements and activation

WhatsApp Cloud, Meta’s AI Terms and the standby webhook.

Configuration

Instructions, audience, connected tools and operator handoff.

Costs

How the agent’s usage is billed.

Project map moved to Project settings → Workflows

The Project map now lives inside Project settings as a new Workflows section, next to General and Meta Business Agent. The previous route /brain/projects/:projectId/workflows no longer exists; the new one is /brain/:projectId/settings?section=workflows. The project picker persists across settings sections, so you can jump between General, MBA and Workflows without losing project context. The map and workflow list functionality is unchanged.

Go to Project map

Review how to access the map from Project settings → Workflows.

Token optimization in the AI Agent

The Tools tab of the AI Agent node now includes the Token optimization switch. When enabled, tool results are sent to the model in the TOON format (Token-Oriented Object Notation), a compact representation that removes JSON’s repetitive punctuation and reduces token usage, especially on large lists of structured data.

Go to Token optimization

Learn how the TOON format works and when to enable it.

Test mode for payment gateways

You can now install a gateway and validate charges end to end without your own account or credentials. When you install a compatible gateway, choose Try with Jelou to test immediately or Connect my account to use your own credentials.Test mode allows up to 10 tests per configuration. When you finish you can open a template with the workflow already set up. When you want to charge for real, connect your account on the same installation without rebuilding the flow.Available for Wompi, Stripe, PayPal, Mercado Pago, Kushki, Conekta, Payphone, and Nuvei, in specific countries. DeUna does not have test mode yet. You cannot change the country later; if you pick the wrong one, reinstall the gateway.

Go to test mode

See the step-by-step, countries per gateway, and how to connect your account afterwards.

`phone` is the canonical user phone field

WhatsApp no longer guarantees the phone number as a stable identifier and is moving to its own identifier, the BSUID. So that the user object keeps delivering a reliable phone number, the canonical field is now phone.

Use phone: it returns the contact’s real phone number, resolved internally even when the conversation arrives identified by BSUID.
phoneNumber is still available as an alias of the same value. It could previously return the BSUID instead of the phone number; it no longer does.
Both fields can come back as null when no phone number can be resolved for that contact, and they only appear on WhatsApp channels. Your integrations must tolerate null rather than assume a number is always there.
The value is normalized to digits only, with no +, spaces or dashes.

If your flows read {{$user.phoneNumber}}, they keep working. We recommend migrating to {{$user.phone}}, which is the field we will maintain.

Go to the User variables guide

See which contact data is available and how to use it in your flows.

New API section: Tag management

The API reference now documents how to manage your conversations’ tags from your own systems, with the complete flow on a single page.

Company catalog: GET /v1/company/{companyId}/tags returns the available tags with their id values, colors, and filters by name, language, teams, and channels.
A conversation’s tags: GET /v1/rooms/{roomId} includes the current tags in the data.tags attribute.
Assign or update: POST /v1/bots/{botId}/rooms/tags takes the roomId and the list of id values.

Keep in mind that the tags array fully replaces the conversation’s current tags: to keep the existing ones you must send them along with the new ones, and an empty array removes them all.

Go to the Tag management reference

Check the parameters, the request and response examples, and the status codes.

Multi-organization: one account across several organizations

A single account can now belong to several organizations. You still have your own, and you can now be invited to others and switch between them from the picker, without signing out or creating another account. If the invited email already exists in Jelou, the person signs in with their usual credentials.You can also leave the organizations you were invited to, and the super admin can remove members. Organizations on the FREE plan cannot invite. Available on the Builder and Business plans.

Go to the Multi-organization guide

See how to switch organizations, invite people, leave, remove members and what happens with two-factor authentication.

AI Routing native tool in the AI Agent

The AI Agent node can now hand a conversation to the right workflow when the user asks for something that agent cannot do. Enable the AI Routing tool in the node’s TOOLS section.

Automatic: the agent passes the message verbatim and the project’s AI Routing picks the destination.
Named destination: if the user says where they want to go, the handoff goes straight to that workflow.
Check destinations: the agent can list the available workflows before deciding, or offer the user a choice.

A successful handoff ends the agent’s turn: the reply comes from the destination workflow. If nothing matches, the agent keeps handling the conversation.

Go to the Native tools guide

See the parameters, argument examples and when not to use this tool.

Coexistence: support from WhatsApp Business

The new Settings → Coexistence section defines, per WhatsApp channel, how your agent behaves when a human agent replies from the WhatsApp Business app.

Pause workflow: the default workflow pauses as soon as the agent steps in. The pause can be Dynamic, restarting with every message, or Fixed time, and lasts from 15 minutes to 4 hours.
Keywords: up to 5 per channel, each mapped to a workflow, to hand control back to the AI without waiting for the pause to end.
Read receipts and typing indicator: marks the customer’s messages as read and shows the typing indicator.

Go to the Coexistence guide

See the pause types, the keyword rules, and how changes are saved.

Clone workflows into another project

You can now copy workflows from one project into another in your company without rebuilding them. In the sidebar’s Workflows section, the icon next to the + turns on clone mode: tick the workflows you want and confirm the destination project.Before copying anything, Jelou checks the selection and tells you what will happen. If something prevents it — a variable or secret that does not exist in the destination, two workflows with the same name, a node calling a workflow that is not travelling — it shows in red and you cannot continue. If there are only warnings, such as overwriting a workflow that was already there, it names which one and leaves the call to you.Each workflow is paired by name: if the destination has none with that name it is created, and if it has exactly one it is replaced, keeping the destination’s name. When you confirm, you can leave the Go to the destination project when the clone finishes checkbox ticked so Jelou takes you there once it is done. If the copy fails halfway through, everything is rolled back.Bear in mind that some nodes arrive with their place and their connections but without the resource that was tied to the source project: HSM templates and WhatsApp Flows are left unselected, and the AI Agent arrives without its knowledge. Review them before publishing in the destination.

Go to the workflow cloning guide

See the step by step, what gets copied and what can prevent the clone.

Jelou CLI: `jelou voice` now covers the whole voice service (v1.93)

jelou voice is no longer just call and number lookup: it now manages the whole voice service end to end from the CLI.

jelou voice schedules — call campaigns end to end: create, activate, watch dialling progress, list the calls placed, pause, resume, cancel. Activating asks for confirmation first and shows how many contacts will be called.
jelou voice contacts and jelou voice contacts lists — manage who a campaign calls, including building a list straight from a CSV or XLSX file.
jelou voice agents — create and tune the agent that speaks on a call (prompt, voice, language, duration cap) and feed its knowledge base from a file or a URL.
jelou voice calls outbound places a single call; jelou voice billing shows what calls cost; jelou voice reports and jelou voice webhooks cover campaign reports and event destinations.

Anything that dials a real number or spends money asks for confirmation first (activating a campaign, a test call, an outbound call, every delete); pass --yes in CI. See Voice.

New channel: Slack

You can now connect your Brain Studio agent to Slack to serve your team and your customers by direct message, without them leaving their workspace.Activation takes four values from your Slack app: the Bot User OAuth Token, the Signing Secret, the App ID, and an app configuration token. With that last token, Jelou sets the Events and Interactivity Request URLs for you, so there’s nothing to paste by hand. The token is used once, is never stored, and expires about 12 hours after you generate it. If automatic configuration fails, the channel is still created and the modal shows you the URL to paste yourself.Slack delivers events and interactions to the same route, so a single URL serves both fields. The Bot User ID is resolved automatically from the bot token and is used to ignore the bot’s own messages. The channel properties keep the Request URL available to copy at any time; the bot token and the Signing Secret stay masked.The channel works over direct messages only and supports text (3000 characters), JPEG and PNG images up to 5 MB, buttons (up to 10), lists (up to 10 options), and CTA URL.

Go to the Slack guide

Check the prerequisites, the step-by-step activation, and the channel limitations.

Project map in Manage workflows

The Manage workflows section of a project with AI Routing enabled now includes a visual map with the AI Router and every connected workflow, alongside the workflow list (same UX as the Settings → AI Routing card, with a Backup badge, per-row hide/show switch and quick links to Settings and Studio). You can filter the map by channel, hide workflows from the router with the switch or by dragging them into the side panel, filter the list between All / AI Router / Hidden, and test routing with real messages without leaving the section.Projects with no AI Routing keep landing on the classic workflow table with no change.

Go to Project map

Read the full guide with interactions, filters and map states.

Create a project from scratch or by cloning another

The project creation modal now lets you choose how to start: From scratch, the behavior you already know, with a first workflow that contains only the Start node, or Clone existing, beginning from the configuration of a project you already have.Cloning copies workflows with all their canvases, nodes, and connections, AI Routing routes, training questions, project settings, and installed integrations. Marketplace app and database nodes are now preserved and keep pointing at the same resources in your company instead of ending up empty. Human agent handoffs also keep their assignment configuration. Nodes that send a WhatsApp Flow and HSM template nodes keep their connections, but you need to pick the flow or the template again.For security, secrets are copied by name only, without their value, and AI Agent knowledge doesn’t carry over. The source project is never modified, you can only clone projects from your own company, and if cloning fails halfway through it rolls back completely.Projects created by cloning are flagged with a Cloned tag, visible in the Projects table, in the project selector, in the Studio top bar, and in the project settings. There, the information card adds two rows: Origin, with the tag, and Cloned from, with the source project’s name and its ID ready to copy. If the original was deleted, the row shows the ID alone.The WhatsApp test channel created alongside the project — both from scratch and when cloning — takes the project’s name as-is, without the Sandbox - prefix it used before.

Go to Create a project

See the full guide with everything that gets copied, the Cloned tag, and what to review before publishing.

New channel: Microsoft Teams

You can now connect your Brain Studio project to Microsoft Teams so your coworkers and customers get fast, personalized answers without leaving Microsoft’s collaboration platform.The connection runs through an Azure Bot app that you register in Azure. Activation takes three required values: the App ID (client ID), the app password (client secret), and the Tenant ID. The channel is provisioned as single-tenant only; the app type is not selectable during activation.When the channel is created, Jelou shows the messaging endpoint URL you need to paste on Microsoft’s side, either in the Teams Developer Portal (Tools → Bot management → Endpoint address) or in the Azure Portal (Settings → Configuration → Messaging endpoint). The URL stays available in the channel properties, where the client secret is masked and cannot be copied.Teams renders messages as Adaptive Cards, so the channel supports text (4096 characters), JPEG and PNG images up to 5 MB, buttons (up to 3 per message), lists (up to 10 options), and CTA URL, which is sent as Action.OpenUrl.

Go to the Microsoft Teams guide

Check the Azure prerequisites, the step-by-step activation, and the channel limitations.

Jelou CLI: docs caught up to v1.89

We refreshed the entire Jelou CLI documentation to match the real behavior of version 1.89.0, in Spanish, English, and Portuguese.Newly documented commands and behaviors that previously had no page:

jelou context — identity, project and workflow drift, model allowlist, secret names, and database inventory in one call. Referenced from Projects and channels and AI models.
jelou tool and jelou test tool — the lifecycle of reusable tools shared across an app’s projects.
jelou channels testers — whitelist phone numbers on the WhatsApp sandbox before connecting a real number. See WhatsApp.
jelou workflow node-spec and the new tools/*.json validation in jelou workflow validate --partial. See Workflows and testing.
jelou databases plans, the get-or-create behavior of collections create, and the new confirmation gates on api-keys regenerate/create. See Databases.
Metrics — new page for jelou metrics list/get/dashboard: catalog by category, per-metric filters, time windows (--last, --period, custom range), and official dashboards via --template.

We also fixed examples that no longer matched the real CLI: trigger events (insert → create), jelou shop products images add (uses --file and a SKU, not --url or a product id), jelou shop coupons show/uses (by code, not id), jelou shop imports upload (flags, not a positional argument), the skill-install flags (--method, not --symlink/--copy), and the note claiming jelou functions tokens create was deprecated — it isn’t anymore.

Business Scoped User ID (BSUID) support

Jelou now supports the Business Scoped User ID (BSUID), the new unique identifier Meta uses to represent users within WhatsApp Business.With the arrival of Usernames, some users will stop sharing their phone number with businesses. In those cases, the BSUID is the identifier that lets you keep sending and receiving messages without disrupting the conversation. No additional configuration is required — the platform supports both identifiers.If your business has external integrations or custom developments that consume data from Jelou, review them to make sure they can process the BSUID as the user identifier.

Go to Business Scoped User ID (BSUID)

Read the full guide on the new identifier and its rollout timeline.

Sending templates via BSUID

WhatsApp template (HSM) sending now accepts a BSUID (Business-Scoped User ID) in destinations, in addition to the traditional phone number. This applies to individual sends, bulk CSV sends, and the WhatsApp Template node in Brain Studio, letting you send templates to users who enabled number privacy in WhatsApp using the identifier Meta assigns them instead of their real phone number.

Go to Individual Send (1-1)

Check out the full guide, including the BSUID format and restrictions for authentication templates.

Jelou CLI: new sections and full command coverage

The Jelou CLI documentation is now its own top-level section, organized by product area: Get started, Projects, Data, Channels, AI and voice, Commerce, and Reference.Four pages were added for the commands that were missing:

Connect — list operators and teams to route the human-handoff nodes in your workflows.
AI models — discover and validate the models available to the AI Agent and AI_TASK nodes, with exit codes built for CI.
Voice — look up calls, download recordings, export to CSV or XLSX, and review your organization’s numbers.
Shop — manage catalog, branches, categories, coupons, orders, imports, and external catalog syncs.

Functions: deferred runs, webhook verification and template validation

Jelou Functions now documents three new capabilities:

Deferred runs: book a future run of your function with ctx.jelou.schedule or with HTTP headers — 24-hour reminders, abandoned-cart follow-ups, post-purchase surveys. Includes lookup, bulk cancellation and ctx.guard to check before acting.
Webhook verification: ctx.verifyStripe, verifyShopify, verifyMeta and verifyHmac validate a webhook signature in one line.
WhatsApp template validation: ctx.templateRegistry lets you confirm the name, language and param count before calling Meta.

On the CLI side we documented jelou functions check (validates the project before deploying), jelou functions defer list/get, jelou functions deployments build (why a deploy failed) and the --dry-run, --agent and --describe flags. jelou skill install is now jelou agent install.

Deferred runs

Book, look up, cancel and check before acting.

Webhooks

Stripe, Shopify, Meta and generic HMAC signatures.

Custom Channel

You can now connect your own application to Brain Studio flows with the Custom channel: configure your webhook with its corresponding authentication in Jelou, send interactions via API, and receive Agent replies in your system. Learn which nodes and tools you can start using with this channel.

Activation guide

Enable the Custom channel in the builder.

Custom Channel API

Endpoints, response formats, and HMAC signature.

Custom payment gateways

You can now connect any payment provider (PSP) that isn’t in the provider catalog by describing its integration declaratively from Payments: credentials, charge request, webhook, and return. With no backend code per new provider, the gateway becomes available to the Payment node in Brain Studio.

Go to Custom gateway

Review the architecture, the setup wizard, the available placeholders, and how to test it before publishing.

Tool composition

You can now use the Tool node inside other Tools. The Tools picker now appears right on the canvas — drag in the one you need and you’re done. Available to all companies.This feature includes:

Anti-recursion safeguards: a Tool can’t reference itself, and a runtime nesting limit prevents infinite loops (e.g. A → B → A). When you publish, an analyzer checks for cycles.
Full traceability: a child Tool’s executionId is now always recorded, even if the execution times out. From the parent, you can debug its children directly.

Go to Tool composition

See the full guide for nesting Tools, safeguards, and debugging.

New payment integrations: Kushki and Conekta

Documentation was added to connect Kushki (Mexico, Colombia, Peru, Chile, and Ecuador) and Conekta (Mexico) in Brain Studio, and collect payments via embedded WebView checkout.

Kushki

Review how it works, configure credentials, and use the payment node in Canvas.

Conekta

Review how it works, configure credentials, and use the payment node in Canvas.

Evaluation conditions in the AI Agent node

The AI Agent node now lets you evaluate its conversations. In the Events tab, define conditions through a prompt; when the agent ends a conversation, it analyzes the history and fires an event for every condition that is met, ready to analyze in your metrics.Key features:

Prompt-based conditions — A name (up to 60 characters) and an instruction (up to 900 characters) per condition.
One event per met condition — The condition’s name defines the name of the event that fires.
Per-condition toggle — Enable or disable each condition without deleting it.
Up to 10 conditions per agent, with search, editing, and deletion from the same tab.

Go to Evaluation conditions

See the full guide in the AI Agent node.

Business Profile

You can now configure your WhatsApp Business Profile directly from Settings: profile photo, name, category, email, website, description, and address. This section is currently available to all companies.

Go to Business Profile

See the full guide to edit your WhatsApp channel’s profile.

Operator Unavailable Reason

Now, when switching their status to Unavailable, operators can share a reason — free text plus an emoji, or one of 7 default reasons (In a meeting, Lunch, Administrative work, Technical issue, Medical leave, Active break, End of shift). The reason is visible to their team and administrators in the operator’s header, the connection history, and operator monitoring.

Go to Unavailable reasons

See the full guide to enable and configure this feature.

Campaign name is now only requested when it applies

We adjusted the campaign (HSM) creation flow: the Campaign name field is no longer requested in the General step. It now appears in the Recipients step, and only when you choose to send to Multiple contacts.Previously, the name was requested regardless of the number of recipients, even when sending to a single contact — a detail that could cause confusion when reviewing send records.If you choose Multiple contacts, the name field appears right below, to fill in before continuing. If you choose Single contact, the field isn’t shown, and a notice indicates that campaign won’t appear in the Campaigns module — its information will only be available in Reports.

Campaign Resending

You can now resend an existing campaign without setting it up from scratch — ideal for recurring campaigns where only the contact list changes.From the actions menu of a Completed or Cancelled campaign, select Resend, enter a name for the new campaign, and upload the new recipients file.The new campaign automatically inherits the original’s configuration —template, parameter mapping, response type, and workflow associations—, shown as a read-only summary both during creation and in the confirmation preview. The uploaded CSV is validated against the original campaign’s expected columns, and parameter mapping is preserved even if you remove and re-upload the file. Resent campaigns are marked with a Resent badge in the monitoring table, referencing the original campaign.

Client Name Editing in PMA

Operators, supervisors, and administrators can now edit a client’s name directly from the PMA interface, without navigating to the Clients (CRM) module.This improvement addresses common cases during support: incorrect names, client-requested updates, incomplete automatic capture, or contacts that only display a phone number. Editing is available both in active conversations and archived cases, and is reflected in the conversation’s PDF download.

Public Channel Interactions in Monitoring

We added a new section to the Monitoring module to view interactions from public channels —such as Facebook Feed, Instagram Feed, and Twitter Replies—, complementing the existing views for private cases and email.This section only appears if the company has at least one public channel configured, so the view stays clean and relevant for each customer. The columns in the new table follow the same structure you already know from private cases and email, for a consistent experience. Downloads are generated from the Download Center, in Excel or CSV format.

Go to Conversation History

See the full guide to the Monitoring module.

Node Events

You can now add an Events tab to your node configuration (HTTP, SOAP, Code, Tool, Conditional, Skill, AI Agent, and others) to register tracking events every time a node runs. Define a name and up to 5 custom properties per event, to feed your metrics and performance dashboards.

Go to Node Events

See the full guide to configuring tracking events on your nodes.

Memory management in the AI Agent

We added a new group of native tools for Memory management in the AI Agent node’s Context tab. With per-source toggles, the agent can save and recall information (per-user working memory), read and update contact data with the Inbox CRM, share variables between nodes with $context, and inject user, channel, and company information as context. We also added the native connection to Databases (Datum v2) in the Tools tab.

Go to Memory management

See the full guide in the AI Agent node’s Context tab.

Memory management in the API

Use the new API section to read and write per-user session memory through the Memory API v2. Retrieve all of a user’s records, fetch a single record by key, or save one or several records by user reference. The endpoints authenticate with an API key (x-api-key) through gateway.jelou.ai and require the memory:read or memory:write scopes.

Go to Memory Management

See the full memory resources guide.

Send messages via BSUID

The message sending endpoint now accepts a BSUID (Business-Scoped User ID) in userId, in addition to the traditional phone number. This lets you reply to or proactively message users who enabled number privacy in WhatsApp, using the identifier Meta assigns them instead of their real phone number.

Go to Send Messages

See the full guide, with use cases and examples.

Operator statistics in the API

We added a new API resource to query your operators’ attention statistics for a date range: total cases, attended, unattended, transferred, and average response times. Filter by a specific operatorId or query the full range for a team-wide overview.

Go to Operator statistics

See the full resource guide.

Genesys Integration

Jelou now integrates with Genesys Cloud to enable interoperability between both platforms. You can transfer conversations, escalate to Genesys agents, and centralize customer support using Open Messaging, Triggers, Workflows, and Data Actions.

Go to Genesys Integration

See the full integration guide.

Bulk product import via Excel

You can now load your Jelou Shop catalog by uploading an Excel file. Download a guided template —with an instructions sheet, headers colored by whether they’re required, and comments on each column— and choose between a simple upload (with an automatically generated SKU) or a complete upload by enabling optional sheets for variations, features and modifiers.See the Import products guide.

WooCommerce connection with OAuth

Connecting your WooCommerce store is now a simple OAuth flow. You no longer need to manually generate the Consumer key or Consumer secret: just enter your store domain, sign in to WooCommerce, and approve access. Jelou creates the credentials for you.Check the WooCommerce guide.

WhatsApp Username

The guide covers:

What a username is and how the @ identifier works.
Benefits for your business privacy and visibility.
Steps to create and configure your username from WhatsApp Manager.
Important considerations about availability and restrictions.

Go to WhatsApp Username

Check the full guide to set up your username.

Workflow templates documentation

We updated the documentation on workflow templates in Brain Studio: how to create and share them with the Jelou Agent, install them in other accounts, and browse the catalog from the home view.See the Jelou Agent and Your first workflow guides.

Promotions and coupons in Jelou Shop

Jelou Shop now includes a Promotions module to create and manage discount coupons, available in the Shop Admin panel and via API.Key changes:

Percentage or fixed-amount coupons — Define the discount as a percentage or a fixed value.
Validity — Set start and end dates, or leave it with no expiration.
Per-branch scope — Apply the coupon to all branches or only the ones you choose, with a per-branch usage limit.
Usage limits — Set a total redemption cap and an option for one use per customer.
Automatic states — Each coupon shows its state (Active, Scheduled, Expired, Depleted, or Inactive) computed from its validity and usage.
The discount is validated and applied at checkout; an invalid, expired, or out-of-branch coupon is rejected with a clear message.

See the Shop Admin → Promotions guide and the Promotions API.

Configurable follow-up message in the AI Agent node

The follow-up message in the AI Agent node is now configurable. When you turn on the toggle under Advanced, a button appears that opens a modal to define how the agent resumes an inactive conversation.Key changes:

Manual and Automatic modes — In Manual you write the instruction that guides the reminder (up to 900 characters); in Automatic the agent writes it on its own.
Wait between wakeups — Sets the idle time before each reminder (from 5 to 120 minutes).
Stop after # attempts without reply — Limits consecutive reminders without a reply (from 1 to 5, default 2).
The options automatically adjust to the agent’s expiration time.

Settings module update

We continue improving the platform to deliver a simpler, more organized, and intuitive experience. Here are the highlights from the Settings module update.Key changes:

Revamped navigation system — Settings are now organized into two main sections: Organization and Product, making it easier to find what you need and reducing option overload.
New “Advanced settings” section — We centralized channel-specific settings, team configurations, and other global options, enabling clearer and more efficient administration.
Time traffic-light improvements — We updated the visualization and management of time indicators to facilitate conversation tracking and improve operational control.

These improvements aim to make platform administration faster, more intuitive, and aligned with team needs.

Monitoring module update

The Monitoring module has been updated with improved navigation, a new interface, and optimized label management.Key changes:

Faster and more organized navigation — Simplified access to the most relevant information, with an improved structure for greater efficiency.
Consistency with the new platform UI — A unified and modern experience across all modules, maintaining visual coherence.
Optimized label management — Label creation has been relocated to the Customers module and will be available directly from the Multi-Agent Panel for greater accessibility.

These changes are part of our ongoing commitment to improving your experience on Jelou.

New documentation pages in Connect — Management

Five new pages were added to the Management section within the Connect tab, expanding the documentation available for supervisors and administrators:

Live Monitoring: monitor active conversations, response times, and cases requiring immediate action in real time.
Operator Supervision: check the status, workload, and activity of each operator, with actions such as impersonation and conversation transfer.
Consolidated Case Management: analyze all cases managed within a period, with support indicators and classification by routing type.
Conversation History: review and audit past interactions with full traceability of messages, events, and transfers.
Quick Replies: create and manage predefined responses to speed up operator support.

Go to Management

Explore the new management sections in Connect.

Inbox Emails — New design available

We completely revamped the Emails module: the experience now feels like a modern email client, with everything more organized, clean, and easy to use.What’s new:

View preferences: choose between the new panel view or the full-screen view with the classic layout.
Split and resizable view: inbox, email, and customer CRM on a single screen.
Refreshed look and feel.
Same functionality, now enhanced and better organized.

Label filter in the chat inbox

A new filter is now available in Inbox → Chats that lets you search conversations by the labels assigned to each chat, quickly and dynamically.

The filter automatically populates from the labels on your active chats.
If none of your chats have labels, the filter will not appear.
Especially useful when handling multiple conversations and you need to quickly find those of a specific type.

New WhatsApp template sending sections

In the Customer profile section and from inbox conversations, you will now see a new icon that displays the list of available templates to send to your customers, both within and outside an active session.This feature is especially useful for re-engaging users. Additionally, you can trigger a workflow that will start automatically when the customer replies to the received message.

Considerations:

The button is only shown for roles with campaign sending permissions.
Templates with QR codes and special templates (LTO, Catalog, Flows, Carousel, etc.) are temporarily excluded.
The selected workflow runs only when the user is not in the inbox.

Connect tab update

We updated the Connect tab with a reorganized structure for customer support operations and management. Key changes:

Reorganized sections: Getting Started, Outbound Messaging, Management, Channels, and Settings.
Unified introduction in Brain Studio presenting the complete platform: builder, human support, and outbound messaging.
Renamed groups: “Campaigns” is now “Outbound Messaging”, “Multi Agent Panel” is now “Management”, “Closure reason” is now “Workflow routing”.

Introduction

Discover the platform from the new home page.

Connect

Explore the new customer support and operations tab.

Ecommerce API — Migration to the gateway

The Jelou Shop API is now consumed through the Jelou gateway. The base URL changes to https://gateway.jelou.ai/ecommerce and authentication is done with an API key sent in the x-api-key header (instead of Authorization: Bearer).Generate and manage your API key from the API Keys module. The app_id (Shop ID) remains unchanged.

View ecommerce API

Check the updated endpoints for branches, categories, and products.

Shopify — OAuth 2.0 connection

Connecting your Shopify store now uses OAuth 2.0. You no longer need to create an app in Shopify or enter your store domain, Client ID, or Client Secret. From the Marketplace (or the Jelou Agent) you click Connect, authorize the Jelou Commerce app in your Shopify account, and the integration is ready.

View the Shopify guide

Follow the steps to connect your store with the new authorization flow.

Workflow Auto-publish

New self-service accounts have auto-publish active by default: every change saved in the canvas deploys immediately to production without clicking Publish. The button disappears from the editor while the feature is active.Existing accounts (enterprise and self-service) keep the manual publish flow. Super admins can enable or disable auto-publish from the context menu (upper right corner).Keep in mind: with auto-publish active, no version history is generated and rollback is not available. It only applies to regular workflows — Marketplace Tools and Skills always require manual publication.

Publish and versions

Learn how to configure your account’s publishing mode.

AI Routing — Takeover

New card on the AI Routing page in Settings that defines when the AI Router takes control of the conversation. Especially useful for projects where users open with a greeting before stating their actual request.Choose between two modes:

Immediate routing: the AI Router attempts to route the user from their very first message.
Filter greetings: ignores short courtesy messages and waits to detect a real intent before routing. Configure the wait time (in seconds) and the message sent when that time expires.

View project-level configuration

See how to configure the takeover strategy.

AI Routing — Migrate from legacy intents

Projects with legacy intents now see a Migrate project button on the AI Routing page in Settings. Clicking it shows the full list of workflows with the number of intents each one has, plus a slider to set the rollout percentage toward AI Routing in a controlled way.

How to migrate your project

Follow the steps to migrate from legacy intents to AI Routing with a controlled rollout.

AI Routing

AI Routing automatically routes each incoming message to the most suitable workflow in the project, using semantic comparison instead of exact keywords. When no workflow matches, the message is sent to the fallback workflow you define.

Key updates

Semantic routing: the system compares the message against the name and description of every workflow and picks the best match without manual rules.
Project-level configuration: enable or disable AI Routing per project, set a fallback workflow, and control the rollout strategy (percentage of traffic between AI Routing and legacy intents).
Workflow-level configuration: from the Start node, each workflow can define its name, description, and advanced execution rules.
Advanced rules: restrict a workflow to a user allowlist or specific country prefixes; or hide it from automatic routing so it only runs via a forced redirect.
Test panel: evaluate single messages, batches, or bulk uploads (CSV) and verify which workflow they would route to before publishing.

See full guide

Project-level config, workflow-level config, advanced rules, and test panel.

HubSpot Integration

You can now connect Jelou with HubSpot directly from the Brain module. The integration enables managing conversations, automating conversational flows, and transferring customer support to HubSpot Inbox or custom channels.

Configurable from the HubSpot node in Canvas.
Supports HubSpot Inbox and custom channels.
Requires an Enterprise account with the HubSpot node enabled.

Setup guide

See the full documentation to install, configure, and use the integration.

PayPal and Wompi available as payment providers

Brain Studio now supports connecting PayPal and Wompi as payment providers from Marketplace, Jelou Agent, the Payments node in Canvas, or compatible templates.Both providers use a WebView embedded checkout experience within the conversational flow.

PayPal — Global coverage. Requires a Business account with API access. Configure with Client Id and Client Secret.
Wompi — Available in Colombia (COP). Requires an active Wompi account with Sandbox or Production keys and secrets.

Connect PayPal

Getting started guide for connecting PayPal in Brain Studio.

Connect Wompi

Getting started guide for connecting Wompi in Brain Studio.

WhatsApp Carousel in messages API

You can now send interactive carousel messages through the POST /v1/bots/{botId}/messages endpoint, without requiring an HSM template approved by Meta.

Highlights

Up to 10 cards per carousel, each with image or video header, body text and action.
Supported actions: CTA URL button (cta_url) or quick reply buttons (quick_reply).
Compatible providers: WhatsApp Cloud API and Gupshup CAPI.

Carousel documentation

Body structure, examples and restrictions of the new message type.

Jelou Connect Mobile App

We have made significant improvements to optimize your app experience. Key changes include:

New interface: A more modern, simple, and easy-to-use design.
Light and dark mode: Adapt the app to your visual preference.
Integrated media capture: Take and send photos and videos directly from the app.
Enhanced security: Google sign-in (SSO) and two-factor authentication (email OTP or TOTP apps).
Improved audio playback: Listen to audio at 1.5x and 2x speed to save time.

Google Play Store

Download Jelou Connect for Android.

Apple App Store

Download Jelou Connect for iOS.

Workflow Tester with unified look

We revamped the Brain Studio canvas Tester to share its interface with the production Debugger. Builders no longer need to learn two different tools: bubbles, node indicators, and the detail panel are identical on both sides.

Key updates

Node indicators interleaved with messages: every executed node appears as a card with its canvas icon and name. Click opens the detail panel with input, output, initial state, final state, and errors.
Messages view / Execution view: toggle at the bottom of the panel to switch between conversation only and conversation + indicators.
Contextual actions: “Go to node” centers the canvas; “Debug Tool” opens the Tool’s internal execution with its logs.
Copy JSON path: clicking any key of the initial/final state copies the path in bracket notation (e.g., finalState.tool.url) to the clipboard.
Clickable interactive bubbles: quick replies, lists, CTA URLs, and carousels are now simulated with a direct click, just like on WhatsApp.
Sub-workflows visually identified: inner sub-workflow nodes render as a separate group to mark the execution boundary.

The execution engine does not change. Available to every plan and user with no migration or feature flag.

See full guide

How to use the Tester, read indicators, and debug step by step.

New Jelou Shop UI

Jelou Shop launches a redesigned interface at apps.jelou.ai/shop-v2/products that brings products, categories, branches, and settings together in a single panel. The product detail now includes a live preview of the webview, collapsible-block forms, and new customization options (variations, modifiers, attributes, and customer notes).The previous version is still available through the Need the previous version? link in the sidebar.

View documentation

Visual walkthrough of the admin panel.

Password change with 2FA verification

The password change flow no longer requires the current password. Now, after entering the new password and its confirmation, the platform verifies the user’s identity through two-factor authentication (2FA), requesting a verification code via email (OTP) or through an authentication app (TOTP).

View documentation

See the profile management guide.

Multi-message response in AI Agent node

The AI Agent node can now split its response into separate bubbles when it contains paragraph breaks (\n\n), media URLs, or lists. This produces more natural conversations on WhatsApp and other channels: introductory text appears first, the image arrives as a separate message, and the follow-up question with buttons is delivered at the end.

How it works

Image, video, audio, or document URLs are sent as independent media bubbles according to the channel.
Numbered lists (1., 2., 3.) and single line breaks are kept together in the same bubble.

Activation

Enabled by default for agents created after April 24, 2026.
Earlier agents keep the previous behavior (single bubble) until you enable the Multi-message response toggle in the Advanced tab.

View documentation

See the AI Agent node guide.

New External Support Panel documentation

Complete documentation for integrating your existing customer support panel with Jelou. Allows external providers to manage human support conversations using webhooks and Jelou’s REST API.

Content included

Integration process: Step-by-step guide with HMAC-SHA256 signature verification for secure bidirectional communication.
Configuration: Endpoints to create, get, update, delete integrations and regenerate the signing secret.
Webhook events: Common event structure and reference for all event types (conversation.start, conversation.close, message).
Messages: Receiving end-user messages and sending text and multimedia responses.
Conversations: Endpoints to start, close and query conversations, with documentation of the webhook event associated with each operation.

Availability

Documentation available in Spanish, English and Portuguese.

See External Support Panel documentation

See the complete External Support Panel integration guide.

New Context tab in AI Agent node

The third tab of the AI Agent node is now called Context (previously Knowledge) and centralizes all agent input configuration in one place.

What the new Context tab includes

Initial message: a selector with three options to control how each conversation starts:
- Pass the last message — the agent receives what the user wrote (default option).
- No user message — ideal for HSM campaigns, automated workflows, or post-data collection.
- Custom — text field with a variable selector for specific contexts.
Remember previous messages: defines how many previous messages to include as context each session (range: 1–50 messages).
Unified upload block: upload documents and images from the same block, with support for drag-and-drop, click, or pasting a URL.
External context: allows resuming the node from an external system via the resume API. Shows the endpoint, required headers, and a payload example (previously called External resume in the Advanced tab).

Changes in the Advanced tab

Save response moved to the top of the Advanced tab.
External resume was renamed to External context and moved to the Context tab.
Follow-up message added to keep the conversation active if the user does not respond.

View documentation

See the complete AI Agent node guide with the new tab structure.

Profile Management

We revamped the profile management experience to make it simpler, smoother, and aligned with the platform.

Key updates

New profile interface: More intuitive, modern design consistent with the rest of the platform.
Notification management: Enable or disable notifications and configure sound preferences.
Automatic login: Configure whether to automatically connect when logging in.
Password change: Update your password quickly and securely.
Two-factor authentication (2FA): Choose your preferred method for added security (email or authentication apps).
Login history: View recent account access for greater control.

View documentation

See the complete profile management guide.

WhatsApp Template node improvements

The WhatsApp Template node in Brain Studio received improvements to make it easier to use and configure. It now features a clearer step-by-step flow that includes template selection, recipients, media, dynamic parameters, and response.

View documentation

See the complete WhatsApp Template node guide.

New Databases node in Brain Studio

You can now read, create, update, and delete records from your Datum databases directly on the canvas — no code required. The node’s panel builds the HTTP request for you based on the database, collection, and operation you pick.

What this change includes

Five operations: list, get, create, update, and delete records.
Per-field type editor: text, number, bool, date, email, url, select (single and multi), rich text, and json — each field with the right control and variable support.
Visual filter and sort builders: operators (=, !=, >, >=, <, <=, ~, !~), AND rules, and pagination.
In-panel testing: run the request and inspect the response (status, body, headers) without leaving the node; the result is cached until you change the configuration.
Direct Datum shortcut: an Open button next to the selector launches the chosen database in a new tab of the Datum app.

Read the full guide

How to configure the node, per-operation examples, and how to access the response.

Webhook node revamp

The Webhook node received a comprehensive update that significantly expands its capabilities. Key changes include:

New UI: Redesigned configuration interface, aligned with the style of other nodes.
New executions without executionId: You can now use the webhook to start new executions without passing an executionId.
New $webhook variable: Similar to $memory and $context, it exposes body, headers, query, method, mode, executionId, and isNewExecution from the incoming request.
Configurable multiple resumes: oneTimeOnly option to control whether an execution can be resumed more than once.
Webhook in Workflows: The Webhook node is now available in Workflows, enabling conversational flows to be started and resumed from external events.
Testing with /test: Dedicated URL to test the draft without affecting the published version.
Full backward compatibility: Existing webhooks continue to work without changes.

View documentation

See the complete Webhook node guide.

Documentation: workflows and the previous skill name

We updated documentation copy to use workflows as the primary term. Where it helps you find your way, we use workflow (skill) so you can still map the previous product name.

What this change includes

Guides and visible text aligned with the new wording (for example, Your first workflow (skill) on introduction and getting-started pages).
No URL changes: page paths and internal links stay the same; only on-screen copy was updated.
Technical references: in API contexts or exact system names, skill is kept when it matches the real identifier.

Connect channels directly from Brain Studio

You can now connect a WhatsApp channel without leaving Brain Studio. The connection flow is available from two points within the canvas:

Publish → Connect your channel: click Publish and then the Connect your channel link at the bottom of the side panel.
Manage channels: open the menu in the top-right corner of the canvas and select Manage channels.

Since you are already inside a project, the modal skips directly to the channel name, Facebook authentication, and terms acceptance steps — no need to choose an integration or project.

View full WhatsApp guide

See the detailed steps to connect your WhatsApp Business number.

New Campaigns UI in Connect

The campaign creation and management experience in Connect has been completely redesigned. You can now complete the entire flow in fewer steps, with clearer navigation that is consistent with the rest of the platform.

What improves with this update

Faster, more organized navigation: the creation wizard guides each stage sequentially and without distractions.
Fewer steps to create campaigns: the flow was simplified without removing any functionality.
Greater clarity at every stage: more explicit names, descriptions, and controls in each step.
Consistency with the new platform UI: visual design aligned with the rest of Connect and Brain Studio.

Default parameters for MCP tools

MCP tools in granular mode now support configuring default parameters from the AI Agent and AI Task nodes. When you open an individual tool’s detail view, you can:

Set default values that the agent will use on every invocation, instead of letting the AI decide.
Inject dynamic flow variables using the variable selector.
Toggle between form view and JSON editor.
Search parameters by name.

Each parameter displays a Manual (value set by you) or Auto (agent decides) indicator.

View documentation

See how to configure default parameters in the AI Agent node.

Increased memory JSON limit

The maximum allowed size for JSON objects stored with $memory.setJson() is 15 KB.

Save response in interactive message nodes

The Buttons, List, and Numbered list nodes now allow saving the option selected by the user directly into a memory variable from the Advanced tab in the configuration panel.

Main highlights

Response variable in Buttons: Enable the “Save response” toggle and enter the variable name. The payload of the chosen button becomes available as {{$memory.variable_name}}.
Response variable in List: Same functionality for the List node. The selected option is stored for later use in the flow.
Response variable in Numbered list: Full support in the new Numbered list node, available on all channels.
New Numbered list node: Interactive message node that presents options as numbered text, compatible with WhatsApp, Web, Facebook, and Instagram.

How to use it

Open the configuration panel of a Buttons, List, or Numbered list node.
Go to the Advanced tab.
Enable the Save response toggle.
Enter the name of the variable where the selection will be stored.
Access the value in subsequent nodes with {{$memory.variable_name}}.

Buttons node

Configure the response variable in interactive buttons

List node

Configure the response variable in dropdown lists

Numbered list

New multi-channel node with response variable support

New channel activation documentation

Complete documentation on how to activate and configure communication channels on the Jelou platform.

Documented channels

Facebook: Connect your Facebook page to manage conversations
Web: Integrate a chat widget on your website
WhatsApp: Set up WhatsApp Business API for customer support

Included content

Prerequisites for each channel
Basic step-by-step configuration
Validations and restrictions
Channel-specific best practices

View activation guides

Check out the complete channel activation guides.

UX improvements in Random node

New improved interface for the Random node with better user experience and visual validation.

Main improvements

Improved empty state: Clear interface when no routes are configured with a direct action button
Visual validation: Color alerts indicating whether the percentage distribution adds up to 100%
Better organization: Redesigned configuration panel to make route management easier
Real-time feedback: Visual indicators of the percentage distribution status

View documentation

Check out the complete guide for the Random node.

New Workflows & Tools interface in Brain Studio

Improved user experience with a new design for accessing Workflows, Tools, and Marketplace.

Main changes

Floating button: Replaces the fixed left sidebar with a floating button showing the workflow name and a folder icon
Popover panel: Clicking the floating button opens a pop-up panel with access to Workflows, Tools, and Marketplace
Cleaner interface: More workspace by removing the permanent sidebar

View updated documentation

Check out the updated guides with the new interface.

Advanced AI Agent configuration

New documentation on the advanced configuration options available in the AI Agent node.

Documented options

Quick reply payload support: Allows processing quick reply payloads in interactive messages
PDF document support: Enables reading and processing PDF files (text only, no images)
Read image with URL and caption: Processes images along with their text descriptions

View documentation

Check out the complete guide for advanced AI Agent configuration.

Memory V2

New version of the memory system with significant improvements for the code node.

Main highlights

File support: Store images, videos, audio, and documents (up to 10MB)
TTL control: Define expiration time for each stored piece of data
Improved API: Specific methods for primitives, JSON, and files
Greater capacity: JSON up to 5KB, files up to 10MB

Quick example

// Primitives
$memory.set('language', 'en');

// JSON with 1-hour TTL
$memory.setJson('profile', { name: 'Maria' }, 3600);

// Files with 1-day TTL
await $memory.setFile('photo', base64Data, 86400, 'image/jpeg');

Full documentation

Check out the detailed guide with all methods, supported types, and migration examples.

​Jelou Agent 2: unified agent, live workspace panel, and approvals

Read the docs

​New version of Metrics: dashboards, insights, and custom metrics

Metrics

​New URL Message node for Facebook and Instagram

See the node documentation

​Meta Business Agent available on the Free plan

Requirements and activation

​Voice: agents that answer and place phone calls

Go to the Voice documentation

​Meta Business Agent available on the free plan

​Highlights

​How to use it

What it is and when to use it

Requirements and activation

Configuration

Costs

​Project map moved to Project settings → Workflows

Go to Project map

​Token optimization in the AI Agent

Go to Token optimization

​Test mode for payment gateways

Go to test mode

​phone is the canonical user phone field

Go to the User variables guide

​New API section: Tag management

Go to the Tag management reference

​Multi-organization: one account across several organizations

Go to the Multi-organization guide

​AI Routing native tool in the AI Agent

Go to the Native tools guide

​Coexistence: support from WhatsApp Business

Go to the Coexistence guide

​Clone workflows into another project

Go to the workflow cloning guide

​Jelou CLI: jelou voice now covers the whole voice service (v1.93)

​New channel: Slack

Go to the Slack guide

​Project map in Manage workflows

Go to Project map

​Create a project from scratch or by cloning another

Go to Create a project

​New channel: Microsoft Teams

Go to the Microsoft Teams guide

​Jelou CLI: docs caught up to v1.89

​Business Scoped User ID (BSUID) support

Go to Business Scoped User ID (BSUID)

​Sending templates via BSUID

Go to Individual Send (1-1)

​Jelou CLI: new sections and full command coverage

​Functions: deferred runs, webhook verification and template validation

Deferred runs

Webhooks

​Custom Channel

Activation guide

Custom Channel API

​Custom payment gateways

Go to Custom gateway

​Tool composition

Go to Tool composition

​New payment integrations: Kushki and Conekta

Kushki

Conekta

​Evaluation conditions in the AI Agent node

Go to Evaluation conditions

​Business Profile

Go to Business Profile

​Operator Unavailable Reason

Go to Unavailable reasons

​Campaign name is now only requested when it applies

​Campaign Resending

​Client Name Editing in PMA

​Public Channel Interactions in Monitoring

Go to Conversation History

​Node Events

Go to Node Events

​Memory management in the AI Agent

Go to Memory management

​Memory management in the API

Go to Memory Management

Jelou Agent 2: unified agent, live workspace panel, and approvals

New version of Metrics: dashboards, insights, and custom metrics

New URL Message node for Facebook and Instagram

Meta Business Agent available on the Free plan

Voice: agents that answer and place phone calls

Meta Business Agent available on the free plan

Highlights

How to use it

Project map moved to Project settings → Workflows

Token optimization in the AI Agent

Test mode for payment gateways

`phone` is the canonical user phone field

New API section: Tag management

Multi-organization: one account across several organizations

AI Routing native tool in the AI Agent

Coexistence: support from WhatsApp Business

Clone workflows into another project

Jelou CLI: `jelou voice` now covers the whole voice service (v1.93)

New channel: Slack

Project map in Manage workflows

Create a project from scratch or by cloning another

New channel: Microsoft Teams

Jelou CLI: docs caught up to v1.89

Business Scoped User ID (BSUID) support

Sending templates via BSUID

Jelou CLI: new sections and full command coverage

Functions: deferred runs, webhook verification and template validation

Custom Channel

Custom payment gateways

Tool composition

New payment integrations: Kushki and Conekta

Evaluation conditions in the AI Agent node

Business Profile

Operator Unavailable Reason

Campaign name is now only requested when it applies

Campaign Resending

Client Name Editing in PMA

Public Channel Interactions in Monitoring

Node Events

Memory management in the AI Agent

Memory management in the API

Send messages via BSUID

Operator statistics in the API

Genesys Integration

Bulk product import via Excel

WooCommerce connection with OAuth

WhatsApp Username

Workflow templates documentation

Promotions and coupons in Jelou Shop

Configurable follow-up message in the AI Agent node

Settings module update

Monitoring module update

New documentation pages in Connect — Management

Inbox Emails — New design available

Label filter in the chat inbox

New WhatsApp template sending sections

Connect tab update

Ecommerce API — Migration to the gateway

Shopify — OAuth 2.0 connection

Workflow Auto-publish

AI Routing — Takeover

AI Routing — Migrate from legacy intents

AI Routing

Key updates

HubSpot Integration

PayPal and Wompi available as payment providers

WhatsApp Carousel in messages API

Highlights

Jelou Connect Mobile App

Workflow Tester with unified look

Key updates

New Jelou Shop UI

Password change with 2FA verification

Multi-message response in AI Agent node

How it works

Activation

New External Support Panel documentation

Content included

Availability

New Context tab in AI Agent node