v4.11.0

[sojan-official]

This release focuses on improving how conversations are handled and resolved inside Chatwoot. We are introducing structured resolution workflows, better AI assistance in the reply composer, and UX improvements that reduce friction for agents.

✨ Captain Editor

We've improved the editor with Captain AI and a better experience for AI-assisted responses. AI reply suggestions now use related FAQs to improve response quality

Agents can now:

• Improve reply
• Change tone
• Fix grammar and spelling
• Suggest a reply
• Summarize the conversation
• Ask Copilot

All of this works directly inside the composer. The agent can review and edit everything before sending.

⚙️ Conversation Workflows

We have introduced Conversation Workflows, a dedicated section under Settings that brings all conversation related logic into one place. This is the foundation for improvement in conversation automation in Chatwoot. Over time, more rules and controls will be added here.

Here are the current options:

• Auto-resolve conversations (existing behavior)
• Required attributes on resolve (new)

📋 Required Attributes on Resolution

Admins can now enforce structured data collection before a conversation is resolved. You can select one or more existing conversation-level custom attributes (Text, Number, Link, Date, List, Checkbox) that must be filled before agents can close a conversation.

How it works

• Admin selects required attributes under Settings → Conversation Workflows.
• When an agent clicks Resolve:
  • If all required attributes already have values → resolution happens normally.
  • If any value is missing → a modal appears asking the agent to fill them.
• Resolution is blocked until all required attributes are satisfied.
• Bulk resolve follows the same rules. Conversations missing attributes will not be resolved, and a notification will indicate which ones failed.
• API-based resolution does not enforce required attributes.

This helps teams collect the right information at the time of resolution and keeps reporting clean and reliable.

🏷️ Remove Label via Right Click

You can now remove labels directly by right-clicking on them inside a conversation. This is a small improvement, but it reduces the friction when organizing conversations and helps your agents move faster.

Other improvements

• Assignment V2 updates for fairer workload distribution and team-safe routing (Enterprise)
• New report downloads and expanded reporting APIs
• Improvements to integrations across channels like Shopify, TikTok, and WhatsApp
• More reliable bot handoffs with smarter conversation state handling
• Better contact and conversation workflows, including faster label actions and smoother contact search
• Numerous bug fixes, security updates, and performance improvements

Thanks to @joao-baza, @TheDanniCraft, @mazenkhalil for the contributions.

Note: Captain v2 now supports GPT-5.2 for self-hosted deployments.

---

This discussion was created from the release v4.11.0.

[haguilarhern]

Hi, wondering if in this release the code from this branch was pushed through?

#13371

I was testing if my business accounts would start processing echo messages after I updated my instance to this latest version and I still don't see the messages sent from my device reflected in chatwoot.

Thanks

[haguilarhern]

Hi!

To answer your questions: Yes, the echo webhooks are definitely showing up in the logs. I can see the server receiving the data and Sidekiq running the job, but the messages still aren't appearing in the Chatwoot UI.

I've sanitized the personal details below, but kept the structure intact:

Rails Log (The incoming webhook):

I, [2026-02-18T20:51:23.567720 #62]  INFO -- : [d78cedae...] Parameters: {
  "object" => "whatsapp_business_account", 
  "entry" => [{
    "id" => "129232189452XXXX", 
    "changes" => [{
      "value" => {
        "messaging_product" => "whatsapp", 
        "metadata" => {"display_phone_number" => "521221180XXXX", "phone_number_id" => "32833503572XXXX"}, 
        "message_echoes" => [{
          "from" => "521221180XXXX", 
          "to" => "1709986XXXX", 
          "id" => "wamid.HBgLMTcwOTk4NjU5NjkVAgARGBQzQjY2RDcwQjg2NzExNTdENUJCQQA=", 
          "timestamp" => "1771447882", 
          "text" => {"body" => "Hola buenas tardes"}, 
          "type" => "text"
        }]
      }, 
      "field" => "smb_message_echoes"
    }]
  }]
}
I, [2026-02-18T20:51:23.569191 #62]  INFO -- : [ActiveJob] Enqueued Webhooks::WhatsappEventsJob (Job ID: 132b6a68-aaa4-4998-8666-a95c286c3843) to Sidekiq(low)
I, [2026-02-18T20:51:23.569396 #62]  INFO -- : Completed 200 OK in 2ms

Sidekiq Log (The background processing):

I, [2026-02-18T20:51:23.570104 #1]  INFO -- : [ActiveJob] [Webhooks::WhatsappEventsJob] [132b6a68-aaa4-4998-8666-a95c286c3843] Performing Webhooks::WhatsappEventsJob...
I, [2026-02-18T20:51:23.574043 #1]  INFO -- : [ActiveJob] [Webhooks::WhatsappEventsJob] [132b6a68-aaa4-4998-8666-a95c286c3843] Performed Webhooks::WhatsappEventsJob in 4.32ms

I’m not very technical, but I noticed the job completes in about 4ms, which might be a clue. I hope this helps! Let me know if you need me to look for anything else.

[sojan-official]

@muhsin-k could you take a look ?

[muhsin-k]

@haguilarhern Could you please check message created or not in DB by running following command

Message.where("source_id LIKE 'wamid.HBgLMTcw%'").last

[haguilarhern]

@muhsin-k when I run
Message.where("source_id LIKE 'wamid.HBgLMTcwOTk4NjU5NjkVAgARGBQzQjY2RDcwQjg2NzExNTdENUJCQQA='").last
it returns nil, so there is no Message record created for that echo event.

When I run the command you gave me
Message.where("source_id LIKE 'wamid.HBgLMTcw%'").last
it returns the outgoing bot message (not the echo one), as shown below:

irb(main):001:1* Message.where(
irb(main):002:1*   "source_id LIKE 'wamid.HBgLMTcwOTk4NjU5NjkVAgARGBQzQjY2RDcwQjg2NzExNTdENUJCQQA='"
irb(main):003:0> ).last
=> nil
irb(main):004:0> Message.where("source_id LIKE 'wamid.HBgLMTcw%'").last
=> 
#<Message:0x000076789bc03520
 id: 24796,
 content: "¡Que tengas un excelente día! 😊",
 account_id: 2,
 inbox_id: 34,
 conversation_id: 466,
 message_type: "outgoing",
 created_at: Wed, 04 Feb 2026 21:29:33.318855000 UTC +00:00,
 updated_at: Wed, 04 Feb 2026 23:21:35.391392000 UTC +00:00,
 private: false,
 status: "read",
 source_id: "wamid.HBgLMTcwOTk4NjU5NjkVAgARGBJENzlCRENDQkIxRTAyNDE3MkQA",
 content_type: "text",
 content_attributes: {},
 sender_type: "AgentBot",
 sender_id: 2,
 external_source_ids: {},
 additional_attributes: {},
 processed_message_content: "¡Que tengas un excelente día! 😊",
 sentiment: {}> 

[haguilarhern]

Hi @muhsin-k, just a quick follow up on this discussion to see if you had any feedback or needed more info from me. Thank you again for maintaining this project.

[dollyzn]

Why can’t I enable the new Advanced Assignment feature? Even after enabling it in the account as a super admin, when I refresh the page it still shows as disabled. Is this feature currently disabled?

[sojan-official]

cc: @tds-1

[tds-1]

@dollyzn Advanced assignment is a Business and Enterprise feature only. When you enable Assignment V2, a default policy is created for your account that supports round-robin agent selection and automatically assigns unassigned conversations starting from the earliest created.

[dollyzn]

@tds-1 What does assignment_v2 mean as a chatwoot_internal feature? Does it need to be enabled to enable advanced assignment?

[luanstartflow]

This version is buggy it blocks every action (sending messages, adding labels, etc.). When I refresh the page, the action finally runs and the message gets sent/arrives.

[sojan-official]

@luanstartflow

Could you check if there are any console errors in the browser (DevTools → Console) when this happens?

Also, please share any relevant browser console logs and server logs if available. That will help us narrow down whether this is a frontend issue or something getting stuck on the backend.

[Ranyelton]

Hi team,

After updating to version v4.11.1, I noticed that the AI features for "Fix grammar" and "Improve writing" have disappeared from the message composer.

Previously, these features were available through the standard AI integration. Now, it seems they are strictly tied to the Captain AI interface.

My questions are:

1. Is this an intended change to gate these basic text improvement features behind the Captain module?

2. Is there a way to keep the AI-powered grammar correction and message improvement active without fully enabling Captain?

3. If this isn't a bug, could you provide guidance on how to restore the previous behavior where the AI button was directly accessible in the editor?

This change significantly impacts the workflow for agents who rely on quick grammar checks but don't necessarily need the full Captain/FAQ automation suite.

Thanks for the great work on this release, and I look forward to your guidance!