Changelog

We’ve added three new frontier AI models to Chatling:

1. GPT-5.5: OpenAI's new flagship. Stronger reasoning and instruction-following than GPT-5.4, with noticeably fewer hallucinations on knowledge-base grounded answers. Best pick when you need the most accurate, reliable responses and action calling for complex support flows.

2. Claude Opus 4.8: Anthropic's most capable model yet. Excels at long-context understanding (great for large knowledge bases), nuanced tone, and following detailed brand/persona instructions. Choose this when conversation quality, safety, and handling of long or ambiguous queries matter most.

3. Gemini 3.5 Flash: The speed option. Near-instant responses, while improving on Gemini 3 Flash in reasoning and multilingual quality.

All three models are now available on Chatling paid plans.

We shipped three improvements that to improve your live chat experience.

1. Canned responses

Create a library of saved replies your team can drop into any conversation in one click from Settings → Canned Responses.

To insert canned responses when chatting, click the bookmark icon in the reply composer and choose a response.

2. Teams

Group agents into teams so you can route conversations, scope visibility, and keep large support orgs tidy. Manage teams under Settings → Teams.

3. Quick actions

Press / in the reply box to bring up a searchable menu of everything you can do without leaving the keyboard:

• Insert a canned response

• Attach a file

• Add an emoji

• Switch between reply and private note

• Pause or resume your AI agent

• Send message

• Clear the current draft

You can now connect Notion as a knowledge source for your AI agents. Sync your existing Notion pages directly into your knowledge base - no more copy-pasting content or maintaining duplicate docs.

To get started, head to your knowledge base and select Notion as a source. Connect your Notion account, select the pages you want to include, and Chatling will import them automatically.

You can now cap how many AI credits an individual AI agent or chatbot is allowed to use each month. Once the cap is reached, the Agent stops generating responses until the next monthly cycle, or until you raise the limit.

This is useful when you want to allocate budgets across multiple AI agents, prevent any single Agent from depleting your project's quota, or build an Agent for a client without giving them unlimited spend.

To set it up, head to your chatbot or AI agent's settings and open the new Limits tab.

Read the docs →

We're rolling out granular access control for team members. You can now limit which AI agents each member can access and interact with inside a project.

When inviting a new member or editing an existing one's role, you'll find a new Restrict access section.

Leave it empty and the member gets access to every agent in the project, just like before. Add one or more AI agents and that member will only see those, nothing else in the project is visible to them.

This new features is useful for:

• Agencies managing multiple clients under one project. Give each client's team access only to their own agent.

• Large teams where support agents should only handle specific products or departments.

• Compliance scenarios where access to certain conversation data needs to be segmented by role.

The new feature is available on all plans. Head to Project Settings → Members to configure access controls. Give it a try and let us know your feedback.

A few practical upgrades to make conversations richer and easier to manage. Here's what's new.

File Attachments

Both end-users and your agents can now share files directly in chat, making it easier to handle support requests, share documents, or send media without leaving the conversation.

• Supported types: images, documents, audio, and video

• Drag-and-drop: just drop a file onto the chat to attach it

• Up to 3 files per message, with optional caption text

• Per-channel support: works on the web widget, WhatsApp, and Instagram. The file picker automatically narrows to what each channel accepts, so agents never have to guess

• Fine-grained controls: pick exactly which document types your end-users can send (PDF only? Spreadsheets but not PowerPoint? Whatever fits your workflow)

Enable it from the widget designer under Configure → File attachments.

→ Read the docs

Block Users

Tired of repeat spammers? You can now block individual users from the Conversations page. Once blocked, the user can no longer start new conversations with your bot or agents, and any in-progress chat is blocked too.

You can review and unblock users at any time from your settings.

Emoji Picker

A small but long-requested upgrade: end-users can now pick emojis from a built-in picker when writing their messages. No more copy-pasting from a separate app or remembering shortcodes.

That's it for this release. As always, let us know what to build next. 👇

We've shipped a handful of improvements to give you more flexibility with AI agents, better visibility into AI performance, and a smoother day-to-day workflow. Here's what's new.

1) New AI action: Custom Form

We’ve released a new Custom Form action for AI agents that lets you display forms with custom fields. Previously, the only option was the Lead Form, which came with a fixed set of predefined fields.

With the new Custom Form action, you decide exactly what information you want to collect from your visitors, whether that's a support intake form, a booking request, a feedback survey, or anything else your workflow calls for.

2) Confidence Score for AI Answers

When viewing a chat, you can now see a confidence score for each AI answer that comes from your knowledge base. The score reflects how confident the AI is in its response, giving you a clear signal of where it's performing well and where it might be falling short.

You can also filter conversations by confidence score, making it easy to surface low-confidence messages and fine-tune your knowledge base to improve future answers.

3) Sound Notification for Handoffs

You can enable a sound notification that plays whenever a conversation is waiting for agent intervention.

To turn it on, head to Settings → Notifications → Personal Notifications and enable the sound for "A new conversation is waiting for agent intervention" notification.

4) Improved Knowledge Base Filtering and Sorting

We've added new filtering and sorting options to the knowledge base to help you find and manage your content faster. You can also now sort entries by clicking directly on the column headers.

5) Switch Back to AI Chat from Live Chat

You can now hand a conversation back to your AI agent after a live chat session. From the chat screen where the text editor is displayed, click the AI chat toggle at the top to re-enable AI responses.

As always, we'd love to hear your feedback. Let us know what you think and what you'd like to see next.

You can now connect your AI agents directly with Zendesk and take actions automatically.

With Zendesk actions, your AI agent can:

• Create tickets instantly from conversations

• Automatically pass context from chat to your support system

• Create users in your Zendesk CRM

Your AI agent can now escalate issues and keep everything in sync with your Zendesk support workflow.

This means faster resolutions, smoother handoffs, and a better experience for your customers.

Give it a try today and let us know what you think!