Productlane Agent
Get instant answers from your docs, Linear projects, and changelog.
Productlane Agent
The Productlane Agent answers customer questions and resolves conversations across your widget, live chat, and public docs. It is grounded in your help center, roadmap, and changelog, and it replies in the customer's language.
When a conversation is a bug or feature request, it can file a Linear issue, and it can hand harder conversations to your team.
Where to find it
Settings > Features > AI > Productlane Agent.
The header carries a \"How it works\" button that opens the in-app docs. A live widget preview sits alongside the settings so you can see the Agent as customers will.
Controls on this page
Productlane Agent
The main toggle. Turning it on activates the Productlane Agent for your widget, live chat, and docs. It is off by default. This setting saves the moment you flip it, and a toast confirms \"AI Chat enabled\" or \"AI Chat disabled\". The toggle requires the Starter plan and up; on lower plans an \"Upgrade\" button appears in its place. The rest of the page's controls stay hidden until the Agent is enabled.
Generate auto-replies with AI
A toggle under the \"AI auto-replies\" section. When on, your regular email auto-reply is sent with an additional, clearly labeled AI-generated section appended. It is off by default and saves instantly on change. Requires the Starter plan and up. The AI auto-reply never asks the customer a clarifying or follow-up question: if it needs more information to resolve the request, it hands the conversation to your team instead of going back and forth by email.
The AI reads the customer's message text and, if they attached a file or pasted a screenshot into the email, considers that too, up to 5 attachments per message.
Create Linear issues from agent chats
A toggle that appears only when Linear is connected and a default Linear team is set. When on, the Agent files a Linear issue for a chat only when it judges the conversation to be a bug or a feature request. To file an issue for every thread instead, use \"Auto-create issues\" in the Linear settings. Off by default, saves instantly, and requires the Starter plan and up.
Auto-assign Productlane Agent to bugs
A toggle that automatically assigns a coding agent to fix incoming bug reports. It requires a connected coding agent: without Cursor connected, the toggle is disabled and the description links to Settings > Integrations > Cursor to connect. This setting is stored on the workspace and saves instantly. Requires the Starter plan and up.
Exclude manually created issues
A toggle that appears only once \"Auto-assign Productlane Agent to bugs\" is on. When enabled, only bugs created by the AI chat are auto-assigned, not bugs created from manual feedback. It shares the coding-agent requirement, so it is disabled until Cursor is connected. Saves instantly.
Customize (Welcome Message and Suggestions)
This section appears when the Agent is enabled and your plan qualifies. If your workspace has more than one enabled language, a language selector in the section header lets you edit each language's content separately.
Welcome Message: a textarea for the greeting shown at the top of the chat. It defaults to a message that tells customers the AI is trained on help docs, roadmap, and changelog. It saves on blur and auto-saves while you type, after a short debounce.
Suggestions: three input fields for the example questions shown as starter prompts. Each saves on blur. All three are required.
Welcome message and suggestions are per-language. Editing them while a non-default language is selected updates only that translation; editing the default language updates the base content too.
Workspace guidance
A textarea under the \"Workspace guidance\" section where you provide instructions and context the Agent uses when responding. Leave it empty for default behavior. It saves on blur and auto-saves while you type after a short debounce. See the AI custom prompt article for guidance on what to write.
How to set up the Productlane Agent
Open Settings > Features > AI > Productlane Agent.
Turn on the Productlane Agent toggle. On the Starter plan or higher this activates it immediately; on lower plans, upgrade first.
Edit the Welcome Message and the three Suggestions to match your product's voice. If you support multiple languages, pick each one in the selector and translate the content.
Add Workspace guidance with any tone, policy, or product context the Agent should follow.
If you use Linear, connect it and set a default team, then turn on Create Linear issues from agent chats so bug and feature-request chats become issues.
To have bug reports fixed automatically, connect Cursor, then turn on Auto-assign Productlane Agent to bugs, and optionally Exclude manually created issues.
Optionally turn on Generate auto-replies with AI to append an AI section to your email auto-replies.
Use the live preview on the right to confirm the greeting and suggestions read well.
Handing off to a human
A customer can ask for a person at any point in the chat. The Agent recognizes an explicit request like "talk to a human" or "connect me with support" and immediately shows a "Talk with a human" button instead of trying to keep resolving the request itself.
Clicking that button confirms the request right away ("Thanks, our team has been notified and will follow up here") and, if Live Chat is enabled, moves the conversation into Live Chat on the same thread so the customer keeps reading one continuous conversation. The AI Agent stops replying in that conversation once a customer has requested a human, even if Live Chat is off.
The handoff pulls the thread back into your inbox, marks it unread, and flags it as high priority so it does not get missed. Your team is paged over Slack and email, based on each teammate's "AI Agent handoffs" notification setting in Settings > Notifications. See the Notifications article for details.
Handoffs to your team
When a customer asks to talk to a person, or the Agent otherwise can't resolve a conversation, it tells the customer the team will follow up and hands the thread over. A handed-off thread behaves differently from a resolved one:
The thread stays open. Productlane does not mark a handed-off thread as Done, so it keeps showing up in your open-thread views until a teammate handles it.
The timeline shows the handoff. Open the thread and its activity timeline includes an "AI Agent handed this conversation to the team" entry, so anyone can tell at a glance that the Agent handed off rather than resolved the conversation.
Your team gets notified. Turn on the "AI Agent handoffs" toggle for Slack and email in Settings > Notifications to be pinged the moment a handoff happens, instead of relying only on the "Where the AI hands off" chart in Reporting.
Notes
What grounds the Agent: your help center docs, roadmap, and changelog by default. This includes link entries in your docs sidebar, pages hosted outside Productlane that the Agent can read and cite with their real URL. Connecting GitHub adds your repositories as a source, and connecting PostgreSQL lets whitelisted read-only SQL inform answers. For email auto-replies, the Agent also reads any files or pasted screenshots on the customer's message, up to 5 attachments per message.
When it replies: the Agent answers customer messages in the widget, live chat, and docs while it is confident it can help. It hands off to your team when it cannot resolve a conversation or when the customer explicitly asks for a person, and it never leaves a customer with an unanswered clarifying question, it connects them with a human instead. Handed-off threads stay open rather than being marked Done; see Handoffs to your team above.
Activity in the thread timeline: when the Agent runs three or more actions on a thread within a 3-minute stretch, such as connector actions from your custom data setup, the timeline collapses them into a single entry like "AI Agent ran 7 actions" instead of listing each one. Expand the entry to see every action it took. If any action failed, the entry calls it out, for example "AI Agent ran 5 actions, 2 failed", so failures stay visible even when collapsed.
Monthly cap and limits: each confident, non-escalated resolution counts against your plan's included resolutions for the month. Once the free tier is used up, further resolutions stop unless pay-per-use is enabled. Set caps and see usage in Settings > Features > AI > Usage. See the Usage article.
Rate and message limits: each visitor is capped at 30 chat messages per minute per IP address and 10 messages per minute per chat session. A single message can't exceed 4,000 characters. Once a session reaches 50 messages total, the widget tells the visitor to start a new chat to continue.
Outage behavior: if the Agent can't reach its docs, roadmap, or changelog search at the moment a visitor asks a question, it says so and offers to connect them with your team instead of guessing or claiming something isn't planned or documented.
Autosave: every setting on this page saves automatically. There is no Save button.
Plan requirements: the Agent and its sub-toggles require the Starter plan and up. Coding-agent auto-assignment additionally requires a connected Cursor account.
Was this helpful?