Turn a solved customer support conversation into an AI knowledge base entry in 10 minutes
From one closed support ticket to a ready-to-use «question - answer» pair in a Markdown file - no code or paid subscriptions needed; finish with a quick check against the source before publishing.
- 1.Install the WebAct extension for Chrome, confirm your WebAct account, and choose to use AI in the browser - ChatGPT. If needed, log in to ChatGPT in the same Chrome profile. No paid API key is required.
If the service is unavailable in your region, check the tool page - it lists the current access conditions.
WebAct ↗ - 2.Open a browser tab with a completed support conversation that has a documented solution. Save the conditions under which the solution was applied.
Before launching, remove names, contacts, order numbers, addresses, and payment data from the conversation: the tab content will be processed by a third-party AI service. The task only works with data from the active tab.
WebAct ↗ - 3.Enable WebAct on this tab, select the needed context, and switch to «Task» mode. Paste the instruction copied from the tool page there.
The instruction is customizable: specify the fields, tone, or detail level you need, while keeping the original facts intact.
WebAct ↗PromptPrepare a question and an answer for reuse based on one of my completed support conversations. Use only the documented solution. Before starting, ask me for any missing settings and source details. Work only with data from the current active tab. Create a Markdown file named webact-support-conversation-to-faq.md. The full file in UTF-8 must be under 48,000 bytes. Use only observable data and details I provide. Leave missing values as unknown. Preserve identifiers, dates, units, and currencies. Report any skipped, truncated, or unreadable content and do not claim full coverage. Cross-check the result against the source and stop if the necessary source data is missing. Do not add steps that are not in the documented solution. The draft must be reviewed before publishing; do not publish it. Start with no more than 25 entries or a short document. Specify exactly which source content is included.
What this prompt doesA ready-made instruction from the WebAct page: it tells the AI to build a «question - answer» pair from the open conversation, use only the confirmed solution from the ticket, invent nothing, and save the result as a Markdown file. Copy it in full without changes; add your own additions (fields, tone, detail) after it. - 4.Review the proposed actions and launch the task. The AI will create a file webact-support-conversation-to-faq.md with a «question - answer» pair for the knowledge base.
The task is counted against the limit when its first action starts; intermediate steps and retries do not consume the limit.
WebAct ↗ - 5.Cross-check the finished file against the original conversation: the answer should be clear to another customer without access to the ticket, and the solution should still be applicable. Publish the entry in your knowledge base or FAQ only after this check.
Check answers with numbers, deadlines, prices, and limitations - these should be confirmed manually. The task does not guarantee SLA compliance and does not apply changes to future tickets.
WebAct ↗
- freeExporting data from messengers/emailExporting conversation history via built-in client tools is free and has no limits.
- freeAI chat (ChatGPT Free / Claude Free / Gemini)Free plans allow file uploads and text processing, but may have limits on the number of daily requests or file size.
Why today
AI services change fast - interfaces and free limits may differ from what's described.