Hey everyone,
I’m curious how other teams handle building and maintaining their knowledge base for SaaS apps — especially when it comes to automating customer support with AI or chatbots.
Do you use your docs as the main source of truth?
Are you pulling directly from your codebase or a CMS?
How do you keep it updated as your product evolves?
Would love to hear how you’re approaching this — tools, workflows, and lessons learned.
Hey @muhammadanees-4377! This, indeed, is an interesting topic to discuss which is evolving at the speed of light 
small essay incoming, so, thank you for reading →
As a support engineer at Vercel myself, I’m super in awe of what the Vercel support agent has been able to achieve in terms of helping customers, quicker. Documentation, guides, FAQs, are all important data sources to add (btw, I, personally, still believe there is merit to having a “human in the loop” since your customer support rep would be able to review the automated responses for future fixes and ultimately improve the whole flow 
).
Ultimately, a mindset shift for everyone in any org (or even solo-founders) to strive towards is to transform into “editors” of the data source(s) that the AI uses for automated replies to customers vs. remaining “authors” of replies to customers. We do have a few existing insights/learnings around this topic, such as Example of customer support ticket classifying evals and Transforming customer support with AI: How Vercel decreased tickets by 31%.
All in all, I’m very excited for even more learnings to be made public on this topic by the team in the near future; stay tuned! 
Coming soon 
We use our docs as the single source of truth. They live in a CMS, not the codebase, and every product change includes a doc update. The same content powers the knowledge base and the chatbot, so support stays consistent. Regular reviews and support feedback help keep everything up to date.
Interesting! Thanks for sharing What is your tech stack?