https://trybarkan.com/blog Activation, onboarding, and what we learn shipping an AI guide that onboards every user. en https://trybarkan.com/blog/when-your-ai-support-agent-should-stop-talking https://trybarkan.com/blog/when-your-ai-support-agent-should-stop-talking Fri, 04 Sep 2026 00:00:00 GMT Gabriel Lancelot Containment rewards an AI support agent for avoiding humans, even when the customer is still stuck. Design a handoff that preserves context and recovers trust. https://trybarkan.com/blog/ai-agent-approval-is-a-permission-boundary https://trybarkan.com/blog/ai-agent-approval-is-a-permission-boundary Sat, 29 Aug 2026 00:00:00 GMT Maxence Rabrais Design AI agent approvals around consequence, timing, exact action details, and one-shot consent so routine work stays fast without granting broad authority. https://trybarkan.com/blog/the-second-workflow-problem https://trybarkan.com/blog/the-second-workflow-problem Sat, 22 Aug 2026 00:00:00 GMT Maxence Rabrais Teams optimize the first five minutes and lose users in week two. Learn why the second workflow decides activation, how to measure it, and how to fix it. https://trybarkan.com/blog/why-we-read-the-screen-not-the-docs https://trybarkan.com/blog/why-we-read-the-screen-not-the-docs Sat, 22 Aug 2026 00:00:00 GMT Gabriel Lancelot Doc-trained assistants are confidently generic. Barkan grounds every answer in the live rendered interface. The reasoning, what we capture, and the trade-offs. https://trybarkan.com/blog/nobody-reads-your-documentation https://trybarkan.com/blog/nobody-reads-your-documentation Fri, 21 Aug 2026 00:00:00 GMT Maxence Rabrais Docs are written by experts, for generic readers, in another tab. See where activation actually leaks, why chat widgets didn't fix it, and what does.