← back
Don't just slap on a chatbot: building AI that works before you ask
Takeaway
Stop bolting chat onto products — proactive, context-aware AI that acts inside the existing workflow beats reactive chatbots.
Summary
- Tigon (AI issue tracker) ships proactive AI that watches user input and suggests actions inline — three modes: suggestion mode (real-time prompts), action mode (auto-split issues into subissues), and question-plus-action (PM-style timeline reasoning).
- Argues chat UI is the lazy default — proactive contextual AI like a fixed Clippy is the right pattern for professional tools.
- Three rules: AI supplements user agency, never replaces; AI offers recommendations, never forces; AI lives inside the natural workflow, not in a separate chat window.
- One-click revert keeps user in control.
productuxllm-apps
Original description
Everyone's racing to add chatbots to their products, but is that REALLY the best way to integrate AI? In this talk, Arthur Objartel from Evil Martians challenges the status quo of AI integration and shares product design insights from building Tegon, an AI-powered issue tracker. While most companies are making more reactive AI assistants and waiting for user commands, we took a completely different approach. This begs the question: should the future of AI be more proactive? For instance, remember Clippy from MS Word? While everyone loved to hate on that thing, we'll explore how its core concept was revolutionary — just 25 years too early! So, that means we’ve got questions: how AI can now enhance user workflows …without a chat interface? How to build AI that knows what users need before they ask? What are the practical principles to make AI truly useful in your products? Watch Arthur’s talk because he has the answers! https://evilmartians.com/chronicles/dont-just-slap-on-a-chatbot-building-ai-that-works-before-you-ask