I work on AI agents. Not the "here's a ChatGPT wrapper" kind — actual autonomous agents that do tasks on behalf of small businesses.
The thing nobody talks about: there's a massive gap between what AI agents can do and who can actually use them.
A developer can set up an agent, connect APIs, handle auth, debug when something breaks. A restaurant owner who wants AI to handle their booking confirmations? They can't. Not because the tech isn't there — but because every solution assumes you know what an API key is.
This is the gap that matters. The people who would benefit most from AI automation are the people least equipped to set it up. And "just make it simpler" isn't the answer — it's a different product entirely. You need:
• Managed infrastructure (they shouldn't know what a server is) • Guardrails that actually work (the agent can't go rogue with their Twilio account) • Failure modes a non-technical person can understand and fix • Trust signals that don't require reading logs
We've been learning this the hard way. The tech works. The packaging for real humans is the actual product.
For anyone building in this space — what's your experience? Are your users technical, and if not, where do they get stuck?
[link] [comments]