A working manual on proactive agents.

This site is a slow-built reference on proactive agents — what they are, why they’re different from the reactive agents most teams ship, and how to think about building them. Less brochure, more textbook with opinions.

What we mean by “proactive”

A proactive agent doesn’t wait for a prompt. It acts because time passed, data changed, or someone spoke. That shift — from reactive tool to teammate — is the most important architectural choice you make when shipping an agent. Most of the writing here is variations on that one idea.

Who’s behind it

Written by Khaliq Gant, co-founder of AgentWorkforce. Before this, three years as the first hire at Nango— the market leader in third-party integrations — where most of the day-job was getting the right context to AI: webhooks, normalized payloads, the long tail of provider quirks. The hard parts of proactive agents are, mostly, the same hard parts dressed up in a different jacket.

The essays are adapted from positioning docs, postmortems, and design notes that would otherwise live in PRs.

How we publish

Educational over promotional. Specific over abstract. Honest about what’s hard. New essays appear when they’re ready, usually weekly.