Agents are the new SaaS. Here's the whole playbook
Description
Watch: full video.Greg Isenberg argues that AI agents represent a shift from software that helps people do work to systems that can perform parts of the work itself. His practical playbook is to find a repetitive workflow people already pay someone to do, study how the job actually works, build the smallest useful agent, and add approvals, logs and checks before increasing autonomy. The useful lesson is that good agent opportunities start with understanding the work, not with finding somewhere to apply AI.
Official video embed
Greg Isenberg / The Startup Ideas Podcast, YouTube · 2026-07-01
Greg Isenberg / The Startup Ideas Podcast, YouTube · 2026-07-01
This video is hosted by a third-party provider. It will load only after you choose to press Play.
This video can load through the original platform's official player where available.
Open original sourceKey Takeaways
- Agent SaaS sells work, not just software: the important shift is from giving someone another tool to use toward taking responsibility for completing a defined part of their job.
- Start with a workflow that already has a “paycheck attached”: frequent, painful tasks with a clear outcome, existing software and manageable exceptions are stronger candidates than inventing an agent first and searching for a problem later.
- Study the human before designing the agent: watching real people perform the job reveals the context, exceptions, approvals and judgement that a high-level description of the task will usually miss.
- Autonomy should be earned rather than assumed: a useful first agent may draft, triage, coordinate or take one bounded action, while logs, approvals and evaluation provide the controls needed to trust it with more work over time.