$5T opportunity: AI Roll Ups
Greg Isenberg · 2026-09-28
💡 Quick Take
1. Use AI agents now for back‑office work in service firms — they’re already outperforming many humans.
2. Use CodeX, Cloud Code, or Gemini as the core model platform for building your agents.
3. Skip trying to compete with billion‑dollar funds on large roll‑ups; focus on small‑mid‑market businesses.
4. Wait to scale to multiple firms until you have a solid shared‑layer of agents, rules, and a reliable GM in each business.
5. Avoid overpaying for acquisitions; price deals on current cash flow, not on speculative AI upside.
6. Use a weekly corrections‑log review process to continuously improve agent rules before going live with clients.
📊 Detailed Explanation
The speaker argues that AI agents have reached a reliability level where they can handle the repetitive, data‑heavy tasks of accounting, bookkeeping, and other service businesses better than many human staff. He cites Thrive’s accounting firm case where an accountant cut 180 hours of work to 15 hours, saving 31 % of time. This demonstrates that the technology is ready for production use, not just a beta experiment.
For the underlying model stack, Greg recommends using the same tools the big funds employ: OpenAI’s CodeX (or “codeex”), Google’s Gemini, and cloud‑based code execution platforms. These provide the necessary flexibility to build custom “intake,” “prepare,” and “reviewer” agents, and they are publicly accessible, so solo founders can replicate the setup without needing a proprietary partnership.
He cautions against trying to match the scale of Thrive or General Catalyst. Those funds target larger, multi‑million‑dollar firms, while the real opportunity for solo founders lies in the one‑million small businesses projected to change hands by 2035. By focusing on smaller firms, you avoid direct competition and can acquire at valuations based on existing cash flow.
However, he stresses a disciplined rollout: build a shared agent layer once, then plug each new acquisition into that layer. Only after the agents are proven reliable should you expand to additional businesses, and each should have a strong General Manager who receives equity to stay motivated. This staged approach mitigates integration risk and cultural friction.
Finally, the speaker highlights the importance of a corrections‑log and rule‑generation workflow. By spending an hour each week reviewing agent errors, converting repeat issues into immutable rules, you create a moat—your rule library becomes the most valuable asset, ensuring agents improve continuously and reducing the need for human rework.
🎯 Tech Expert Opinion
The core thesis—that AI agents can now drive a 3‑4× profit uplift in traditional service firms—is well‑supported by the real‑world examples Greg shares. The data from Thrive’s tax‑return processing and General Catalyst’s property‑management roll‑ups show tangible time savings and margin expansion, confirming that the technology is past the “rough draft” stage and ready for production deployment.
Greg’s advice to start with small‑mid‑market businesses and avoid overpaying is prudent. Early adopters who price acquisitions on current earnings rather than speculative AI upside reduce financial risk and give them time to perfect the integration playbook. This aligns with classic private‑equity best practices, but with the added advantage that AI reduces the cost base dramatically.
Overall, the recommendation is to adopt AI agents now for back‑office automation, use accessible model platforms like CodeX or Gemini, and grow cautiously with a strong governance layer (GM, rules, corrections log). The risks—cultural resistance, regulatory compliance, and potential over‑automation—are mitigated by the speaker’s suggested human‑in‑the‑loop checkpoints and gradual scaling strategy. This makes the opportunity both viable and attractive for solo founders willing to invest the operational discipline.
Kanal: Greg Isenberg