A practical knowledge base for VPs of Finance, FP&A leaders, and hands-on CFOs at growth-stage companies. The goal is not to survey every AI tool. It's to help you figure out where AI can actually help your team today and in the future.
A few notes on how this was created: AI Adoption Framework
1. The State of AI in FP&A — the five usage patterns and my read on where things are heading
2. Identify Your Pain Points — a self-assessment to calibrate your starting point and the FP&A diagnostic across eight workflows
3. Put AI to Work — practical use cases organized by the five patterns
4. Where to Start — a couple practical tips
Appendix — Experiments — real things built and tested
This is a living document. AI in FP&A is moving fast enough that anything written six months ago is already partially out of date. What follows is my best current understanding, with sources where I have them and my own opinion clearly labeled.
There are five ways people are using AI right now. The first is personal. The other four describe a progression specific to FP&A work:
These patterns describe both how teams use AI and how that usage matures over time. Most teams move down the table as they build capability, from personal productivity toward agents. Within each pattern, they also deepen from left to right, early stage to mature, as they invest in structure and governance. Section 4 covers what to put in place before you start moving through this progression.
| Pattern | Early Stage | Mature Stage |
|---|---|---|
| Personal productivity | Ad hoc: email drafts, meeting prep, personal research | Projects, personal agents (like OpenClaw), and building ‘second brains’ |
| **Prompting | ||
| (in an FP&A context)** | One-off questions | Structured and shared prompt libraries |
| Workflows | One manual workflow, human-triggered each time | Connected workflows: multiple workflows linked to live data, faster cycle times |
| Automation | Simple macros that operate automatically and scripts | Bounded automation: governed automations with audit logs and approval steps |
| Agents | One narrow agent, tightly scoped | Agent orchestration: multiple agents working together across workflows |