Key insights
- The post discusses the use of AI tools like Codex and Claude Code for company valuation, specifically for tasks like structuring DCF models, sanity-checking assumptions, and generating written explanations. While acknowledging the potential benefits in separating deterministic work from narrative, the author highlights risks related to mathematical inaccuracies and weak assumptions. The overall sentiment suggests cautious optimism regarding AI's role in enhancing valuation workflows.
Curious if anyone here has started using Codex, Claude Code, or other agent-style tools for company valuation.
I don’t mean "Tell me what stock to buy" type prompts. More like using it to structure a DCF, sanity-check assumptions, compare margins/reinvestment/growth, or write up the reasoning in a way that is easier to audit?
I’ve been experimenting with this locally, and I’m finding it useful, but also a bit dangerous if the model is allowed to make up the math or gloss over weak assumptions. The useful part seems to be separating the deterministic valuation work from the written explanation.
Has anyone here built a workflow they actually trust? What do you let the model do, and what do you absolutely keep outside the model?