LLM guidance moved simulated households toward broader participation in diversified equity funds.
Research-informed financial guidance
Better questions lead to
better financial decisions.
FinPath turns an incomplete money question into a structured financial brief—then creates a practical plan and shows its assumptions.
No bank connectionNo product commissionsAssumptions made visible
The research behind FinPath
AI advice can help.
The question is the bottleneck.
A 2026 working paper by researchers at MIT and Stanford simulated the lifetime effects of financial guidance from GPT-5.2 and Gemini 3 Flash.
Most participants moved closer to established household-finance principles—but outcomes varied with the information and framing in their prompts.
Read the 86-page working paper ↗Structured “academic” prompts improved consumption smoothing and reduced reliance on simple heuristics.
Differences in questions and model responses compounded into 4–5% retirement-wealth gaps between groups.
The paper evaluates simulated adherence to advice—not real-world returns or a guarantee that LLM advice is correct. FinPath uses its findings as design evidence, not as an endorsement.
Designed around the evidence
Not another blank chat box.
1
Fill the gaps
We collect the facts people commonly leave out: cash flow, debt cost, time horizon, income stability, and risk behavior.
2
Normalize the question
Your words become a complete, neutral brief. Demographics never stand in for risk tolerance.
3
Generate and audit
The plan must expose assumptions, balance monthly cash flow, address shocks, diversify, and state a rebalancing rule.
Clarity before recommendations.