Conference affiliation: Principal Product Manager · Intuit · 2026
On this page
Udi Menkes is a principal product manager at Intuit developing financial intelligence and advisory systems that help small businesses make better decisions. He also creates GenAI PM and open-source tools that give AI product assistants specialized expertise, persistent memory, and practical workflows.
From financial data to AI products
Menkes studied at Tel Aviv University and worked on a machine-learning project at Payoneer that modeled customer profitability. His career continued through financial-crime detection at ThetaRay and AI and data leadership at OpenWeb, where his work included conversational intelligence, moderation, audience targeting, and revenue optimization.
At Intuit, he shifted into product leadership for financial language models and advisory experiences. By 2024, he had co-authored work on a custom language-model leaderboard for financial applications, evaluating models against the specific demands of the domain.
The ideas and projects defining his work
The fluent bluff. Menkes uses this term for polished financial advice unsupported by evidence about what happened when comparable businesses made similar decisions. A general-purpose model might recommend expansion to a cash-strapped landlord, a price increase that endangers a supplier’s dominant customer, or cutting a vendor responsible for most of a company’s revenue. Access to financial statements alone does not teach a model which intervention actually works.
State-action-outcome trajectories. Menkes’s approach to outcome-grounded financial AI connects business conditions, decisions, and subsequent changes in profit, revenue, or cash flow. Drawing on financial records associated with QuickBooks, TurboTax, Credit Karma, and Mailchimp, the system combines causal inference, conditional average treatment effects, and reinforcement learning. General-purpose models suggest possible actions; outcomes-grounded models evaluate those actions against the experience of comparable businesses. The resulting advisory experience explains recommendations and helps customers develop action plans.
Product expertise as executable skills. His awesome-pm-skills project converts publicly available product-management podcast transcripts into reusable skills for Claude Code and Cursor, covering strategy, discovery, prioritization, design, and executive communication. Menkes has described his goal as helping AI assistants learn from experienced product and design leaders.
Persistent memory for product assistants. His genaipm-claw project extends NanoClaw into a product-management operating system with reusable skills, briefings, dashboards, and retained context. His separate claude-never-forgets project exports and indexes Claude Code sessions, allowing later conversations to retrieve relevant information through semantic search.
