01 / SharpBench
Assessed quality, not popularity.
An independent, nightly benchmark across three categories of services AI agents consume—scored on four separable axes, with every score traceable to a raw run.
Applied AI · 0→1 products · real-world systems
Ideas tested inthe real world.
I’m Rishabh Singhal, a principal applied-AI product leader and venture builder. I turn frontier capabilities into products people trust, deploy, and pay for.
See how I build Currently building companies at AI Fund
Selected evidence
2funded companies incubated
10×product adoption growth
$1M+new recurring-revenue business
5technology domains shipped in
Selected work · 01—03
Proof, not pitch decks.
Products and ventures where the decisions, evidence, and my exact role are visible.
Portfolio reconstruction · synthetic documents02 / TrueLumen
Every document should agree. This finds where they don’t.
I originated the idea, led validation, wrote the PRD, and built the pre-build prototype. The founding engineer built the production MVP, evals, and evidence graph.
Visit public site
Portfolio reconstruction03 / VocalBridge
Voice that stays fast without giving up deep reasoning.
I led validation, product definition, the ten-week co-build, and the funding structure. The founding engineer built the platform; the core pre-response idea came from the partner team.
10 → 250+ organic developers during the build
From the venture notebook
Earlier-stage bets, stated plainly.
These are active or recently incubated explorations—not all are launched companies or proven businesses.
04
Strata
Incubated ventureRegulatory intelligence built around living deliverables—work that stays current as the underlying record changes.
My role: PRD, validation system, investment case, CEO search
05
Vidably
Live pilotVerified-buyer video turned into structured product-page evidence, with measurement designed in from day one.
My role: Idea, UGC pivot, validation, schema, build leadership
06
AA Firewall
Working prototypeA developer-first control plane for governing the actions AI agents take across tools, files, shells, and browsers.
My role: Concept, prototype iteration, one-click install, validation
Systems shipped · 2016—2025
The operating range.
Different industries, one recurring job: turn technical possibility into a system that works in the world.
2023—24
Dialectic AI
Cofounder
Shipped three generative-AI products for healthier online discourse, writing product code, improving the model and UI, and growing daily use 10× in one month.
Founder loop · research → build → adoption → funding
2024—25
Terawatt
Lead Product Manager · Heavy-Duty
Led product strategy for Class 8 EV charging infrastructure, including charging optimization and predictive planning across vehicles, sites, and the grid.
Software × hardware × capital infrastructure
2021—22
Sofar Ocean
Principal Product Manager
Created a historical ocean-data SaaS offer that reached approximately $1M+ in recurring revenue and helped launch an open connectivity standard for ocean hardware.
New business · 5× org revenue · open ecosystem
2018—21
Kodiak Robotics
Senior Product Manager
Founded the product function and helped turn an autonomous-trucking engineering MVP into a commercial product across perception, safety, hardware, and operations.
Safety-critical AI · 60% fewer human interventions
2016—18
Uber
Strategy & Product Analytics Lead
Led a global analytics team and used product data to improve driver onboarding, including 50% signup growth across an in-person program spanning 700+ locations.
Marketplace scale · analytics · team leadership
Field note · 001
7 min read
Product judgment
The demo is the easy part.
Four tests I use to tell whether an AI capability can become a product: repeatability, workflow ownership, earned trust, and measurable value.
About · beyond the chronology
I like the messy middle between “possible” and “working.”
My work lives where new technology meets a real workflow, a skeptical user, and a business that has to make sense.
Today I help create applied-AI companies at AI Fund. Before that, I worked on autonomous freight, ocean data and hardware, heavy-duty EV charging, generative AI, and marketplace systems. The settings change. The through-line is finding the wedge, building the right thing, and earning adoption.
Away from work, I build small hardware and art experiments, backpack, play squash, and remain delighted by anything that makes the digital world feel more physical.
A useful conversation might start here
Building something technically ambitious?
I’m always interested in hard applied-AI problems, founder-quality product roles, and teams working where software meets the physical world.