Company

Serious infrastructure for serious AI.

Built by people who have been on your side of the review, running the data platforms, the risk architecture and the research programs inside regulated enterprises. That order of events shows up in the architecture.

Why we built this

A note from the founders.

Rhodes & Kang · Oct 2025

We’ve spent our careers shipping software inside regulated enterprises: at Morningstar, at IBM Research, and on risk and trading desks at RBC and TD. Generative AI is the most useful tool the enterprise has seen in a decade. It is also the most ungovernable.

Drop a generic AI assistant into a regulated enterprise and the compliance team can’t tell what was said, which model said it, or which policy was applied. Risk can’t review at the rate the work is generated. Audit trails are an afterthought. The result is a tool that’s useful in spite of the enterprise’s governance, not because of it.

AlphaBitCore exists because that tradeoff is artificial. Enforcement, supervision, and audit belong inside the execution layer, not bolted on after the fact. Local model options, firm-defined policies, per-stage model selection, and cryptographically sealed Proof of Execution are not features we added; they are what the platform is.

Regulated enterprises need to control how AI runs, not just what it produces.

Founders

Built by people who have shipped enterprise software at scale.

Deep operating experience in data infrastructure, research systems and applied AI, across asset management, capital markets and global research labs.

James Rhodes

Co-founder & CEO

Background

James led one of the largest data and software businesses in financial services as President of Morningstar’s Direct Platform, owning an $800M+ P&L and a 4,000-person organization across 29 countries. Before that he served as Morningstar’s CTO and CDO, and was Partner & CTO at Rocaton Investment Advisors, where he built and spun out a big-data wealth platform Morningstar acquired in 2016. He spent 13 years at IBM Research as a Master Inventor with 30+ patents in financial modeling, text analytics, and data mining.

Track record

Oct 2025 → NowCo-founder & CEO · AlphaBitCoreChicago
2022 → 2025President, Direct Platform · Morningstar$800M+ P&L · 4,000+ people across 29 countries
2018 → 2022Chief Technology Officer · Morningstar1,500+ technologists · cloud migration · ML/AI program
2016 → 2018Chief Data Officer · Morningstar500+ across data, science, and engineering
2011 → 2016Partner & CTO · Rocaton Investment AdvisorsBuilt wealth-tech spinout acquired by Morningstar in 2016
1998 → 2011STSM / Master Inventor · IBM Research30+ patents · youngest Master Inventor in IBM history

George Kang

Co-founder & CTO

Background

George brings two decades of building risk, trading, and wealth-tech systems for global financial institutions. He founded ITECH (ItechChoice) in 2015, a global engineering studio delivering wealth-tech and AI-native advisory platforms for banks and asset managers across APAC and North America. Earlier he was CTO at RequiSight, Associate Director of Risk Architecture at RBC Capital Markets covering liquidity, market and credit risk, Senior Solution Architect at TD Securities on the strategic risk reporting platform, and a Tech Lead at IBM Research.

Track record

Oct 2025 → NowCo-founder & CTO · AlphaBitCoreRuntime, Gateway, and execution architecture
2015 → NowFounder · ITECH / ItechChoiceWealth-tech & AI-native advisory · APAC + North America
2014 → 2015Chief Technology Officer · RequiSightRightPond institutional DB/DC analytics platform
2011 → 2014Associate Director, Risk Architecture · RBC Capital MarketsLiquidity, market, and credit risk
2011Senior Solution Architect · TD SecuritiesStrategic risk reporting · VaR · P&L attribution
2003 → 2011Tech Lead, Sr. Research Software Engineer · IBMEngagement tools for IBM GTS · DB2 UDB SVT framework

Principles

Four ideas the platform is built on.

Proof, not posture

Governance dashboards describe what a system is meant to do. A Proof of Execution shows what it actually did. The control plane is built around producing that proof at runtime, not reconstructing it from logs afterward.

Enforcement is execution

Policy, identity, scope and authority are not downstream checks. They run inside the Gateway, as part of the same execution that produces the work. Denied actions produce zero state change.

The enterprise controls the enterprise

Model choice, local execution options, firm-defined policies and per-stage routing exist because regulated enterprises need to control how AI runs, not just what it produces.

Category, not feature

Provable AI is bigger than any one company. It is a posture every serious enterprise use of AI will eventually require. AlphaBitCore is the first implementation. Our job is to make that future usable now.

The company

Research-backed. Enterprise-ready.

Agentic systems in this industry fail on data before they fail on reasoning. If the numbers underneath a client document cannot be traced to a licensed record, nothing built on top of them is deployable, however good the analysis reads.

Standing as an Approved Agentic Solutions Provider is what lets provenance run all the way from a figure in a finished proposal back to the record it came from. It helps that the team spent a decade building the platform that data comes from.

Founded: October 2025
Headquarters: Chicago, IL
Stage: Seed
Backed by: Hyde Park Venture Partners
Built for: Regulated enterprise

Standing

Where we are, publicly.

Every figure in a finished client document traces back to the licensed record it came from.

Morningstar

An Approved Agentic Solutions Provider, with the deepest integration of Morningstar data into an agentic system.

Live in market

A commercial deployment with paying advisors, in production since April 2026.

Morningstar wording is exact and deliberate: an Approved Agentic Solutions Provider, with the deepest integration of their data into an agentic system. That is a statement about our standing as a provider and the depth of the integration, not a claim about who else may license the underlying data.

Enforcement, supervision and audit belong inside the execution layer, not bolted on after the fact.

Talk to us.

The most useful first conversation is usually the one where you bring the objection that killed your last pilot.

Request a demo Read the argument