Move 32 Labs
Building the trust layer for high-stakes AI.
We're auditors and engineers building multi-agent systems for work that can't afford to be wrong — starting with Veritas AI, our financial-statement reviewer, and scaling toward the oversight of autonomous AI itself.
Multi-agent review · every finding traced to its source · illustrative preview
Review is where high-stakes work quietly breaks.
A single mis-cast subtotal, a broken cross-reference, or one missing IFRS disclosure can survive every human pass and surface at the worst possible moment. Skilled reviewers burn days tracing numbers across tables instead of exercising the judgment only they can provide. The work is slow, manual, and unforgiving — and it doesn't scale.
Financial Statement Review, rebuilt as a team of specialists.
Veritas AI isn't one model guessing at everything. It's a multi-agent system where each agent owns one job and does it well, then a reviewer agent reconciles the findings, filters false positives, and reports only what matters. Every finding traces back to the exact figure, table, or standard that triggered it — so your reviewers can verify the machine, not just trust it.
Numeric Validation
Verifies every calculable figure, in-table and across tables, so subtotals, movements, and reconciliations actually tie.
Logic Consistency
Flags claims that are numerically correct but logically impossible (e.g., 500% revenue growth alongside an 80% headcount cut).
Disclosure Compliance
Checks the statement against applicable IFRS/IAS standards and surfaces only confirmed missing disclosures.
External Signal
Verifies the report's claims against the outside world (news, registries, public filings) to catch what the document alone can't tell you.
Veritas AI is the first of four.
The same multi-agent approach extends across high-stakes review and reporting work. Each product targets a different team and ships on its own timeline.
Veritas AI
Reads financial statements and flags numeric and logical inconsistencies before sign-off.
Demo readyManagement Report Review
Reviews any management report for accuracy and consistency.
On the roadmapFinancial Statement Preparer
Produces accurate statements so reporting teams move faster.
On the roadmapCompany Brain
Synthesizes a company's key information and makes it instantly queryable.
On the roadmapBuilt and validated by working auditors.
In controlled testing with senior-auditor oversight — synthetic statements seeded with known errors — Veritas AI reached 87–94% average accuracy across its four validation dimensions, with its strongest performance in disclosure compliance and numeric validation.
Results from controlled evaluations on illustrative/synthetic data; not an audit opinion.
Trusted by teams at
Placeholder for partner logos — shown once consent is confirmed
What we don't do tells you who we are.
A focused strategy is defined as much by what it excludes.
We don't ship thin LLM wrappers dressed up as “AI tools.”
We don't sell vague “AI transformation” where the value is impossible to verify.
We don't build outside our circle of competence — we start with finance, controls, and audit, and earn the right to go further.
We don't chase short-term sales over the long-term mission.
Two engines. One flywheel.
Every deployment earns revenue and teaches us how multi-agent systems behave under real pressure. We channel that capital and hard-won knowledge into the hardest problem in AI — alignment: building systems humans can actually trust with autonomous work. Our long-term goal is EMICA — Evaluation, Monitoring, Internal Controls, and Audit — for autonomous AI agents. Product funds and grounds the research; research makes the products harder to replicate. Financial oversight is where we start, not where we stop.
Not an ad-hoc team.
Auditors and engineers who've worked together for years — not strangers assembled around a trend.
18 years in audit, leading compliance and oversight.
10 years in software engineering, leading the AI systems.
Backed by seasoned assurance specialists and senior engineering talent.
See what your team stops missing.
Book a 30-minute walkthrough on a sample of your own statements.