Making procurement corruption economically irrational.
An AI trust layer for government procurement. Detection that cuts through the noise, trust scores that restructure incentives, and a shared intelligence network that makes every jurisdiction harder to defraud.
is spent on public procurement every year with no trust infrastructure built in. The world's largest market is also its least protected, and it leaks on an industrial scale.
Fifty years of reform has treated corruption as a moral failure.
Better people, stronger culture, harsher punishment. Decades of this playbook have produced limited sustained movement on Transparency International's index, while the detection systems meant to catch fraud drown investigators in noise: 90–95% of alerts are false positives (Datos Insights, 2025).
Corruption is not a moral phenomenon. It is an equilibrium, and equilibria don't respond to sermons. They respond to payoffs.
The honest contractor faces economic death.
Today, bribery is the dominant strategy. When competitors pay and you don't, you lose the contract and the appeal. Rational actors converge on corruption, not because they are bad people, but because the payoffs reward it.
Watch the equilibrium move. The players don't change. The game does.
Credit bureaus already solved this, for lending.
They made good behaviour economically rational by tying it to future access, and they made reputation portable. A defaulter could no longer walk into a new bank with a clean slate. The game changed not when individual lenders got smarter, but when the system learned collectively.
Procurement has no equivalent infrastructure. WITIA builds it.
Five years ago, impossible. Five years from now, infrastructure.
Statutory rails
The UK Procurement Act went live in February 2025: a central debarment register, new transparency duties, a Procurement Review Unit. Reciprocal debarment, WITIA's core mechanic, now has legal rails to run on.
The analyst bottleneck is gone
LLMs make an entire procurement landscape interrogable in plain language. Detection tools once sat unused because every question needed an analyst and a query language. That bottleneck no longer exists.
Machine-readable spend
Open contracting standards have matured into machine-readable spend, contract and supplier data at national scale: exactly the substrate a cross-jurisdiction network needs to learn from.
The rails, the interface and the data now all exist. No incumbent has claimed the layer that connects them.
Ask. Detect. Protect.
Ask
Interrogate an entire procurement landscape in plain language. Surface anomalous spend, hidden vendor relationships and cross-jurisdiction risk in seconds, with no analyst or query language required.
Detect
Ensemble AI flags the fraud single-method systems miss and scores every vendor across eight trust dimensions, with explainable evidence trails built for audit defensibility, not black-box flags.
Protect
A cross-jurisdiction intelligence exchange. Fraud caught in one authority trains the model that protects the next, so every jurisdiction on the network makes the others harder to defraud.
Interrogate billions in spend in a sentence.
Live product · synthetic demonstration data. Live-data validation is what the pilot delivers.
Signal, not noise.
Legacy tools drown investigators: 90–95% of alerts are false positives (Datos Insights, 2025). Watch what WITIA's ensemble does to the same alert queue.
- 01Shell vendor network0.94
- 02Invoice splitting0.91
- 03Bid rotation0.87
- 04Conflict of interest0.82
- 05Change-order abuse0.78
In a live demonstration, WITIA surfaced a shell-company network hidden across four tiers of subcontracting while processing one million transactions.
Detection catches fraud after the fact. Trust scoring makes it irrational before it starts.
Tap any dimension to flip its signal. Scores are ranked across vendors and confidence-weighted: a thin history cannot game an extreme score.
Caught once. Blocked everywhere.
Flagged once, in Europe's largest local authority. Blocked across the network before the next bid is filed.
Anonymised fraud patterns and debarment signals travel across the network. Standalone detection from day one; the exchange is additive.
Why won't the incumbents fix this? They can't.
The ERP giants
If Oracle shipped a tool proving Oracle Financials enabled the leakage, it would be selling against itself. The vendors who run government finance are structurally unable to audit it.
Palantir
Wins through $20M national deployments, an order of magnitude above what a local authority can pay. The exact buyers Palantir can't reach are the ones WITIA is built for.
Point detection tools
Spend Network, OpenTender, SAP's fraud modules address detection in isolation. None build trust scores that restructure incentives. None learn across jurisdictions.
The moat is not the AI. It is the network: trust scores travel, so every authority that joins makes WITIA harder to replace, and impossible to copy without cannibalising the deals that fund the incumbents.
Priced to remove friction. Built to compound.
Prove
£100k engagement fee + 10% of fraud surfaced. The authority pays meaningfully only when we find what others missed. Outcome pricing removes buyer friction and gets us into live data fast.
Expand
A flat platform fee tied to procurement volume under management, a smaller outcome-linked component, and modular add-ons: vendor risk scoring and pre-bid integrity.
Platform
Full platform pricing, with fraud detection as one of five product lines on the trust layer: predictable budget lines for the authority, compounding revenue for WITIA.
The 10% is designed to disappear. Outcome pricing wins the pilot; once officers trust the platform they want a predictable budget line, and revenue has to compound.
Illustrative, using UK National Audit Office loss-rate estimates: a $2B authority leaks tens of millions a year to fraud and error. The outcome fee is priced against recoveries, not new budget.
Land one node. The networks carry the rest.
Local authorities buy through shared procurement frameworks, so one pilot creates a contractual expansion path to neighbouring councils without restarting the sales cycle. The first node sells the region; the region sells the nation.
Following WITIA's invitation to and acceptance at NACo, America's largest government conference, the same motion runs in parallel through county procurement cooperatives. Expansion is simultaneous, not sequential.
We don't sell one authority at a time. We sell into networks that already exist.
Jordan Unokesan
Cambridge Land Economy (Law, Economics & Policy). Won the university-wide valedictorian prize, breaking four records with the highest attainment in the subject's recorded history. His dissertation surfaced $2B in anomalous procurement spend and changed national government policy, using the same fraud engine that powers WITIA. That work earned him fellowships from Emergent Ventures and Schmidt Futures Prometheus X, and a place among Powerlist Magazine's top 10 Future Leaders.
Jamie Ogundiran
MSc AI, King's College London. Award-winning builder and four-time hackathon winner who has judged and spoken at competitions across Europe, Saudi Arabia and San Francisco. Built the world's first culturally aware LLM, spanning 15+ African languages, and helped build the first ever unified MCP server, which hit #1 trending on GitHub and earned a keynote at GitHub HQ.
Raising $3M to begin the era of trust.
Corruption is an equilibrium.
Equilibria can be redesigned.
Backed by Emergent Ventures and the Schmidt Futures Prometheus X Talent Fellowship.