About Fideleo
Built to release value on proof, within human authority.
Fideleo is an AI-born venture of Future Thesis Lab, a DIFC venture studio. It is a neutral escrow and conditional-release ledger: value is held in regulated custody and released only when fulfilment conditions are met and machine-readable human authority limits are respected.
Where the venture stands today.
The controls that govern value were built for people, not for AI colleagues.
Value increasingly moves between software acting on behalf of organisations. The controls that authorise that movement were designed for human approvers working at human speed. Fideleo addresses a well-evidenced problem in human B2B — value leaving a business before fulfilment is proven — and extends the same discipline to transactions between AI colleagues acting for different companies.
For a finance leader accountable to the board and auditors, the practical question has changed. When procurement and finance agents can initiate payments, 'claw it back later' no longer offers reliable protection, and pre-funding a full escrow ties up working capital that the business would rather keep available.
What the evidence shows
The evidence base is drawn from human B2B, not from AI-agent commerce. In 2024, 79% of surveyed organisations experienced attempted or actual payments fraud. The dominant vector is counterparty impersonation, at 45% and up 11 points year on year. Over the same period, recovery weakened: the share of victims recovering 75% or more of their losses fell from 41% to 22%.
Taken together, these point to a shift in where value is protected — away from clawing losses back after the event and toward controlling release before value leaves custody.
The wedge: authority and liability, not custody alone
Fideleo enforces machine-readable human spend and authority limits and binds release to verified fulfilment. Routine, within-policy releases clear automatically; anything that exceeds a value threshold, is irreversible, or breaches a delegated-authority limit is held with its evidence for an accountable human to approve or reject.
On the research provided, no surfaced competitor combines this governance layer with regulated fiat custody and KYC/AML. That combination — verified counterparty identity bound to release, segregated value, and a documented failure scenario — is what lets a compliance officer defend the arrangement to auditors and the regulator.
An AI-born company under accountable human governance.
AI colleagues perform the operating work continuously. Named human governors approve material movements and own fiduciary duty. No material or irreversible release occurs without either an explicit policy authorisation or a human approval, with a complete audit trail behind every action.
AI colleagues run the operational core
They monitor fulfilment conditions against agreed terms, ingest and validate delivery artefacts and oracle confirmations, reconcile balances across escrow accounts, and route exceptions. Within-policy releases execute automatically.
Human governors hold authority
Accountable humans define policy-as-code, approve threshold breaches and irreversible releases, adjudicate contested cases, and retain the fiduciary and regulatory obligations for what their agents commit to.
A regulated home in the DIFC
Entry is planned via a DFSA licence: Category 3D at USD 200,000 base capital, or Category 3C at USD 500,000 where value is custodied. The intent is to establish a trusted, regulated position before machine-to-machine volume scales.
Every transaction is auditable
Each release produces a record suitable for dispute resolution and compliance review — evidence a treasury controller and a risk officer can reconcile against policy after the fact.
Earn revenue from human B2B today; build the regulated position before machine volume arrives.
Why now
Two forces converge. Prevention-at-release is displacing claw-back, because fraud is near-universal, counterparty impersonation is rising, and post-loss recovery is deteriorating. At the same time, agent-payment protocols are standardising, creating the substrate that machine-to-machine escrow will need. Authorisation, however, is not settlement — the conditional-release and governance layer above the rails is largely unbuilt for autonomous agents.
How the opportunity is sized
The market is anchored to the reported machine-to-machine payment pool of USD 9.24bn in 2025, rising to USD 54.95bn by 2034. On stacked assumptions, addressable conditional-settlement TAM is estimated at USD 1.4–2.8bn, regional SAM at USD 60–120m, and a three-year obtainable revenue proxy at USD 0.6–3.6m. These are assumption-based planning ranges, not forecasts or commitments.
The risk we are underwriting
There is no direct evidence of demand for conditional-release settlement between AI agents. We state this as a material, open risk rather than a solved problem. Fideleo therefore earns revenue today from human-B2B conditional settlement while it validates inter-agent willingness-to-pay, and builds a regulated, trusted position ahead of the volume it is designed to serve.
This is the discipline of Future Thesis Lab: B2B infrastructure that generates per-transaction fees before the ecosystem reaches full scale, presented to investors without over-claiming.
Assessing Fideleo as a counterparty or an investor?
We can walk through the regulatory perimeter, the segregation model, the defined failure scenario, and how authority limits are enforced at release. Write to us and we will respond directly.