/ AI is about to run the world, not assist it.
The only thing that makes that safe is proof. /
Aelethion builds Athena so that every consequential thing an AI does inside a company is seen, held for a named person where it matters, and sealed into a record an outsider can check without trusting us. That is safe, accountable AI, and it is what lets a company that could not or would not use AI on real work do it now. We are a small team in Tallinn. We say tamper-evident, never tamper-proof, and we ship the verifier so nobody has to take our word for it.
The unconcealment of truth. Not truth as an opinion, truth as what is no longer hidden. Making the world’s AI answerable is what Aelethion is for, and the name is the standard we hold the product to: nothing an AI does on your behalf stays concealed from the people accountable for it.
/ The accountability layer
for a world run by AI. /
Every AI action binds to the named human who authorised it. Not a log near an approval, the same action.
Evidence an outsider verifies offline, with no network and no trust in us. We ship a verifier, not a promise.
You keep your data, keys, logs and evidence. We never take custody, and verification never needs our servers.
The approved action, who it counts against, and proof anyone can check, in one sealed record.
/ Meet the people.
Behind Athena. /
Then there is the team. Reuben built the verification and identity stack that seals every Athena record. Brendon spent twenty years in B2B software learning how to build products that companies actually adopt. Estonia is where digital trust was invented, and it is where we are building the layer that lets the rest of the world hand real work to AI without losing command of it.
CEO and engineering. Built Athena’s verification and identity stack: the post-quantum signing (ML-DSA-65), the sealed-record substrate and the kernel-level confinement. Authored the open evidence-format standard published as an IETF Internet-Draft (draft-burls-mtac) and filed the five provisional patents behind the binding method. Before Aelethion, algorithmic and high-frequency trading, where the governing lesson was that confident systems break expensively, so you prove things rather than trust them.
COO. Twenty-plus years in B2B software. Founded Newstrack Online, which runs Athena live in production today; Brendon founded it, so it is a disclosed related-party reference and we say so. Owns operations, finance, legal and the commercial motion, and brings a twenty-year C-suite network across regulated enterprise.
GTM adviser, not a founder. Formerly Head of Partnerships at Estonia’s e-Residency programme. Books and runs the Tallinn meetings the pipeline is built from, and opens enterprise, channel and investor doors.
/ The giants have the resources.
A pre-seed team shipped the fusion. /
Microsoft, Google, Palo Alto, Cisco, Fortinet and Zscaler have every resource a company can have, and each has shipped a piece of this: a governed workspace, an offline-verifiable substrate, a detection stack. Everyone has a piece. Nobody has the combination. A small team in Tallinn shipped it, run as an EU-sovereign product: control over what AI sees and does, delegated human authority on every action, and a sealed record anyone verifies offline.
Today Athena is the control room for the riskiest AI decisions inside an enterprise. Next, it is the neutral layer every AI system writes to, from a copilot to an autonomous agent, each one leaving a record the world can check. A platform cannot notarise itself; ours is the record that verifies without us. When AI runs the business, someone has to prove it did only what it was allowed to. Athena is how.