Partner Story Β· in conversation with Justin Dobson of sebbi.pro. This is a written interview submitted through Submit Your Story. The claims below are his own, presented as they were given to us. No payment was involved.
Justin Dobson with his son. Photo: courtesy of Justin Dobson.
Four years ago, Justin Dobson was sitting by a river in the north east of England with his son, skipping stones across the water.
After a few throws, the boy stopped and looked down at the rings spreading across the surface. Why, he asked, were the ripples the same shape as the rings inside a tree, and the spiral on the back of a snail?
"That single observation blew my mind and turned into an obsession," Dobson says. "Nature doesn't guess, negotiate, or hallucinate. It leaves an immutable signature at the exact point of origin."
He looked at where artificial intelligence was heading and saw the opposite. Millions of automated decisions spreading out across finance, infrastructure and ordinary lives, and almost no reliable record of what had actually happened at the moment each one began.
That question became sebbi.pro. He named it for two people: his late brother and his son, both called Seb.
A founder from the trades
Dobson did not come out of a research lab or a venture-backed incubator. He spent 25 years in painting and decorating, much of it working alongside his father, Joseph, and he builds sebbi.pro from Blyth in Northumberland under his company, Monop Content.
The trade, he says, taught him the principle the whole product rests on. "In decorating, you learn very quickly that you can't talk your way out of bad prep work. Either the foundation is square, or the whole wall falls over."
The problem: nobody can prove what the agent was allowed to do
AI systems are moving from drafting text to taking actions: approving payments, changing access rights, running code. When something goes wrong afterwards, the question every auditor and regulator asks is simple, and surprisingly hard to answer. What exactly did the system decide, under what authority, and can you prove the record has not been changed since?
Most organisations answer with their own logs. Dobson's point is that a log the company controls is a log the company could, in principle, have edited. He separates two things that are easy to confuse: the integrity of a record, meaning it has not been altered, and the validity of a decision, meaning it was the right one under the rules. His product is built to guarantee the first, so that people can argue honestly about the second.
What AILeash actually does
The product, which he calls AILeash, is an API that sits underneath a platform running AI.
Each time a user or an agent acts, the platform sends the event to sebbi.pro. The engine scores it across nine signals in under 30 milliseconds and returns one of three answers: allow, challenge or block. Before the reply goes back, the decision is sealed into a chain of records secured with SHA-256 hashes.
The clock on that chain is deliberately kept outside the company's own control. Records are timestamped through OpenTimestamps, which anchors them in the Bitcoin blockchain, and other independent platforms witness the chain as well. The aim is that nobody can quietly rebuild the history afterwards, not a hacker, not an employee, and not sebbi.pro itself.
For a business, the practical promise is one sentence on the site: when a regulator asks what your AI decided and why, you answer in one API call. The product is built to support the record-keeping and human oversight duties in Articles 9, 12, 13 and 14 of the EU AI Act, alongside the UK Online Safety Act and the Children's Code.
Where the money is
The pricing is published and plain. The full engine is free for 90 days against a platform's real traffic, with no card and no contract. After that it costs 50p per active device per month, metered on the devices that actually used the key, with no tiers and no sales calls.
The comparison Dobson draws is with dedicated compliance tooling, which he puts at tens of thousands of pounds a year. The pressure on the other side of the ledger is regulatory: under the EU AI Act, fines for the most serious breaches can reach a percentage of global turnover, and the obligations for high-risk systems arrive in stages through 2027. A tamper-evident record of every decision is the kind of evidence those obligations ask a company to be able to produce.
How it was tested
Dobson has put the architecture through mutual witnessing tests with independent peer systems, including ThePraesidium.ai, Shango and MIR, in which each side exchanged and verified the other's cryptographic records.
He also took part in ReCon Challenge #001, a public exercise that asks governance tools to reconcile executed payments against the authority limits that applied when they were made. Different tools approach that task in different ways. Dobson's design goal was narrow and specific: that every step should end in a verifiable receipt rather than an open question, and he reports that sebbi.pro produced one in each of the scenarios it ran.
Honest about where it stands
sebbi.pro is live. Anyone can get an API key and run it. What it does not yet have is customers. Dobson is now looking for its first enterprise pilots: system integrators and developers building AI agents with real stakes, who need an audit trail they can hand to an auditor and have checked without asking anyone's permission.
He is also growing a network of independent witness nodes, because the value of a shared record rises with the number of parties who hold a copy of it.
The lesson
Asked what building this has taught him, Dobson goes back to the riverbank.
"While billion-dollar companies were trying to build smart AI that guesses what humans want downstream, I built the lock at the water's edge: an architecture that anchors the origin, so the truth can never be altered, denied, or faked."
It is a large ambition for a one-person company in Northumberland. It is also, in a sense, the most practical idea a decorator could have brought to AI. Get the preparation right at the start, and everything built on top of it has something solid to stand on.
sebbi.pro is at sebbi.pro.
---
This is a Partner Story: a written interview with a company building with AI, submitted through Submit Your Story and published free of charge. Statements about the company's product, tests and results are its own. Building something with AI? Tell us about it.
Would you try this product?
