About
Codicx is an implementation firm, not a strategy practice and not a prompt shop. We take responsibility for the part of an AI programme where most of them quietly stop.
Mission
Codicx started in 2021 with a specific irritation. Our founder had watched three employers in a row spend serious money on AI initiatives that produced genuine excitement, a well-received demo, and nothing anybody could use. Each time the post-mortem blamed the technology. Each time the actual cause was that nobody had done the implementation work.
That work is not glamorous. It is defining what a correct answer looks like with the people who adjudicate it. It is authenticating against a system nobody has touched since 2011. It is building the review queue, wiring the evaluation gate, and agreeing who gets paged. None of it demos well, and all of it determines whether the thing survives its first quarter.
Five years on, we have put more than forty systems into production across twenty-three clients. The pattern has not changed. The teams that succeed are the ones that treat AI as an operational commitment rather than a technology purchase, and our job is to make that commitment a manageable one.
Values
If the data will not support the use case, if the timeline is wrong, if the process needs fixing before it needs automating — we say so in week one, when it is cheap. We have talked clients out of engagements we were commercially happy to take.
A demo can be built in a week. A system that holds up for a year cannot. We optimise for the second, which sometimes makes our first showing less exciting than the alternative you are also evaluating.
Success is your engineers maintaining and extending the system without us. We pair throughout, document properly, and treat a dependent client as a delivery failure rather than a retention win.
We commit to measurable outcomes before we start and report against them honestly afterward — including the engagements where the number came in under what we projected.
Team
Founder & Principal
Spent eleven years building decision systems in insurance and freight before founding Codicx, including four leading the automation group at a national carrier. Started the firm after watching a third consecutive employer spend seven figures on AI pilots that never reached a production queue.
Head of Engineering
Previously staff engineer on a payments platform processing several billion dollars a month, where he learned what it costs to get correctness wrong at volume. Owns the firm's position that the model boundary should be as small as the problem allows.
Principal, Applied ML
Research background in evaluation methodology and calibration, with a doctorate in statistical machine learning. Designs the graded evaluation sets that anchor every engagement, and is the reason clients can answer their auditors.
Principal, Platform
Fifteen years in infrastructure across regulated healthcare and financial services. Builds the shared platforms that make a client's fifth agent dramatically cheaper than their first, and has strong views on Terraform state.
Director of Delivery
Ran transformation programmes at a global consultancy before deciding she preferred teams that ship. Keeps engagements honest about scope and is the person who tells a client, early, when a plan is not going to work.
Principal, Security & Assurance
Former application security lead with a decade of red-team experience, now focused on adversarial testing of agentic systems. Writes the model cards and threat assessments that get Codicx builds through client compliance review.
Working together
No qualification gauntlet and no discovery call that is really a sales call. Describe the process that is stuck and we will tell you what we think, including when the answer is that you do not need us.