Services
Production AI Implementation
- Duration
- Scoped per engagement
- My role
- I design and build it hands-on, in your repository, and hand it over. I do not run it for you long term.
When a prototype has proved out, or when you already know what needs to exist, this is the engagement that takes one workflow all the way to a running production system. It covers the parts a demo skips: integration with the systems the workflow touches, verification and escalation paths, per-decision monitoring of accuracy, cost, and latency, documentation, and a handover your team can own.
Discuss this engagement
Signs this is the right engagement
- A prototype sprint or internal proof of concept has answered the feasibility question
- The workflow touches documents, transactions, or support cases where mistakes are expensive
- Your team can own the system afterwards but does not have the capacity or experience to build the first version
- Compliance, audit, or data-residency requirements have to be designed in from the start
How it runs
- Discovery (1 to 2 weeks): scope one workflow, agree success metrics, map the systems and constraints involved, produce a written plan and fixed quote
- Build: the pipeline, its verification and escalation paths, and the evaluation harness, developed in your repository with weekly written progress notes
- Integration and deployment: connect to your systems, deploy to your environment, instrument accuracy, cost, and latency per decision
- Hardening and handover: run against real traffic with your team, document the system and its runbook, transfer ownership
What you get
- A production workflow running in your environment, with verification and human-escalation paths
- Evaluation harness and per-decision monitoring of accuracy, cost, and latency
- Architecture documentation and an operational runbook
- Structured handover sessions so your team owns the system
What I need from you
- A decision owner and an engineering contact who can answer questions within a day
- Access to the data and systems the workflow touches, in an environment you control
- Agreement on the success metric before build starts
When this is the wrong engagement
- Feasibility is still open (start with a prototype sprint)
- You want a vendor to operate the system indefinitely
Afterwards
Your team owns the system. Many teams move to an advisory retainer for design reviews as they extend it.
Relevant work
Fintech, SME accounting · In production
100K transactions processed, over 90% automated end to end
An AI-native accounting platform that has matched, categorized, and posted 100K transactions across 10+ businesses with over 90% automated end to end, escalating only what it cannot resolve.
Fintech, insurance and annuities · In production
Processing time cut from 4 hours to 8 minutes per application
A four-stage vision pipeline that reads handwritten annuity applications, cutting processing time from 4 hours to 8 minutes, with a monitored optimization loop gated by held-out evaluation.
Related writing
Essay · September 2, 2026 · 12 min read
Why the most important design decision in a production AI system is deciding when the model is not allowed to answer, and how to engineer that refusal.
Essay · September 2, 2026 · 10 min read
One model doing the work and an independent mechanism checking it is the single most reliable pattern I know for production AI. Here is how to build it.
Pricing
Every engagement is a fixed quote agreed in writing before work starts. The quote moves on four things:
- How many systems, data sources, and environments are in scope
- Whether the work happens inside regulated or restricted environments (data residency, audit trails, access approvals)
- How fast you need it: a compressed timeline costs more than a steady one
- How much evaluation data and system access already exists versus needs to be built first
Start with a short note
Describe the system and where it hurts. I reply within two business days; if there is a fit, we schedule a 30-minute call and I send a written scope within a week.