Services

What we build, and what it takes.

Six services, grouped the way an engagement actually runs. Most clients start in one box and move right.

Decide

Before anything gets built

AI Strategy and Roadmap

Work out what to build, in what order, before spending money on any of it.

A current-state assessment across data, systems and delivery capability. Use cases scored on business value against feasibility. A phased roadmap with milestones, resourcing and a risk register. An executive briefing to get the room aligned.

Technical detail
  • Build against buy against configure analysis for each candidate use case
  • Model selection rationale documented with the tradeoffs, so a future team can revisit the decision
  • Total cost of ownership modelling including inference cost at production volume
Best for
A leadership team with more AI ideas than capacity, or a program that needs restarting
Timeline
4 to 8 weeks

Data Foundations

Fix the data problems that would otherwise surface halfway through a build.

An audit of what data exists, what condition it is in, and who owns it. Then the pipelines, access controls, and quality checks needed before a model can be trusted with it.

Technical detail
  • ETL and ELT pipeline design: moving data between systems on a schedule, with failure handling
  • Data lineage: a record of where each number came from, which is what makes an output auditable
  • Feature stores where a model needs a managed home for its inputs
  • Labelling and annotation workflow where training data has to be built rather than found
Best for
Anyone whose AI effort has stalled on data quality, access, or a system that will not give it up
Timeline
4 to 12 weeks

Build

The system itself

Process Automation

Stop paying people to retype invoices, route approvals, and chase exceptions.

Document extraction and routing, data entry and validation, inquiry triage, reconciliation and anomaly flagging. Every step leaves an audit log, because the finance and compliance questions come later and they always come.

Technical detail
  • Human-in-the-loop review on anything the system is not confident about, with the threshold tuned to your error tolerance
  • Accuracy measured the way your process cares about, not on a benchmark score
Best for
Three or more people doing repetitive document or data work
Timeline
3 to 10 weeks

Custom AI Development

Build the thing that does not exist off the shelf, on your own data.

Systems that answer from your own documents, classification and prediction models, document and image processing, and the evaluation harness that tells you whether any of it is actually working.

Technical detail
  • Retrieval-augmented generation where answers must come from your documents and cite them
  • Language model fine-tuning where prompting has provably hit its ceiling, and not before
  • MLOps: the plumbing that keeps a model running after launch, including versioning and rollback
Best for
A defined use case that a general-purpose tool has already failed at
Timeline
6 to 20 weeks

Integrate and run

Where most AI investments quietly fail

Systems Integration

Put AI inside the ERP and CRM your team already works in, not beside them.

AI that lives outside the systems your operations run on creates a second place to check. We embed it into the workflow people are already in, with the authentication, permissions, and audit trail that implies.

Technical detail
  • CRM and ERP integration, data warehouse and BI connections, API layer design
  • Real-time pipelines where a batch job would be too slow to be useful
Best for
Anyone whose AI pilot works in a browser tab nobody opens
Timeline
4 to 16 weeks

AI Operations

Keep it working after launch, and know the moment it stops.

Models get worse as the world moves away from the data they were trained on, and they do it silently. Monitoring, scheduled retraining, incident response, and a monthly report in the terms your operation already measures.

Technical detail
  • Drift detection: catching a model getting quietly worse before it produces a costly error
  • Scheduled retraining and evaluation against a held-out set
  • Integration health monitoring, because the platform underneath will change without asking
Best for
Anyone with an AI system already in production and nobody watching it
Timeline
Monthly, ongoing

How an engagement runs.

Whichever service you start with, the shape is the same. Each phase ends at a point where you can stop.

Week 1

Readiness review

A 45-minute conversation, then a short written assessment. Two to three use cases scored, one recommended.

You get Scored use cases and an honest cost range
Weeks 2 to 4

Production Readiness Assessment

Data, systems and process detail on the recommended use case, scored against the six controls. Fixed fee, quoted before it starts.

You get A score out of 18, a build plan, or a written recommendation not to build
Weeks 4 to 16

Build and integrate

Staged releases into the systems your team already uses. Each phase ends at a point where you can stop.

You get A working system, and everything needed to run it
Ongoing

Operate

Monitoring, drift alerting, retraining and a monthly report. Or your team takes it, which is the point of the handover.

You get Monthly review in your operating metrics

The Production Readiness Assessment, and how it is scored.

Each of the six is scored from 0 to 3 against the same four bands, for a total out of 18. The score is not a grade. It decides what the next phase is, and one of the possible answers is that there should not be one.

Each of the Six Controls scores 0 to 3

Nobody in the room can point to it. This is the normal starting position for at least two of the six and it is not a criticism.
It happens because a particular person makes it happen. It is not written down and it is not owned, so it leaves when they do.
Written down, owned by a named role, and agreed. Not yet tested against production volume, production data or a real incident.
In place, owned, and evidenced by something you could hand to an auditor without preparing it first.

What is scored

  • Production constraints The pilot never scales
  • Data readiness The data was not ready
  • Adoption Nobody changed how they work
  • Monitoring The model decayed and nobody noticed
  • Handover The vendor owns the outcome
  • Measurement Success was measured in the wrong units

What the total decides

0 to 6
The recommendation is usually not to build yet. The gap is not the model, and buying one will not close it.
7 to 12
Build, with the low-scoring controls as named workstreams inside the plan rather than as things to get to afterwards.
13 to 18
Readiness is not what is holding this up. The useful conversation is about which use case, not whether.

Questions we get asked.

What does this cost?
The readiness review is free and the assessment that follows is a fixed fee, quoted before it starts. Build costs depend on which use case you pick, and the honest answer is that nobody can give you a number before knowing that. What we can commit to is that you will leave the first conversation with a range for your specific situation.
We tried AI before and it did not work. What is different?
Most of the time the model was never the problem. It is usually unclear success criteria, data that was not ready, or a system nobody integrated into the actual workflow. We start by working out which of those it was, because that determines whether the next attempt is worth making at all.
How do we know if our data is ready?
You usually do not, and that is normal. Assessing it is part of the work rather than a prerequisite for starting. In practice most organisations have more usable data than they think, and the gaps that matter are narrower than feared. You get a plain answer either way.
What if the assessment says we should not build?
Then that is what it will say, in writing, with the reasoning. Every phase ends with a written review, and you can stop at any of them. A firm that only ever recommends building is not assessing anything.
Do we need to hire AI staff?
Not to start. We provide the engineering capability during the build and can run the system afterwards. Many clients do build internal capability over time and we actively support that, because the handover is the point. Source code, model weights, prompts and documentation transfer to you.
How long before we see results?
Automation work often reaches production in six to ten weeks. Custom development and integration into an established system landscape typically run three to six months. Milestones are set at the start and you get progress against them, not a status update.
What size company do you work with?
Size matters less than shape. What matters is real process volume, an ERP or CRM worth integrating with, and a team that can see the whole picture. Sectors so far include manufacturing, professional services, logistics, financial services and healthcare administration.

What you are agreeing to.

Selling risk mitigation while carrying none of it would be a strange position to hold. These are the terms, stated before you ask for them.

Fixed fee on phase one

The assessment is quoted as a fixed fee before it starts. No hourly overrun on the part where the scope is least certain.

You can stop at the end of any phase

Each phase ends with a written review against criteria set at the start. Stopping there is a normal outcome, not an awkward one. If the assessment says do not build, that is what it will say.

You own what we build

Source code, model weights, prompts, infrastructure configuration and documentation transfer to you. No proprietary runtime and no licensing dependency.

MSA and NDA on request

A standard mutual NDA is available before the first conversation, and a master services agreement before any work begins.

How we handle your data

Not sure which of these you need?

That is what the review is for. We will tell you where to start, including when the answer is nowhere yet.

Book an AI Readiness Review See a sample 90-day plan

You'll leave with 2 to 3 scored use cases, an effort estimate, and an honest cost range, whether or not we work together.

Contact us.

A person reads every message and a person answers it, within one business day.

Used to reply to you.

This site is protected by reCAPTCHA. The Google Privacy Policy and Terms of Service apply.

Or call +1 (844) 844-0097, or write to [email protected].