Skip to main content

AI Consulting

AI Consulting Services

Most AI programmes stall between the pilot and the P&L. We find the two or three use cases that actually pay, prove one in a matter of weeks, and stay until it is running in production with your team owning it.

What you can rely on

  • Written scope and cost range before you commit
  • Working software in the first phase, not a report
  • Your engineers on the team from day one

Why AI Programmes Stall

The blocker is almost never the model. It is everything around it — unclear value, unusable data, and no route to production.

A Pilot That Never Becomes a Product

The demo works on a laptop and dies on the way to production because nobody scoped evaluation, monitoring, failure handling, or who owns it on Monday morning.

Use Cases Chosen by Enthusiasm, Not Economics

Teams build what is technically interesting rather than what moves a number. Without a value model, there is nothing to defend at the next budget round.

Data That Is Not Ready and Nobody Will Say So

Access, quality, lineage, and consent problems surface halfway through the build. We surface them in week one, when they are still cheap.

No Way to Tell Whether It Is Working

Without an evaluation harness and a baseline, "it seems better" is the only available verdict — and that is not enough to keep a system funded.

Start here

AI Discovery Sprint

From a vague ambition to a funded, de-risked plan in three weeks.

Duration
3 weeks
Price
Fixed, from $18,000
Team
3 specialists

What is included

  • Opportunity map across your workflows, scored on value and feasibility
  • A data readiness assessment that names the blockers, with fixes
  • One use case taken to a working prototype against your real data
  • An evaluation harness and baseline so progress is measurable
  • A costed delivery plan and a build-versus-buy recommendation

Fixed scope and fixed price from day one. Everything produced is yours, whether or not you continue with us.

Book a Discovery Call

Deliverables

What You Actually Receive

Artefacts you can act on, hand to another supplier, or take to a board.

Opportunity map
Every candidate use case scored on business value, data readiness, technical feasibility, and regulatory exposure.
Value model
A defensible per-use-case model of the cost saved or revenue gained, with the assumptions written down and challengeable.
Data readiness report
Sources, quality, lineage, access paths, retention obligations, and the specific work needed to make each one usable.
Working prototype
A running system against your real data — not a mock — deployed somewhere you can use it.
Evaluation harness
A test set, metrics, and a baseline, so every later change can be shown to help or hurt.
Delivery plan
Phased scope, team shape, timeline, cost range, and the decision points where you can stop.
Governance pack
Model and data inventory, human oversight design, and the disclosure obligations that apply to your use case.
Build-versus-buy recommendation
An honest comparison against off-the-shelf options, including the case for not building anything.

Where We Help

AI Strategy and Roadmap

Portfolio-level prioritisation, funding cases, and sequencing across multiple teams.

Use Case Discovery

Workflow-level analysis to find where automation or augmentation actually pays.

Proof of Concept Delivery

A working system against real data, built to be extended rather than thrown away.

LLM and RAG Architecture

Retrieval design, context strategy, model selection, routing, caching, and cost control.

Agentic Workflow Design

Tool use, planning, guardrails, and the human checkpoints that keep autonomy safe.

Evaluation and Observability

Offline evals, online monitoring, regression suites, and drift detection.

MLOps and Platform

Deployment pipelines, versioning, rollback, and the operational base a model needs.

AI Governance and Assurance

Risk classification, documentation, oversight design, and audit readiness.

Team Enablement

Pairing, code review, and structured handover so your engineers own what we build together.

How an AI Engagement Runs

  1. 01

    Diagnose

    Week 1

    Workflow interviews, data access review, and constraint mapping. We finish with a scored opportunity map and a recommendation on what to prove first.

  2. 02

    Prove

    Weeks 2–3

    One use case built against real data, with an evaluation harness so the result is a number rather than an impression.

  3. 03

    Harden

    Weeks 4–10

    Production concerns: latency, cost per call, failure handling, monitoring, access control, and human oversight.

  4. 04

    Operate

    Ongoing

    Live support, evaluation against real usage, model and prompt iteration, and structured handover to your team.

What We Build With

Models

  • Anthropic Claude
  • OpenAI
  • Google Gemini
  • Meta Llama
  • Mistral
  • Open-weight models on your own infrastructure

Platform

  • AWS Bedrock
  • Azure AI Foundry
  • Google Vertex AI
  • Kubernetes
  • Terraform

Data

  • Postgres and pgvector
  • Snowflake
  • BigQuery
  • dbt
  • Airflow
  • Kafka

Evaluation

  • Custom eval harnesses
  • LLM-as-judge with human calibration
  • Regression suites
  • Live tracing and cost telemetry

Engagement models

AI Consulting Pricing and Engagement Options

Indicative starting points, not quotes. You get a written price for your specific scope before any work begins.

Lowest-Cost Entry

01 Option

Discovery Sprint

Fixed-scope diagnostic ending in a prototype and a costed plan.

Starting Price

From $18,000

Timeline
3 weeks
Billing
Fixed price

Best For

You know AI matters but cannot yet defend a specific investment.

Discuss This Option
02 Option

Fixed-Scope Build

A defined system delivered to an agreed scope, date, and price.

Starting Price

From $60,000

Timeline
8–20 weeks
Billing
Fixed price, milestone-billed

Best For

The problem is understood and the requirements are stable.

Discuss This Option
03 Option

Embedded Team

Our specialists working inside your team, on your board, to your priorities.

Starting Price

From $14,000 per specialist per month

Timeline
3 months minimum
Billing
Monthly per person

Best For

Scope will move, and you want capability transferred as you go.

Discuss This Option
04 Option

Retained Advisory

Fractional AI leadership: architecture review, hiring input, and governance.

Starting Price

From $6,000 per month

Timeline
3 months minimum
Billing
Monthly retainer

Best For

You have a team but no senior AI voice in the room.

Discuss This Option

Every option is scoped and priced in writing before delivery begins. No surprise invoices or automatic upgrades.

FAQ

AI Consulting: Your Questions Answered

The things prospective clients ask us most often about this service.

Ask us something else
What Does an AI Project Actually Cost?
A discovery sprint is fixed at $18,000. A first production use case typically lands between $60,000 and $180,000 depending on how many systems it touches, how clean the data is, and how strict the compliance requirements are. Integration work and data remediation drive cost far more than model choice does. You get a written range before you commit to anything.
How Long Before We See Something Working?
A working prototype against your real data inside three weeks. Production, including monitoring, access control, and human oversight, is usually eight to twenty weeks after that.
Who Owns the Models, Prompts, and Code?
You do. All work product, including prompts, evaluation sets, fine-tuned weights, and infrastructure code, transfers to you on payment. We do not retain rights and we do not reuse your data to build anything for anyone else.
How Do You Measure Whether It Is Working?
We build an evaluation harness and a baseline before we build the system. Depending on the use case that means task accuracy, containment rate, cost per resolved request, human review load, or time saved per case — agreed with you up front and tracked in production.
Will You Use Our Data to Train Models?
No. Your data is used only to deliver your engagement. Where we use third-party model providers, we configure enterprise endpoints with training disabled and zero or minimal retention, and we tell you exactly which providers process what.
Can You Work with Our Existing Engineers?
That is the preferred arrangement. We pair, review each other's code, and run a structured handover. The goal is that your team can extend and operate the system without us.
What If Discovery Says We Should Not Build It?
Then we say so and explain why, and you have saved a great deal more than the sprint cost. That has been the honest recommendation on real engagements more than once.

Tell Us What You Are Building, and We Will Put Real Numbers on It.

Thirty minutes with the engineer who would run your project. You leave with a scope, a cost range, and a recommended next step — whether or not you work with us.

A 30-minute call with an engineer, not a salesperson. You leave with a scope, a rough range, and a recommended next step — whether or not you work with us.

  • Written scope before you commit
  • NDA signed on request
  • You own all work product