Services
From AI strategy to AI in production, with an operating model that keeps it there.
One path in four offers: define the strategy, roadmap, operating model and team, then prepare your data, your Microsoft 365 estate, your agents and your evidence, and keep them governed. Every offer is fixed-fee and ends with a hand-over to people in your organisation.
01 · For executive and data leaders
AI strategy, roadmap and operating model
The strategic offer. I define where AI should create value for your organisation, turn that into a roadmap your teams can execute, and design the operating model and team that sustain it. The aim is AI that keeps paying back long after the first pilot.
Is this you?
- You have AI pilots and ideas, but no shared view of which ones matter most.
- The roadmap is a slide, not a sequenced backlog with owners, costs and acceptance criteria.
- Nobody clearly owns the data and data products that AI depends on.
- Shared platform costs and individual initiative costs are argued over every time.
- Delivery is request-driven, and the team that started the work is changing.
The four components · what I defineWhat you can act on
StrategyA clear view of where AI creates value for you and where it does not: ranked opportunities tied to business capabilities, which uses stay as personal productivity tools and which should become engineered, workflow-embedded AI, what to build, buy or partner for, the principles and risk limits you set, and the measures that show it is working.
Actionable roadmapDelivery waves, not a slide. Each step names the business capability it unlocks, the data product and sources behind it, the readiness gaps and the minimum maturity needed, owners, dependencies, effort and cost with stated assumptions, milestones and acceptance criteria. The first step is broken into tasks, and I can build the first data product with you.
Operating modelHow AI work is funded, chosen, built and supported: a funding model that separates shared foundations from individual initiatives, intake and prioritisation, decision rights, a hand-over from delivery to support, lean governance checkpoints and an adoption plan. Built on the forums you already run.
Team structure and compositionThe roles AI needs and how they fit together: business and data-product owners, stewards, technical custodians, platform and AI engineering, security and governance. The skill mix and size, what sits in the business and what stays central, what to hire, borrow or partner for, and the path from request-driven delivery to proactive business partnership.
The transition I navigate
From personal productivity AI to engineered, workflow-embedded AI
Personal productivity AI helps individuals work faster, but its value is hard to see or repeat. Engineered AI is built into a business process, runs on governed data products, and is owned, tested and measured. I plan the move, and I build what it needs.
- Take stockWhere personal AI is used today and what it actually delivers.
- ChooseThe workflows worth engineering, ranked by value and readiness.
- BuildThe data products, guardrails and ownership each workflow needs.
- EmbedPut it into the process, measure it, and hand it to its owner.
Strategy sets the direction. Readiness and governance and agent assurance prepare the estate, the agents and the evidence to carry it out.
Tested on your own workBefore hand-over I apply the operating model to two of your live initiatives as worked examples, so the roadmap, owners and decision rights are proven on something that exists.
Handed over, not heldEverything ends with named owners in your organisation. Accountability stays with your management, and I plan the bridge to internal ownership from the first week.
Honest about progressA blocker that is not resolved is reported as a blocker, never as finished work. Roadmaps are plans your teams can execute, not a promise to build every solution within the engagement.
Architecture direction, where neededThe target architecture for your AI platform and agents, the evaluation and guardrail requirements, and a review of scalability and running cost. A readiness assessment separates what is demonstrated today from the gaps that remain.
Time-bound transition supportWhen people or leadership are changing, I keep agreed initiatives moving to a logical conclusion: decisions, issue resolution, user validation, adoption and hand-over, with a closure record of business acceptance, adoption evidence and benefits.
Pricing and limitsEach phase is a fixed fee, or a stated monthly capacity for a stated period, scoped in a 30-minute call. No hourly billing. This is not interim management or routine support. I build data products under separate fixed-fee scopes; your teams or partners build the AI applications, and I define, plan, align and assure.
02 · For CISOs and CIOs
AI readiness and managed governance
The readiness offer. Four to six weeks, one fixed fee, and an evidence pack your security and compliance reviewers can accept or challenge line by line.
Is this you?
- Your Copilot or agent pilot has stalled at security, identity or data-access review.
- Nobody can say which agents exist, what they can reach, or who owns them.
- Permissions on SharePoint and Teams are older than your AI plans.
- Your risk team wants evidence, and the pilot team has screenshots.
The assessment · what I review and buildWhat you receive
Identity for people and agentsA map of who and what can act in your tenant, with gaps ranked by risk.
Permissions and oversharingA prioritised clean-up list, with the scripts to do it.
Data maturity and architectureHow mature your data is for AI, whether the architecture underneath is fit for purpose, and what must be fixed first.
Data products, builtWhere the assessment finds a gap, I build the data product that closes it: sourced, quality-checked, owned and documented, ready for AI and BI to use.
Security review evidenceA ready-to-submit evidence pack for your security and compliance reviewers.
Rollout and measurementA staged rollout plan and the metrics that show it works.
Then, managed governanceMonthly. The agent inventory kept current, access reviews on a schedule, evidence refreshed for audits, and a short report for your risk function.
With Microsoft's own controlsI use what you already license, including Entra and Purview, and cover what those do not: agents from other vendors, evidence written for your regulator, and an independent view.
PricingA fixed fee for the assessment and a monthly fee for governance. Data products are scoped as separate fixed-fee builds. I scope it in a 30-minute call and send one price. No hourly billing.
03 · For risk and compliance · Early access
Evidence that your AI agents are controlled.
An independent assessment of each AI agent you build or buy, for heads of risk, compliance and model risk at insurers, banks, asset managers and life-science companies.
- EV-01Access scope. Every system, dataset and action the agent can reach.
- EV-02Failure testing. Documented tests of wrong, unsafe and out-of-scope behaviour.
- EV-03Human override. How, and how fast, a person can stop or correct the agent.
- EV-04Audit trail. What is logged, who can read it, and how long it is kept.
- EV-05Mapping to your obligations. The duties you name, such as DORA and the EU AI Act, and your internal policy.
Design partners wanted
I am looking for a small number of design partners among regulated organisations. Partners shape the evidence pack and receive their first assessment at a founding-partner rate.
What this is notIt is an independent assessment. It is not a regulatory certification and not insurance, and I cannot promise a regulator will accept it. It is built so your own risk function can rely on it.
Cadence and priceOne fee per agent assessed, with an annual reassessment. Quoted after a scoping call.
04 · For Microsoft partners and system integrators
Add governed AI delivery to your Microsoft practice, without hiring a team.
Your customers ask for Copilot and agents. The pilots stall at security and data-access review. I do the governance part with you, and you keep the relationship.
White-labelled assessmentMy fixed-scope assessment under your brand, delivered with your account lead.
Co-sold governanceMonthly governance for your customer. I deliver, you own the account.
EnablementTraining for your team and a first three workflows taken to production.
Strategy and operating-model packageThe strategy, roadmap and operating model for a customer's leadership team, delivered with your account lead.
Playbooks and templatesEvidence-pack templates, checklists and scoping sheets, in English.
Tell me about a customer whose pilot is stuck
I am an independent specialist. I do not compete for your customer's Microsoft licences or your managed-services contract.
Questions
Asked before most engagements
- Do you build things, or only advise?
- Both, within limits. I build the data products AI depends on, from sources to a governed, owned product. Your teams or implementation partners build the AI applications on top, while I define the strategy and roadmap, align the people involved, and assure the delivery.
- Do you take a leadership role in our organisation?
- No. Accountability stays with your management. I build the plans, name the owners and hand over.
- Can I start with one offer and add another later?
- Yes. Each offer is scoped and priced separately and ends with a hand-over, so you can start with whichever is most urgent.
- Will you need access to our data?
- For an assessment I need read access to configuration and permissions, not to business documents. Building data products needs agreed access to a development environment.
- Do you replace our Microsoft partner?
- No. I do the governance part and work alongside your partner, or through them.
- Can you assess agents from any vendor?
- Yes. I assess behaviour and access, not the vendor.
- Do you test our production systems?
- Only in a scoped, agreed environment.
- Do you certify AI systems or give legal advice?
- No. I provide an independent assessment and evidence. Legal conclusions stay with your counsel.
Not sure which one fits? Start with a call.
Thirty minutes, no charge. You leave with a view on whether an assessment is worth it.