Every engagement, whether it is a firewall, a tenant migration or an AI tool, follows the same principles: understand first, build second, measure everything. No surprises, no hidden scope, no jargon.
Current as of 1 October 2026 — the six steps are unchanged. What is new is a standing item in every engagement on Copilot credit budgets, model choice and who reviews the sub-processor list, set out below.
Most consultancies jump straight to tooling. We don't. Governance and readiness come first, because an AI tool deployed into a confused business creates confusion at scale. Every step has a defined output and a written decision you can point to later.
A full engagement runs end-to-end, but individual steps can also be commissioned on their own — a standalone readiness assessment, a governance sprint, a Copilot rollout.
We start with your business, not the technology. A focused session to understand your workflows, pain points, data landscape, and what success actually looks like for you.
We map what's viable, what's valuable, and what needs sorting first. You get an honest view of AI readiness, risks, and a prioritised list of opportunities.
We build the policies and guardrails first. Acceptable use rules, data handling frameworks, approval workflows, and compliance alignment — before any AI tool goes live.
We build the solution — whether that's a custom AI tool, a workflow automation, or a governed Copilot rollout — integrated into your existing environment with your business rules baked in.
Deployment with proper training, documentation, and change management. We don't throw tools over the wall — we make sure your team knows how to use them and actually does.
We track the outcomes agreed upfront. Monthly support is available to optimise, troubleshoot, and identify new opportunities as your confidence grows.
Copilot agents and Cowork are metered in Copilot Credits, and the models on offer differ from one surface to the next. Both are now settled at the Governance step, in writing, before anything is built.
Credits are consumed while an agent is being built, not only when it runs, so the budget starts at the pilot. We set a credit limit per agent, agree a monthly figure for the tenant, and read the tenant-wide spend view at each review. If an agent is heading over, you hear from us before the bill does it for us.
The newest models are available in Cowork and Copilot Studio only, while Word, Excel, PowerPoint and Copilot Chat have their own options. So we record a model for each surface and each agent, with the reason, rather than one answer for the whole tenant. For a bespoke tool the same decision covers the API model and where inference runs.
Microsoft’s sub-processor list changes regularly. We agree who on your side owns the review, how often they check it, and what gets updated when it moves: the acceptable use policy and the record of processing. During the engagement we do the first review with them.
Not every engagement is an AI one. Network, security and Microsoft 365 projects follow the same order: read what is there, design it on paper, change it in phases, hand it over documented. See infrastructure & security and Microsoft 365 migrations.
Switch configurations, firewall policies, tenant settings, backups and licences are read directly, not taken from a diagram that may be years old. You get a written view of what is there, what is missing and what is exposed.
Migrations and network changes are pre-staged, piloted on a small scope and cut over in phases, each with a runbook and a rollback plan. Work is scheduled around your busy periods, whether that is a match day, month end or peak season.
Every phase ends with a handover document your in-house team or managed service provider can operate from: what changed, why, and how to undo it. Then either a standing engagement or a clean exit.
Three representative engagement shapes. Your scope and scale will be different — but the structure and level of transparency stay the same.
Mid-size advisory firm, ~50 employees.
A mid-size advisory firm knows AI matters but has no clear roadmap. They're worried about compliance, unsure which processes are suitable, and don't want to waste budget on the wrong tools.
Half-day discovery workshop, process mapping across core workflows, data readiness review, and a prioritised roadmap. Governance requirements and compliance considerations addressed upfront — before any tools are selected.
Professional services firm, ~120 employees.
Staff spending 15+ hours a week manually reviewing and summarising client documents. Inconsistent output, human error, and a bottleneck that affects fee earner utilisation and client turnaround.
Build a custom AI tool that reads incoming documents, categorises them by type, extracts key facts, and produces summaries — with human-in-the-loop review built in. Full audit trail for compliance and client confidence.
Retail or e-commerce business, ~30 employees.
The team has heard about AI but has no idea how to use it safely or effectively. No policies in place, worried about data risks, and uncertain which tools are appropriate for their work.
Full-day training workshop with hands-on exercises using real workflows. We create an approved tools list and AI usage policy aligned with data safety requirements. Follow-up sessions measure adoption and identify new opportunities.
Assessments and governance work typically take 1–2 weeks. Custom tool builds run 4–8 weeks depending on complexity. You'll always know the timeline before we start.
Everything is scoped and priced before work begins. No hourly rates, no surprise invoices, no scope creep. You know exactly what you're paying for and what you'll get back.
Regular check-ins, clear documentation, plain language. You'll always know where the project stands. We work with your preferred tools — Teams, Slack, email, or phone.
No pitch, no jargon. Just an honest discussion about where AI could make a real difference in your business.