For enterprise & corporates
Many pilots, little production. We close that gap.
Large organisations rarely lack AI initiatives. They lack the decisions that let initiatives become systems: which platform, which governance, who owns the outcome, how to measure it. We work as a small, senior, vendor-independent unit alongside your architecture, security and procurement functions, and alongside the system integrators you already have.
Why enterprise AI stalls
The pilots work. What fails is everything around them: ownership, platform choice, governance, procurement and the patience to industrialise one thing before starting the next.
A typical picture: thirty proofs of concept across business units, three competing platforms, a Copilot rollout that nobody measures, an AI council that meets monthly and approves nothing, and a security board that has not yet seen a single agent architecture. Each pilot was reasonable on its own. Together they form a portfolio that nobody manages.
Turning this into production is not primarily a technology problem. It needs a platform decision the architecture board can defend, a governance model that satisfies legal, risk, procurement and the works council without slowing every project to a halt, and business-unit programmes with an owner, a baseline and a target.
We are not a large system integrator, and we do not want to be one. We are the senior, independent specialist unit that helps you decide, design and set the standards, then works with your integrators and internal teams so that the standards are actually applied. Our AI governance and AI strategy work is written for exactly this setting.
- Typical clientBusiness units and group functions of corporates with 2,000+ employees
- RoleSenior specialist unit alongside your SIs and internal teams
- PlatformsMicrosoft Copilot and Azure AI Foundry, Google Gemini Enterprise, AWS, Salesforce Agentforce, SAP Joule, ServiceNow, open-weight models
- OutcomeA governed, measurable path from pilot portfolio to production
Where corporates usually stand
A portfolio of pilots without a portfolio owner
Every business unit has experimented. Results are anecdotal, costs are scattered across cloud bills and licences, and nobody can say which three initiatives deserve the budget to go into production.
Platform decisions that are postponed or made by default
Microsoft Copilot and Azure AI Foundry, Google Gemini Enterprise, AWS, Salesforce Agentforce, SAP Joule and ServiceNow each come with their own agents. Without a target picture, the company ends up with all of them, integrated with none.
Governance that exists on paper
An AI policy has been published and an AI register started. What is missing is the operating part: intake, risk classification under the EU AI Act, model and vendor assessment, monitoring, and a clear path from idea to approved system.
Centre of excellence versus federated teams
Central teams set standards but cannot scale; federated teams move fast but diverge. Most organisations need a deliberate mix and have not yet decided where the line runs.
Boards, integrators and the works council all have a say
Architecture and security boards, procurement, the data protection officer, the works council and two system integrators are all involved. Each has legitimate questions. Without someone who speaks all of their languages, every project waits on the slowest.
What we do for corporates
We take on the pieces that large programmes tend to leave unowned: decisions, standards and the first measurable results.
Portfolio triage and production roadmap
We review the existing pilots against value, feasibility, risk and platform fit, retire the ones that should stop, and sequence the rest into business-unit programmes with owners, baselines and targets. Read about our AI strategy work.
Platform and reference architecture
An independent evaluation of Microsoft, Google, AWS, Salesforce, SAP, ServiceNow and open-weight options against your landscape, data residency and existing contracts. The result is a reference architecture your architecture board can adopt and your integrators can build to. See also build versus buy.
AI governance at scale
AI register, risk classification under the EU AI Act, model risk management, vendor and procurement assessment, monitoring and audit trails. Designed with legal, risk, information security and the works council so that it is used rather than circumvented.
Operating model: centre of excellence or federated
We help you decide what stays central (standards, platform, governance, evaluation) and what belongs in the business units (use cases, ownership, change), and we set up the intake process that connects the two.
Production programmes with your integrators
We design and lead the first business-unit programmes, from agent development to hand-over, working with your system integrators rather than against them. We set the standards, review the work and make sure the metrics survive the go-live.
Change management and AI literacy
Role-based training, the AI-literacy duties under the EU AI Act, communication with the works council and a realistic plan for how work changes in the affected teams. See AI training.
A typical path
Enterprise engagements are modular. Most start with a short diagnostic and grow from there.
Diagnostic
Portfolio review, interviews with business units, IT, security, procurement and legal, and a look at current platforms and contracts. Output: a written assessment of where production is blocked and what would unblock it.
Decisions
Platform and reference architecture, governance operating model, centre-of-excellence design and a sequenced production roadmap, taken through your architecture, security and AI boards with us in the room.
First production programmes
Two or three business-unit programmes with owners, baselines and targets, built by your teams and integrators under the new standards. We lead, review and measure, and we adjust the standards where reality disagrees with them.
Scale and step back
The operating model runs on its own: intake, approval, build, monitoring. We stay available as a senior review function, as a fractional Chief AI Officer, or not at all.
Frequently asked questions
How do you work with our existing system integrators?
As a specialist unit alongside them, not as a replacement. Typically we own the decisions and standards (platform, architecture, governance, metrics), review their designs and deliveries, and build the reference implementations that show what good looks like. Integrators usually welcome this because it removes ambiguity from their scope.
We already have a Copilot licence for everyone. Is that not enough?
A general-purpose assistant improves individual productivity; it does not automate a process. The measurable gains come from agents embedded in specific workflows, with access to your systems and clear guardrails. We help you get value from the licences you have and decide where purpose-built agents are needed on top.
What does AI governance at enterprise scale involve?
An AI register of all systems in use, a risk classification aligned with the EU AI Act, model and vendor assessment, defined human oversight, monitoring and incident handling, and a documented intake process. It must integrate with existing information security, procurement and data protection processes rather than duplicate them. We describe obligations in general terms and recommend confirming the current legal status with your counsel.
Should we centralise AI in a centre of excellence?
Usually a mix: a small central team for platform, standards, governance and evaluation, and business-unit teams that own use cases and results. Fully central teams become bottlenecks; fully federated teams produce thirty incompatible pilots. We help draw the line for your organisation.
How do you measure a business-unit programme?
Every use case starts with a baseline (cycle time, cost per case, error rate, throughput, revenue effect) and a target agreed with the business owner. We instrument the system so that the metric is reported automatically and reviewed monthly. Programmes that cannot be measured do not start.
How do you handle the works council in a corporate setting?
As a stakeholder from the beginning, with clear descriptions of what each system does, which employee data it touches and how work changes. Framework agreements on AI use often make individual projects faster; we help prepare them together with HR and legal.
Related services
AI governance, EU AI Act & GDPR
An AI register, risk classification under the EU AI Act, GDPR-aligned processes and a usage policy your teams will actually follow, built together with your lawyers and your data protection officer.
AI strategy & roadmap
A prioritised, costed roadmap of AI and agent use cases, a target architecture and a governance model, in four to six weeks.
Managed AI operations (AgentOps)
Monitoring, evaluation, model upgrades, cost control and incident handling for agents and LLM applications after go-live, with a monthly review and documentation your auditors can read.
Why AI pilots fail
The recurring reasons pilots stall, what production-ready actually means, and a phased path to an agent that runs on real work.
Let's find the first workflow worth automating.
A 30-minute intro call, no slides and no obligation. We listen, ask about your processes, and tell you honestly where AI agents would pay off and where they would not.