A control plane for the AI you're already running.
Real cost numbers, real governance, real audit trails. You can't control what you can't see — and most organisations can't see their AI spend clearly.
Built for organisations running AI they can't fully account for
Visibility and control across every AI you run
The control plane is bespoke — built on your stack, owned by you. These are the consistent components we deliver on every engagement.
Unified cost tracking
Spend across all AI providers — Anthropic, OpenAI, Google, Azure, internal models — in one place. Per-team, per-project, per-user. The number your finance team can rely on.
Budget enforcement
Hard and soft limits with alerts before breaches, not after. Automatic enforcement stops runaway spend. No more surprise invoices that arrive three weeks after the damage is done.
Governance layer
Approval gates for high-cost or high-risk calls. Role-based access controls. Full audit logging. Every AI action is recorded in a format your compliance team can actually use.
Multi-audience dashboards
Three different views — finance, engineering, compliance — from the same underlying data. Each audience sees what they need. Nobody drowns in the other team's numbers.
System integrations
Connected to your existing SIEM, data warehouse, and identity provider. The control plane doesn't sit in isolation — it plugs into the infrastructure you already manage.
Ongoing operational review
Monthly review with the engineering team who built it. Quarterly tune-up. Incident response when something unexpected happens. You're not left to maintain a system you didn't design.
From inventory to live control plane in six to nine weeks
Setup timeline depends on the number of providers and systems in scope. Ongoing operations continue on a written SoW after the initial engagement completes.
Inventory
Map every AI tool and provider currently in use across the organisation — sanctioned, shadow, and embedded in vendor platforms. This is the baseline. Everything else is built from it.
Integrate
Connect provider APIs, internal systems, identity and logging infrastructure. Each integration is tested before we move to the next. No big-bang deployment at the end.
Configure
Budgets, approval gates, alerting thresholds, and dashboards — configured for your organisation, not templated from a default. Three audiences, each with what they actually need.
Operate
Monthly review, quarterly tune-up, incident response when needed. Priced transparently on a written SoW — not a vague retainer.
Transparent from the start. No percentage-of-spend pricing.
Initial setup from £35,000, depending on the number of providers and integrations. Ongoing operations are priced per month based on usage volume and review cadence — set out in a written SoW before you commit.
No per-seat pricing. No percentage-of-AI-spend pricing. You pay a fixed amount for defined work, not a number that grows as your AI usage grows.
- AI provider inventory
- Unified cost tracking across all providers
- Budget enforcement and alerting
- Governance layer and audit logging
- Multi-audience dashboards
- System integrations (SIEM, data warehouse, IdP)
A scoping call is free and takes 30 minutes. No obligation.
What this is not
You can't control what you can't see. Most AI spend happens in the blind spots — tools procured by individual teams, costs buried inside SaaS subscriptions, usage that nobody has mapped.
Delivered personally by Richard O'Flynn & George Lovejoy.
Twenty years of shipping bespoke production systems together. No associates, no delivery team.
Ready to see what your AI is actually doing?
A 30-minute scoping call is free. We'll confirm whether the engagement is the right fit and give you a clear scope and cost estimate before you commit to anything.
Common questions
Will this force my teams to stop using their preferred AI tools?
No. The control plane sits above the tools — it instruments them, it doesn't replace them. Your teams keep their existing tools and workflows. What changes is that those tools are now visible, measurable, and governed.
How does this integrate with our existing FinOps tooling?
We integrate with the data warehouse and reporting infrastructure you already run. If you're using a FinOps platform, we connect to it — AI spend becomes a line item in the tooling you already use rather than a separate conversation. Specific integration requirements are scoped in the Integrate phase.
What visibility will individual users have into their own costs?
Configurable. We can give individual users full visibility into their own spend, team-level aggregates only, or no visibility at all — depending on your culture and governance requirements. The default is to expose data at the team level, but we configure this before go-live based on your preferences.
Who owns the data the control plane collects?
You do. The control plane is built on your infrastructure and connects to your data stores. We have no access to your usage data after the engagement closes. We don't aggregate, anonymise, or use your data to train anything.
What happens when a new AI provider appears?
Adding a new provider is an integration task — scoped and priced as part of the ongoing operational SoW. The architecture is designed to accommodate new providers without rebuilding the core system. New providers typically take one to two weeks to integrate, depending on their API maturity.
Can we self-host this?
Yes. The control plane is built to run on your infrastructure — cloud or on-premise. Self-hosted is the default, not an option. There's no cloud service component that requires external hosting.