Introducing Forge: LAB3’s Secure Agentic Platform

A Secure Agentic Platform for a New Way of Working

Today we are introducing Forge to the world, LAB3's accelerator for deploying and scaling AI agents across real business processes, with governance and human oversight built in. It provides the secure agentic platform to put AI agents to work at scale.

Forge brings agents, teams of agents and repeatable workflows together, with cost control and human oversight from the start. Together, these form a new operating system for the way LAB3 works, for ourselves and, now for our clients.

From Conversation to Capability

Forge began as a conversation between Mark Cayley, LAB3's Chief Engineer, and myself. We were talking about where AI was taking our industry, and what LAB3 should look like on the other side of it.

We kept coming back to the same problem. Building a single impressive AI demo is easy. Running hundreds of agents safely, across real business processes, with costs under control and humans still accountable, is hard. Nobody was solving that in a way we'd trust in a regulated client environment.

So Mark built a prototype, fast. Within weeks we had agents doing real work, governed by a policy engine that decided what each one could see and do.

That prototype became an initiative run by Mark within the LAB3 CTO Office, with help from people right across the company. Over the past six months we've built and shaped it with our engineers, consultants and managed services teams. What started as an experiment is now a fully fledged capability we're delivering with clients today.

The Building Blocks of Forge

Forge is built around a small set of concepts. Together they turn agentic AI from a series of experiments into a dependable way of working.

Agents

An agent is an AI worker with a defined role, a set of skills and clear boundaries. A Forge agent might be a pre-sales architect, a security analyst or a migration engineer. Each one knows what it is responsible for, which tools it can use and which data it can see.

Teams of Agents

Complex work rarely belongs to one person, and the same is true for agents. In Forge, agents work together in teams, each contributing its specialism. A team can research, draft, review and hand off work between members, much like a well-run project squad.

Workflows

Workflows are where agentic and deterministic work meet. A Forge workflow combines agent steps with proven, rules-based logic for anything that must be calculated, repeatable and auditable. Workflows and patterns are reusable, so a process proven once can be applied across many engagements.

Token Economics and Budgets

Every AI action has a cost. Forge tracks token consumption by agent, team and workflow, and lets leaders set budgets that cap spend before it becomes a surprise. Dashboards show live activity, cost and outcomes, so the value of AI can be measured rather than assumed.

Governance and Security

Governance runs through every layer. Role-based access control determines who and what can act. Agents run in secure container runtimes, a policy engine enforces what each agent may do, and high-impact actions require human approval. Every step is logged for audit and assurance.

Built on Microsoft, Designed for What Comes Next

Forge is built on Microsoft Foundry and the Microsoft Agent Framework, and runs on Microsoft Azure. It is deployed within each client's own environment and managed by LAB3.

We designed Forge to be portable, so it can grow with our clients and adapt as the platforms underneath it evolve. That matters, because the way we work has been changing quickly since AI arrived. Forge is our answer to that shift: a foundation for new ways of working, rather than a tool tied to a single moment in time.

The Operating System for How We Work

Forge is more than a platform. It is how LAB3 will work: agents doing the work they do best, people doing the work only people can do, and governance holding it all together. It is already making a difference in four places.

With Our Clients

Forge is already in production with clients in highly regulated sectors. In security operations, governed agents work alongside security teams to correlate alerts, triage incidents and draft response actions for analyst approval. The same foundation then extends into wider business processes, from reporting to operations.

In Managed Services

Forge is designed to scale LAB3's managed services capability. Agents take on high-volume, time-critical operational work, while our engineers focus on the decisions and improvements that need human judgement. Clients get faster response and more consistent outcomes.

In Professional Services

In our professional services engagements, Forge agents support discovery, architecture, migration and modernisation work. Reusable patterns mean each engagement benefits from what we have already proven elsewhere.

Inside LAB3

We use Forge ourselves. Agents already support pre-sales, sales reporting, workforce planning and architecture review across the business. Every workflow we prove internally becomes a pattern we can bring to clients.

The result is a more efficient LAB3: smaller teams delivering more, faster, with the governance enterprise clients expect.

This is just the beginning. Over the coming months we will share more about Forge, the patterns we are building and the outcomes our clients are seeing.