Skip to main content

Methodology

How Claude Activation Works

Four modules. Ten business days. One agent live at the end of the sprint, with the governance pack to support it.

What is the activation model?

The Claude Activation Programme is a fixed scope, fixed timeline engagement that delivers a working AI agent inside your business in ten business days. The work is delivered across four sequenced modules, each one a dependency for the next.

The activation is sector tuned. The same four modules apply to a law firm and a wealth manager, but the workflows, connectors, governance posture, and proof points are different. That tuning is baked into the sector specific delivery plan before kickoff.

The four modules

How the ten days are spent

01

Module 01: Knowledge Foundation

Knowledge Foundation indexes your documents, precedents, and process guides into a structured retrieval layer Claude can query. It runs across days 1 to 3 and decides the quality ceiling of everything built on top of it. Your knowledge lead and one IT contact carry the load on your side.

Output

A working retrieval layer connected to your document repositories with documented chunking, embedding, and access rules.

What it is not

Not a content audit, not a SharePoint migration, not a discovery workshop. We index what is already there.

02

Module 02: Workflow Automation

Workflow Automation maps the specific workflow the first agent will handle and connects it to your existing systems. We run a structured workflow design session, then build the connectors and prompt scaffold the agent will use. This module sits across days 4 to 6.

Output

A defined agent workflow with tested connectors into your document management system, CRM, or matter management platform.

What it is not

Not generic process mapping. Not a Zapier setup. The workflow is the one the production agent will run.

03

Module 03: Agentic Intelligence

Agentic Intelligence is where the agent is built, tested, and shipped. The agent runs multi step tasks against the workflow defined in Module 02, calls tools, retrieves knowledge, and produces structured outputs. This module covers days 7 to 9 and ends with the agent in a controlled production environment.

Output

One working agent that handles a real workflow end to end with structured outputs, error handling, and a documented prompt library.

What it is not

Not a chatbot. Not a demo. The agent ships into a real environment with real data and real users.

04

Module 04: AI Governance

AI Governance covers data handling, audit trails, human oversight checkpoints, and data residency configuration in plain language. It is delivered alongside the other modules, with the final governance pack handed over on day 10. Your compliance, privacy, or risk lead reviews and signs off.

Output

A governance pack covering data flows, residency, audit logs, escalation paths, and human review checkpoints, written for executive sign off.

What it is not

Not a 60 page policy document. Not a generic AI ethics statement. The pack matches the specific agent shipped in Module 03.

What do we need to provide?

The activation runs on three inputs from your side: IT access to the systems the agent needs to read or write, stakeholder time from the workflow owner, and a defined data input the agent will process. We bring the rest.

  • IT access: read or write API access to the document management system, CRM, or practice management platform the agent will touch.
  • Stakeholder time: roughly four hours per week from the workflow owner across the ten day sprint, plus a sign off review in week two.
  • Data inputs: a representative sample of the documents, records, or events the agent will process in production. Real data, not test data.

What does ongoing look like?

After day ten you own the agent. Ongoing support is optional, with most teams choosing a monthly review cadence covering performance, prompt tuning, and the next workflow to onboard. Expansion is scoped one agent at a time rather than as a multi year programme.

Within sixty to ninety days, most clients add a second agent inside the same business unit. By month six, the original team is running two to four agents and a documented internal capability, without ongoing reliance on us for incremental work.

What does it cost?

Pricing is scoped to your team size, workflow complexity, and integration footprint. We publish ranges rather than fixed prices because the same agent for a fifty person law firm is not the same engagement as the same agent for a five hundred person wealth manager.

Scoped to your team size. Discovery call required. We share a scoped proposal with a fixed activation price within seventy two hours of the discovery call.

Book a discovery call

Frequently asked questions

A consulting engagement typically ends with a recommendation document. A Claude Activation ends with a working agent in production. The deliverable is software, not slides, and the timeline is fixed at ten business days for the first agent.
Yes. Your organisation holds the Anthropic relationship and pays Anthropic directly for model usage. We work inside your tenancy so your data, costs, and audit trail stay under your control from day one.
Yes. The default configuration uses Anthropic AWS Sydney endpoints for Australian resident infrastructure. If you require an offshore region for a specific reason, we configure accordingly during Module 04 and document the residency decision for your compliance team.
Yes. The agent, the prompts, the connectors, and the governance pack all sit inside your environment. There is no vendor lock in, no platform licence, and no ongoing dependency on us for the agent to keep running.
Yes. The activation is designed around one workflow as the starting point. Once that agent is stable, most teams expand into a second workflow within sixty to ninety days, often inside the same business unit.
We pre flight system access before day one with a written checklist your IT lead signs off. If access is delayed, the activation timeline shifts rather than the work cutting corners. Most delays are resolved within a week with the right escalation path.