What Is a Claude Activation Sprint?

Most organisations experimenting with AI have the same problem. Promising pilots, useful chat sessions, nothing in production. A Claude Activation Sprint is designed to close that gap in 10 business days.

What is a Claude Activation Sprint?

A Claude Activation Sprint is a fixed scope, fixed timeline engagement that ends with a production AI agent running inside your business, connected to your real systems, handling a real workflow. It is not a discovery project, a strategy review, or a proof of concept. The output is working software and the governance documentation to support it.

The sprint runs across four modules delivered over 10 business days. Each module builds on the previous one. By day 10, one agent is live. By week 6, the firm has a documented AI programme it can extend without depending on external consultants for every increment.

The programme is designed for ANZ organisations with 20 to 500 staff that have identified a specific workflow problem and need to move from intent to production without a 12 month internal project.

What are the four modules?

The four modules are: Knowledge Foundation, Workflow Automation, Agentic Intelligence, and AI Governance. They are delivered in sequence because each one is a dependency for the next.

Module 01: Knowledge Foundation. Before any agent is built, the organisation's relevant knowledge needs to be structured and accessible to the model. This means indexing documents, precedents, internal databases, and process guides into a retrieval layer the agent can query. For a law firm, this is the precedent library and matter archive. For a financial services firm, this is product documentation and compliance frameworks. The Knowledge Foundation module takes 2 to 3 days and determines the quality ceiling of everything built on top of it.

Module 02: Workflow Automation. Workflow Automation maps the specific process the first agent will handle and connects it to the systems the organisation already uses. This is where document management system connectors, CRM integrations, and practice management system hooks are built. The output of this module is a defined agent workflow with tested system connections, not a theoretical diagram.

Module 03: Agentic Intelligence. The agent is built, tested against historical data, and refined. Agentic Intelligence means the agent runs multi step tasks without a person supervising every action. It retrieves documents, applies a decision framework, generates structured output, and routes exceptions to a human reviewer. This is the core build phase and runs across days 4 to 8 of the sprint.

Module 04: AI Governance. No production agent ships without a governance layer. Module 04 covers data handling policies, audit trails, human oversight checkpoints, escalation protocols, and where applicable, data residency configuration. The deliverable is a governance document your legal, privacy, or compliance team can review and sign off on. It is written in plain language, not vendor jargon.

Who delivers the work?

David and Goliath delivers the sprint. The team is small by design: one lead AI systems designer and one integration specialist per sprint. We do not staff projects with graduates supervised by a principal. The person who scopes the sprint is the person who builds the agent.

We work across Anthropic's Claude model suite as the primary model layer. We do not resell licences. The Anthropic API relationship sits with your organisation after activation, so there is no vendor dependency on David and Goliath for ongoing model access.

What do we need to provide?

You need to provide four things: a defined workflow problem, access to the relevant systems, a nominated internal owner, and two to four hours of subject matter expert time across the 10 days.

The workflow problem should be specific. "We want to use AI" is not a workflow problem. "Senior associates spend 12 hours per due diligence matter on document triage before a single issue is surfaced to a partner" is a workflow problem. The more specific the problem definition, the faster the sprint moves.

System access means API credentials or read access to the document management system, CRM, or database the agent needs to connect to. We work with your IT team in the first two days to confirm access and flag any blockers before the build phase starts.

The internal owner is the person who will supervise agent outputs and take responsibility for the programme after the sprint ends. This does not need to be a technical person. It needs to be someone with the authority to make workflow decisions and the context to recognise when the agent is producing useful output.

What does ongoing look like after the sprint?

After the sprint, you own the agent and the governance documentation. David and Goliath offers a monthly retainer for organisations that want ongoing optimisation, additional agent builds, or a standing review of AI system performance. The retainer is optional, not a condition of the sprint.

Most organisations run a second sprint within 60 to 90 days of the first, targeting a second workflow once the team has seen what production AI actually looks like. The second sprint is faster because the Knowledge Foundation and system connections from the first sprint carry forward.

The programme is designed to produce internal capability, not dependency. By the end of the first sprint, your nominated internal owner understands how the agent works, how to adjust its decision framework, and when to escalate a problem to the systems team.

How is this different from a consulting engagement or a vendor licence?

A consulting engagement ends with a report. A vendor licence gives you access to a platform. A Claude Activation Sprint ends with a working agent.

The distinction sounds simple but it matters in practice. A report tells you what you could build. A platform gives you tools and expects your team to configure them. An activation sprint removes both gaps: the strategy is embedded in the workflow design, and the configuration is done by the team delivering the sprint.

The cost structure is also different. Consulting engagements in this space typically run for months and deliver recommendations. Platform licences run on annual contracts with significant implementation overhead. An activation sprint is a fixed price for a fixed deliverable in a fixed timeframe. You know what you are getting before you start.

Who should not run an activation sprint?

The activation model is not the right fit for every organisation. You should not run an activation sprint if you do not have a specific workflow problem identified. Starting a sprint without a defined problem means spending the first week on discovery that should have happened before engagement, and the sprint will not produce the right outcome.

You should also not run a sprint if your IT environment has access restrictions that cannot be resolved in the first two days. Some organisations have security configurations, legacy systems, or procurement rules that make API access take weeks to approve. A sprint cannot absorb that timeline.

Finally, if your organisation is looking for a board level AI strategy document, that is a different engagement. The activation sprint is for organisations that have already made the decision to build and want to get to production, not for organisations still deciding whether to invest.

If you are ready to scope a sprint, start at davidandgoliath.ai/claude-activation. If you have a specific sector in mind, davidandgoliath.ai/claude-activation/start takes you directly to the scoping intake.