A map of your team's processes and constraints: where Claude creates value, which systems are involved and who decides what.
Claude Forward Deployed Engineer (FDE)
One of our engineers inside your team, building agents, automations and skills with Claude and Cowork.
Preparation → immersion → prototype → production → operations → handover
- phases, from access to handover
- 6
- one of our engineers inside your team
- FDE
- kinds of deliverables: agents, automations and skills
- 3
Base/labs is Certified in the Claude Partner Network. Our team is Claude certified.
Companies that have already chosen Claude and Cowork and need an engineer inside the team to turn the potential into agents, automations and skills.
A Forward Deployed Engineer (FDE) is a software engineer who works inside the customer, builds in production on their systems and follows the result until it works. The model started at Palantir and is now used by AI labs and technology companies.
Companies with Claude and Cowork hit the same wall: the licenses exist, but the use cases depend on knowing the process, the systems and the people. A consultant hands over a document and leaves; a fixed project ends when its scope ends, and the queue of opportunities is left without an owner.
Our offering places a Base/labs engineer in your team, focused on Claude and Cowork: agents, automations and skills. Claude Code comes in when it helps, not as the center of the work.
Licenses with no day-to-day owner
- Claude and Cowork available, but few use cases leave the drawing board
- A consultant who hands over a document and leaves
- A fixed project that ends with the opportunity queue left without an owner
With an allocated FDE
- An engineer inside the team, building in production
- Cases validated with evaluations before the build
- A backlog of new cases that continues after the first delivery
How the FDE works, week by week
The FDE works at your team's pace, with a fixed cadence so the sponsor always knows what was done and what comes next.
Weekly alignment
A short conversation with the sponsor and the case owner: what was delivered, what is blocked and what changes in priority.
Short weekly report
An objective record of progress, decisions made and risks, for the team and for leadership.
Biweekly review
A demo of what already works, with code and evaluation review by whoever will maintain the solution.
Reuse
Part of the time turns what worked into skills, templates and connectors that other departments can reuse.
What we deliver
Claude agents in production on your systems, with human approval at critical points.
Automations of repetitive flows, with quality measured by evaluations.
Corporate skills and Cowork set up in the departments, with the connectors needed.
Periodic results and cost reports, and a prioritized backlog of new use cases.
Documentation and training for your team to take over operations.
How it works
- 01Preparation
- 02Immersion and scope
- Prototype and validation
- 04Build and production
- 05Operations and evolution
- Handover
Preparation
We settle scope, sponsor and an owner for each case, and set up access, environments and the confidentiality and data terms.
Immersion and scope
The FDE observes users, maps systems and constraints and prioritizes cases by value and feasibility, defining the first version and the success criteria.
Prototype and validation
A prototype with real or synthetic data, validated by evaluations with examples from the users themselves, ends in a decision: go, adjust or stop.
Build and production
The agent, automation or skill is built on your systems and goes live gradually, with human approval at critical points and a clear way to switch it off.
Operations and evolution
We track metrics and cost, fix failures and open new cases from what usage reveals.
Handover
Documentation, training and a runbook for your team to operate, and a formal decision: renew, expand or close.
- Prioritized use cases, with the first version and the success criteria defined.
- Agents, automations and skills in production, each with its own evaluations.
- Periodic results and cost reports.
- Documentation and training for your team to take over.
We define metrics per case with you, such as cycle time, rate of cases resolved without intervention and cost per case, always against the baseline measured before we start.
- A sponsor who prioritizes and unblocks, and an owner for each use case.
- Access to the systems, data and environments each case needs.
- Users available to be observed and to validate what is built.
Frequently asked questions
What sets an FDE apart from a consultant?
The FDE writes and ships code on your systems and stays until the solution works, instead of recommending and leaving. They also record what they learn so it can be reused in other departments.
How is the offering contracted?
In two blocks. The preparation, immersion and prototype phases can be contracted as a fixed package; build, operations and handover are quoted separately or contracted by the hour. Hourly contracting from the start is also possible.
Does the FDE use Claude Code?
When it helps build the solution, yes, but the focus is Claude and Cowork. To develop software with Claude Code in your own codebase, with a team and a harness, see Software Engineering with Claude Code in 4 Weeks.
What happens if the case turns out not to be viable?
The prototype phase ends in a decision backed by evidence: go, adjust or stop. You only enter the build after seeing the validation.
What access does the FDE need?
The minimum necessary for each case, defined in the preparation phase, with confidentiality and data terms and LGPD compliance.
Want to talk about this offering?
We start with a conversation to understand your scope. Each offering has a closed scope and price per phase.
