Claude Code enablement for 600 Morgan Stanley leaders and engineers
A one-week advanced and hands-on Claude Code program built on spec-driven and test-driven development, covering subagents, agent teams, skills, hooks, governance and security through the firm's own use cases.
Sector: Financial services · Morgan Stanley
- Participants, from C-suite executives to senior engineers
- 600
- Advanced and hands-on tracks
- 1 week
- The working method behind every lab
- Spec + TDD



The brief
Morgan Stanley wanted Claude Code adopted as a disciplined engineering tool, not used as a smarter autocomplete. The audience ran from C-suite executives and senior leaders to senior engineers and product teams, so one program had to serve the people deciding on adoption and the people shipping code with it.
In a regulated financial firm, speed alone is not a win. Every workflow taught had to come with governance, security and permission boundaries the organization could stand behind.
Program design
Mindela's founder designed and led a one-week program with an advanced track and a hands-on track. Participants worked in Claude Code themselves rather than watching slides.
Spec-driven and test-driven development were the backbone. Participants wrote the spec first, had Claude Code turn it into failing tests, then let the agent implement until the tests passed, so output was checked against intent instead of trusted on sight.
Every concept was explained through worked examples and the firm's internal use cases. Labs ran on a sample risk-engine repository built for the program, with trading books, stress scenarios and risk limits, so teams practiced realistic finance workflows on safe sample code.
What we covered
Workflow: spec-first sessions and test-driven loops, from written requirements to failing tests to a passing implementation.
Customize: project instructions and reusable skills, such as a stress-test skill that runs a market-shock scenario against a trading book and reports any limit breach.
Enforce: hooks and permission settings that block unsafe actions and apply team standards automatically, plus the governance and security model for rolling Claude Code out across a regulated organization.
Connect: MCP servers that link Claude Code to internal tools and data sources under controlled access.
Orchestrate: subagents for focused, parallel work and agent teams that split larger tasks across coordinated agents.
Outcome
600 participants, from C-suite executives and senior leaders to senior engineers and product managers, completed the program in one week.
Participants left with a repeatable method for their own repositories: write the spec, write the tests, then let Claude Code implement inside governed boundaries.
It was a really good session. Our people learned a lot, discovered new things in Claude Code, and are now able to use Claude effectively to increase their efficiency and productivity.
Planning a Claude Code rollout?
We run this program for engineering organizations in India, the GCC and the United States, adapted to your stack, use cases and governance rules.
Plan your programMore work
Multi-agent systems for autonomous driving at Uber
How fleets of models coordinate for autonomous driving, with agent communication, simulation-first evaluation and verification before road decisions.
An AI assistant that resolves queries end-to-end
A tool-using AI assistant grounded in the client's own knowledge. It answers, acts, and escalates to humans when confidence drops.