AI-Augmented Engineering: 12-Week Programme for Engineers, Architects, and Data Engineers

What it is: a 12-week, cohort-based programme for software engineers, architects, and data engineers that builds toward production-ready AI workflows. It covers agentic development tools (Claude Code and equivalents), context and repository setup, spec-driven development (specify what the code must do and how it will be verified, let the agent implement, check the result), testing and review patterns for AI-generated code, and the security and governance constraints the organisation requires. Teams finish with a working AI-augmented workflow deployed against a real codebase.

Why does engineering need its own AI programme?

Because the shift for developers is different in kind from the shift for knowledge workers.

For most employees, human review of AI output stays constant: the model drafts, the person verifies and owns the result. For software teams using agentic tools, the work itself moves up a level. Engineers increasingly specify what the code must do and how it will be checked, let the agent implement, and spend their judgement on architecture, review, and verification. That is a change in method, tooling, and team process, and a general-purpose AI fluency session does not address it.

Who

Software engineers, architects, data engineers; teams of 8–25

Format

12 weeks; weekly live sessions with hands-on labs against the team's own repositories

Tools

Claude Code and comparable agentic development tools; the organisation's CI, review, and security tooling

Prerequisites

Working engineers; no prior AI tooling experience required

Output

A production-ready AI-augmented workflow per team, with tests, review process, and governance documented

Support

Instructors with production engineering backgrounds; AI coaches between sessions

Agentic tooling foundations

How agentic coding tools differ from autocomplete; setting up Claude Code or equivalent against a real repository; what the agent can and cannot see.

Context engineering for codebases

Repository instructions, documentation the agent reads, conventions files, and how to scope a task so the result is specific rather than generic.

Spec-driven development

Writing specifications that state behaviour and acceptance criteria; letting the agent implement; verifying with tests the agent did not write alone.

Review, testing, and verification

Review patterns for AI-generated code, test generation and independence, catching the failure modes that look correct.

Security, data, and governance

What code and data may enter the tool, secrets handling, licence and provenance questions, and audit trails the organisation requires.

Team workflow and production

Integrating the new method into sprints, CI, and code ownership; deploying the team's capstone workflow and documenting it for the wider organisation.

How is this different from vendor product training?

Product training teaches a tool. This programme changes a team's method and leaves evidence behind.

Method over features. Spec-driven development, verification, and context engineering transfer across tools as the tooling changes.
Your codebase, your constraints. Labs run against the team's own repositories within the organisation's security policy.
Governance built in. The security and data module is designed with the client's security and risk teams, not appended afterwards.
Measured. We agree in calibration what to measure (cycle time on the targeted workflow, review findings, test coverage, adoption at 90 days) and report it.

How does it fit with the rest of an enablement programme?

It is the technical persona's track in a persona-based architecture.

Most organisations now run a multi-tool environment with a general-purpose assistant for all employees and agentic tools for engineers and a few other roles. A persona architecture (leaders, technical workers, business workers) with a business goal per persona is what makes that coherent. This programme is the technical track; the AI Impact Lab and executive enablement serve the other two.

Software development in which engineers specify behaviour and verification, agentic tools such as Claude Code implement much of the code, and the engineer's effort shifts to architecture, review, and testing. The programme teaches that method and deploys it on a real workflow.

A method in which the engineer specifies what the code must do and how it will be verified, rather than how it should be written, lets an AI agent implement it, and checks the result against independent tests.

Claude Code and comparable agentic development tools, together with the organisation's existing CI, code review, and security tooling. The method transfers across tools; the specific tool is confirmed in calibration.

Twelve weeks, with weekly live sessions and hands-on labs against the team's own repositories, ending in a deployed capstone workflow.

Yes. A dedicated module, designed with the client's security and risk teams, covers what code and data may enter the tool, secrets handling, licence and provenance, and audit requirements.

Working software engineers, architects, and data engineers in teams of 8 to 25. No prior experience with AI tooling is required.

Featured:

Turn AI anxiety into action

Featured:

Bolstering technical skills development in MENA

Featured:

Explore new digital career pathways

Featured:

Take action on generative AI

Featured:

Hear what the experts have to say

AI-Augmented Engineering

Why does engineering need its own AI programme?

What does the programme cover?

Agentic tooling foundations

Context engineering for codebases

Spec-driven development

Review, testing, and verification

Security, data, and governance

Team workflow and production

What do teams produce?

How is this different from vendor product training?

How does it fit with the rest of an enablement programme?

Frequently asked questions

Claude training for enterprises

AI Impact Lab

Claude or Copilot? How enterprises decide

Enterprise AI Enablement overview

Bring agentic development to your engineering teams