Evidence and operating lessons
Case studies
Case studies are where the consulting proposition becomes concrete: the
problem, the constraints, the architecture, the organisational choices
and the outcome.
These are initial public placeholders. They name the likely themes and
structure, but avoid invented metrics or confidential detail.
Problem
Recruitment workflows involve high-context matching, communication, follow-up and judgement spread across many small operational steps.
Context
A production recruitment setting with human-in-the-loop decision points, operational constraints and a need to keep candidate and client interactions accountable.
Approach
Combine agent orchestration with explicit workflow boundaries, human approvals and operational visibility rather than treating the whole process as a single autonomous black box.
Technology
LLM systems, agent orchestration, email automation, semantic search and production operations tooling.
Outcome
Publicly shareable metrics are not listed yet. The case study is structured so verified outcomes can be added without changing the page.
Lessons
Agentic systems need interfaces, review points and operational controls as much as they need model capability.
Problem
Coding agents create leverage only when their work can be assigned, reviewed, tested and integrated without losing engineering control.
Context
A software engineering environment experimenting with named agent roles, persistent task ownership and multi-agent coordination.
Approach
Treat agents as participants in an engineering system: explicit roles, durable tasks, review loops, MCP-enabled tools, testing and clear ownership boundaries.
Technology
Coding agents, MCP, multi-agent coordination, code review workflows, test automation and engineering controls.
Outcome
The public version should only include measured effects once they have been verified.
Lessons
The useful question is not whether agents can write code. It is whether the surrounding harness makes their work inspectable, reversible and worth trusting.
Problem
Recruitment platforms have to connect product workflow, enterprise SaaS requirements, large-scale data, reliability and user trust.
Context
A product and architecture setting with scale, delivery pressure and the need to balance business outcomes with maintainable systems.
Approach
Use architecture, product judgement and team design together, so platform decisions support the operating model rather than becoming isolated technical choices.
Technology
Enterprise SaaS architecture, large-scale data systems, product delivery and platform foundations.
Outcome
Specific metrics and commercial details should be added only where approved for publication.
Lessons
Technical architecture and organisational architecture usually fail or succeed together.