Prompt engineering and custom assistants
Custom GPTs, Claude projects, and shared prompt libraries - tuned, evaluated, and version-controlled for your team.
Prompts are the new SOPs. We build a shared library of evaluated, version-controlled prompts and custom assistants so the whole team works from the best version, not the one each person typed at 9am.
What we deliver
From a single high-value workflow to a library of dozens, with evaluation harnesses so prompts don't silently drift.
- Custom GPTs and Claude projects scoped to a job
- Shared prompt library with version control
- Evaluation harnesses and regression testing
- Team training on prompt design and iteration
- Integration with internal tools and knowledge bases
More in Efficiency
AI strategy audit
A clear, prioritised plan for where AI will actually move the number.
RPA (robotic process automation)
Automate the rules-based work nobody enjoys doing.
Agentic AI
AI agents that complete multi-step work end-to-end.
Voice AI
AI receptionists, callbacks, and outbound voice agents.
Internal AI assistants
AI assistants trained on your data and policies, for your team.
AI analytics and dashboards
Turn scattered data into live dashboards your team can act on.
Workflow automation
Automate the manual, cross-system work that's costing your team hours every week.
Reporting dashboards
Live dashboards your team opens every morning, not month-end static reports.
Integrations
Connect the tools you already pay for so data flows without manual re-entry.
Fractional advisory
Fractional Head of AI, CMO, or growth advisor on a monthly retainer.
AI training and enablement
Hands-on training programmes for marketing, sales, and ops teams.
Operating model design
Redesign teams, roles, and ways of working for an AI-augmented business.
Business strategy and operating model
Strategy refresh and operating model design built for the modern, AI-shaped market.
Transformation advisory
Independent advisory on large change programmes, from business case to delivery.
Frequently asked questions
It is the practice of designing the instructions you give an AI model so it produces useful, consistent results. Good prompts plus a clear workflow are what separate a team that gets real value out of AI from one that gets random output.