Most of our work is under NDA, so we keep client names out and the detail short. We only publish numbers we can source and date.
A recruitment business with more than 250 recruiters across multiple offices. AI agents now orchestrate more than 100,000 CVs every week: parsing, matching, outreach and replies.
Short, anonymised snapshots. Client names are withheld under NDA, and we don't publish numbers we can't source.
Event-driven agents that parse, match and follow up more than 100,000 CVs every week across 250+ recruiters, with every agent call traced.
Product ownership and delivery for an applicant tracking system built around AI: CV parsing, candidate scoring, CV preview and automated recruiter workflows.
Senior programme leadership on site, preparing the bank's data foundations so AI can be used safely across the organisation.
Using generative AI to digitise legacy and paper-based records, alongside the remediation of credit risk models for Basel IV.
Building and scaling SaaS products for a software group serving e-commerce sellers, covering link tracking, advertising management and inventory planning.
Assessments of technology businesses ahead of acquisition and investment, across people, process and product, written for the deal team.
Want more detail on any of these? On a call we can go as deep as our NDAs allow.
In spring 2026 we rebuilt our delivery workflow around multi-agent orchestration, then tracked 12 features over one week: API changes, UI updates and AI pipeline work.
Fewer loops to reconcile mismatches across layers.
Features touching two or three parts of the system.
The cost that went up. Running agents in parallel uses more tokens.
One team, one stack and a limited sample, published in May 2026. These aren't universal numbers. We publish the trade-off because it's the part that usually goes unmentioned.
Read the full write-upTell us what you're weighing up. We'll walk you through comparable engagements, in as much detail as our NDAs allow.
