Mangrove logoMANGROVE

Proof

What one expert and AI actually ship.

Every figure below comes from our own measure tool, run against the real codebase: lines of code, pages, endpoints, AI features and commit cadence, counted directly. Client names are withheld; the engineering is not.

PropTech · Client engagement

A market-intelligence platform for letting agents

A public directory and multi-source data engine: it ingests and reconciles property listings across five portals, enriches them against company, EPC and registration data, and grades them with AI. A public site, a data pipeline, and an AI proof-of-concept, delivered end to end.

£25,000 + VAT

fixed price · 8 weeks, one engineer directing AI

A traditional team of 4-6 engineers would take 4-6 months and cost £200K-£400K.

8-16x

lower cost

10-15x

output per engineer

56,348

lines of production code

1,753

lines of tests

15

application pages

6

external integrations

8

AI capabilities

675

commits / 48 working days

DORA Elite throughout: 13-minute median commit interval.

More from the workbench

Products built in-house with the same model. They show the range and scale one expert and AI can hold, costed against what a traditional team would charge to build the equivalent.

Cybersecurity · SaaS

A cybersecurity guidance and posture platform

A multi-tenant SaaS with RAG-grounded AI security guidance, automated risk suggestions, and a large application surface.

prod LOC
104,195prod LOC
test LOC
6,822test LOC
pages
52pages
API endpoints
204API endpoints
AI capabilities
10AI capabilities

Equivalent to a £300K-£600K traditional build. Delivered solo + AI, roughly 5-7x faster.

Cybersecurity · Endpoint

A real-time ransomware detection engine

Filesystem-monitoring ransomware detection, network anomaly detection and key-material preservation, behind a heavily automated CI pipeline.

prod LOC
32,325prod LOC
pages
9pages
endpoints
9endpoints
AI capabilities
6AI capabilities
CI/CD steps
133CI/CD steps

Equivalent to a £300K-£500K traditional build. Delivered solo + AI, roughly 5-7x faster.

Legal-tech · Stealth

An AI legal-workflow platform

Recursive legal-issue decomposition with structured tool use, and case-law retrieval grounded in vector embeddings.

prod LOC
22,172prod LOC
pages
17pages
API endpoints
50API endpoints
AI capabilities
10AI capabilities
commits
209commits

Equivalent to a £200K-£400K traditional build. Delivered solo + AI, roughly 5-7x faster.

How this is possible

  • One senior engineer with AI replaces a coordinated team. No handoffs, no context switching, no ceremonies.
  • AI writes code, tests and documentation alongside features. The engineer directs and reviews, applying 20+ years of judgment to every decision.
  • Proven infrastructure patterns are reused across engagements. No ramp-up, no architecture debates.
  • Fixed price, with scope committed up front.

How we measure

The scale figures (code, pages, endpoints, integrations, AI features, commits) are counted directly from each project's git history and infrastructure by our measure tool. Nothing is estimated.

The traditional-team comparison is benchmarked against published 2026 UK contractor rates and project data. The elite-team-plus-AI midpoint is extrapolated from reported productivity research and should be treated as indicative. Our cost benchmarks are currently based on a small number of completed projects and will be refined as more engagements complete.

AI API spend, third-party vendor spend, and documentation effort are tracked separately and are not included in these figures.

Have a build in mind?

Send us your spec or idea. We'll come back with a clear scope and a fixed-price proposal. No obligation.