The work, in detail.

—20 launch markets, 13 fully localised languages, AI-translation QA gated by parity checks in CI

—Combinatorial personalisation: 20 markets × 449 species × 8 climate archetypes × 4 soil pH bands × 3 organic-matter bands × 11 garden goals × 4 experience levels — over 37 million distinct advice contexts, before live weather modulates them further

—Live birdsong recognition across 1,173 species, with per-market geo-priors lifting precision from 0.63 to 0.89

—49 ledgered capabilities on Supabase/Postgres — 246 migrations, 31 edge functions, row-level security throughout

—AI features metered per user, with spend caps enforced in the database

—30+ governed agent roles running support, moderation, content, finance and go-to-market — with hard never-autonomous red lines

—~640,000 lines of shipped code, 899 automated test files, three production repos

—Blueprint, build, embed — governance included

—Adversarial review on money-adjacent changes; verification gates before any “done”

—Reference build: Gardn’s AI-operated back office, demonstrated live to clients

—£1.3m → £3.25m → £4.8m revenue over three years — reaching a £6m annualised run-rate by month 21, with a best month of £900k

—Five institutional rounds — Foresight ×3, Haatch ×2 — plus angels

—50+ team across two Nottingham fulfilment centres; founder-led sales against incumbent 3PLs

—FEBE Growth 100 · Barclays business accelerator 2023

—Turnover grown to a projected £12m — 401% growth on 2017 — with headcount tripled

—Pandemic pivot supplying PPE to councils and the NHS; built and launched the Trusted PPE brand