Agentic growth systems
Full-funnel growth, executed by agents I build.
I'm a statistician and growth operator. Paid, web, organic and paid social, and lifecycle all run on one loop: hypothesise, ship, measure against revenue, log, sharpen. The repetitive execution is done by agents I build and supervise, so experiment velocity scales with compute, not headcount.
One loop, every surface
execution layer = agents- 01
Hypothesise
An agent reads the joined data layer - analytics, ads, CRM, support - and proposes the change with evidence, confidence, and expected impact.
- 02
Ship
The variant goes live. Success metric and measurement window agreed up front, not retrofitted.
- 03
Measure
Against revenue and resolved problems, not clicks alone. One identity layer ties every touch to the deal.
- 04
Log
Wins and losses both log. A loss tells you which model was wrong and sharpens the next call.
- 05
Sharpen
Each pass pulls from prior results. The surface gets sharper every round instead of degrading silently.
Human-in-the-loop, with a guard. Hypothesis selection, copy approval, and ship-to-production are human gates. Every variant is checked against the current value prop and target audience before it goes live.
What runs on the loop
Paid media and ad accounts
Google, Meta, LinkedIn. Creative and targeting refresh on the same data layer; underperforming spend surfaced before it leaks.
~£50k+ managed · 3x conversion uplift on a bespoke in-product placementWebsite and CRO
Landing pages, on-page experiments, and SEO / AEO run continuously rather than as one-off redesigns.
~25k pages consolidated across 4 brand estatesOrganic and paid social
Each post's settled performance measured; the playbook rewritten from what actually landed, not what felt right.
one launch: ~250k impressions · ~2,000 MQLsLifecycle and email
An agentic engine: list segmented, campaign assembled and QA'd, sent on approval, engagement measured to sharpen the next send.
~20% download rate on a lifecycle upsell assetProof it compounds
1,000+
firms onboarded to an agentic product, in <18 months
~10x
ARR on a product across channels
<48h
to build an interactive product walkthrough, mid-campaign
~500/mo
ICP contacts sourced by an AI pipeline
Agents in production across
Questions
I own full-funnel growth as an operator and build the AI agents that run the repetitive execution. Paid media, website and CRO, organic and paid social, and lifecycle email all run on one loop - hypothesise, ship, measure against revenue, log, sharpen - with agents doing the execution and a human approving every variant.
An agency briefs work out and bills retainers. I embed as the operator who builds the system: the data layer, the agents, and the experiment loop stay with the company. Experiment velocity scales with compute rather than headcount, and every result compounds into the next hypothesis.
No. It is human-in-the-loop with a guard. Hypothesis selection, copy approval, and ship-to-production are human gates. Agents read the joined data, draft variants, and measure outcomes; a person checks every variant against the current value proposition and target audience before it goes live.
Highlights: 1,000+ firms onboarded to an agentic product in under 18 months, roughly 10x ARR on a product across channels, trial-to-paid conversion tripled from 3.1% to 9.4%, and an inbound engine taken from zero to about 100 leads a month on £0 paid budget.
AI-native startups that want growth run as experiments, and VCs or incubators who want an operator dropped into the portfolio companies where growth is stuck - someone who builds the growth systems rather than just briefing an agency.
Deploy this across your portfolio.
An operator who builds the growth systems, not just briefs an agency - dropped into the companies that need it most.