The Founder as Proof
Claude Cowork at work in the business, a living record. This is not a forecast and it is not a pitch. It's a case study of my own business, measured, four months in and still moving.
Christopher Duffy, Founder, IgniteAI Solutions · As at 16 June 2026
3.6x
Spring revenue vs January (conservative)
5.3x
June run-rate vs January
489
Deliverables in four months
9
Automations running without me
Since February, Claude Cowork has been working across the business. What follows is what changed: the revenue, the work, and the operating layer that now runs underneath it. The revenue is read straight from Xero. The time saved is modelled and measured from a four-hour deep overnight audit, and I say so where it appears. Nothing here is a number without a source.
The work a consulting team would scope over weeks, a discovery that scores dozens of opportunities and lands the hard findings, hyper-personalised, now ships in days. That gap is the case.
The Business Result
My business started in January 2025. I went full time in May 2025, and it took six months to start generating regular revenue. As at January 2026, we were running at a loss. In February 2026, Claude Cowork went live and I started moving the business over.
Things started to improve after just four weeks. Revenue sat flat through the run-up. From March it stepped up and held. Indexed to January, monthly revenue ran about 3.6 times higher across the spring, and June reached a run-rate near 5 times the January base.
The dashed line marks the March to May plateau, about 3.6x. June shown at run-rate.
Source: Xero.
This is correlation, not a controlled test. Revenue in this period has more than one driver, a scaling delivery team and the partner network among them. Claude is one driver among several. February reads low and is kept out of the headline, because anchoring the multiple there would flatter it.
The Work Behind It
Counting end products only, with versions stripped out, the business produced 489 distinct deliverables in four months. The first four weeks show nothing: setup and learning. Output begins in mid-March and climbs from there, with the busiest build weeks reaching into the eighties.
First four weeks: setup and learning. Source: dated work files, 16 Feb to 16 Jun.
The two spikes are large engagements landing at once, full discovery packs and overnight builds, rather than a uniform week of steady output. The shape is a ramp, not a switch flipped on day one.
The climb came from building processes and workflows first, then an operational flywheel of continual improvement. Learning from people like Allie K Miller, Hayden Miyamoto, Nate Herk and Brock Mesariah. In February 2026, nothing ran without me. By mid-June, 9 of them run the recurring work: the briefings, the triage, transcript harvesting, backups and edge sweeps. The system now checks its own output.
From zero in February to nine by mid-June.
Each one freed the time that built the next. That is where the compounding actually sits, in the stack of working machines rather than a single headline number.
The Operational Flywheel
A signal lands and feeds the knowledge base, the daily brief sorts the day, skills and the bench build the work, it passes the quality and voice gate, the branded output ships and the CRM updates, then every output sharpens the skills for the next pass. I sit in the middle: one hour a day, the review gate. Nothing client-facing or financial ships without my yes.
1
CAPTURE
Every call feeds the knowledge base
2
TRIAGE
The brief sorts the day
3
BUILD
Skills draft, the bench delivers
CHRIS
1 hour a day
the review gate
4
QA
Advisory board, voice gate
5
DELIVER
Branded output ships
6
LEARN
Outputs sharpen the skills
Take two from the portfolio. The CSA environmental discovery scored 36 opportunities and landed every hard finding. The Aardvark Safaris discovery produced a full findings report for a luxury tour operator. Each is the kind of engagement a traditional consultancy scopes as a small team over several weeks, at a cost to match. Each was produced here by one architect, in days.
Cost comparator
A consulting day rate for one of the big four runs £1,200 to £2,500 a day for manager and senior consultant roles. Multiple weeks of scoping and interviews comes to £40,000 to £50,000. My rate is £5,000 for the whole discovery engagement.
The most recent month nets back a measured 5 to 9 hours a week, modelled from signed-off baselines minus attended time. Working backward over six months, the compounding time saved is closer to 34 hours a week. That is a founder running a business on the most fluent operating system we have ever had, and the worst it will ever be.
Ignite OS · plugged in, running, guarded
Governance layer
Root CLAUDE.md and MEMORY.md, inherited by every workspace. The Lens · voice always on · scope before draft · 3-pass routing · memory overrides rules.
Content & Thought Leadership
Public writing as Chris
The Brain
Knowledge, ingestion, research. Self-improving
CBFY
Productised builds and offer
Skills Engineering
Build and tune skills
Business Operations
Run the company
Client Delivery
Discovery and delivery · ~15 projects
The Firewall
Draft only · money read only · records ask first · nothing leaves without my yes
Deliverables
Branded DOCX + PDF · discovery reports · LinkedIn & Substack · live Vercel sites
Amplifiers
Enrichment inputs · never method
What it proves
The output of a consulting team can be produced by one person with a system, quickly, and the business grew strongly over the same four months.
What it does not
It does not prove Claude alone caused that growth. The honest claim is correlation with a clear mechanism. The revenue trend is real and read from Xero. The time saved is modelled and labelled.
The 3.6x is my conservative reading. The bolder multiples sit in the chart for anyone who wants to check them.
The same operating system I build for clients, I built for myself first. The standard way in is a SPARK Discovery. From £5,000, we baseline your culture and map where AI fits your real work before a penny goes on the build.