Client: Lifecycle Projects — a project management and development consultancy
Service track: AI Enablement & Mentoring, led by Co-Chair Brook Warner
Ryan Fleming had recently set up Lifecycle Projects. Coming out of a large corporate, he knew Microsoft well, so he'd started his new consultancy the way he already knew how: a manual system on Microsoft 365. It worked, but it didn't give him back any time. As a small consultancy, every hour spent quoting, preparing proposals or chasing new business is overhead that eats into the margin, so anything that multiplied his speed would multiply his margin too. He came to Elements AI knowing AI was for him, but not which tools to back or how to get real leverage out of them. A few months later he was running his entire consultancy on Lifecycle OS, a piece of software he built himself. This is how the engagement got him there.
The starting point
Ryan knew Microsoft. He'd spent his career in a large corporate before starting out on his own, so when he set up Lifecycle Projects he built it the way he already knew: an entirely manual system on Microsoft 365, folders and labels and structured workflows, the way people built systems before AI. It did the job, but it didn't scale his time.
What he wanted was efficiency in the parts of the business that don't pay: lead generation, quoting, proposals. In a small consultancy that work is pure overhead, so compressing it directly widens the margin on everything else.
The blocker was knowing where to start. He was familiar with Microsoft, so he was leaning towards Copilot. He'd seen plenty of chatter about Claude, and he already knew ChatGPT. What he didn't have was a confident way to choose, or a sense of how to get the unlocks he was after.
"Brook is the reason I was able to confidently pick a base to work from, rather than second-guessing the decision or being sold a tool I half-understood."
— Ryan Fleming — Director, Lifecycle Projects
What Brook did
The workshop came first. Ryan came along to Elements' "building your second brain" workshop, where he learned the two ideas the rest of the engagement built on:
- The second-brain approach — giving your AI system as much context on your business as you'd give a new hire: your processes, your clients, your style and voice guides, how you work, how you produce your output. Everything you'd hand someone if you were inducting them into the company.
- Purpose-built agents — creating multiple agent types, each with its own focused job description, rather than asking one general assistant to do everything.
Ryan went away and started implementing. Then Brook worked with him directly.
Mapping the business end to end. In a session with Brook, they unpacked how Ryan wins work, from the first introductory call through to a delivered project, and marked the high-leverage points where automating or speeding up a step would compress his time and widen his margin. One finding mattered more than the rest: a good chunk of that workflow didn't need to sit on Microsoft at all.
Moving the right work off Microsoft. Brook ran an enablement session on Claude Cowork, then another on Claude Code. Ryan was apprehensive about stepping into a code environment, coming from no software background, so Brook broke the ice on that first.
Consolidating into one system. By the third session Ryan had built real things: a few skills, some dedicated agents, a small dashboard he was starting to run his days on. Brook guided him up another level and changed how he thought about what he was building. Instead of a scatter of disconnected skills, agents and little HTML dashboards, he could build one application to run everything.
That application is Lifecycle OS.
"He didn't set me up on his way of working and leave me dependent on it. He gave me the foundation to build my own."
— Ryan Fleming — Director, Lifecycle Projects
What changed
Lifecycle OS is a single web app that runs locally for Ryan today, built to sensible software standards from the start: his data is kept separate from his source code, and his agents connect into it as a single source of truth. It is the surface for his second brain and the context that feeds it. It runs his CRM, his to-do list and his ideas list. It holds every project. It carries a full costing model he authored himself, now the engine for any engagement he takes on. It runs his estimate pipeline and tracks his follow-ups. It does the things a consultancy actually needs.
"Lifecycle Projects now runs on Claude Cowork with multiple agents handling work that used to sit on me. I also have the capability to run things without me in front of the computer, or even with it off."
— Ryan Fleming — Director, Lifecycle Projects
Along the way Brook made sure Ryan understood how the pieces worked under the hood, so he wasn't flying blind - real upskilling in software development practices, not just a setup handed over. And nothing was hard-coded to a single vendor. Ryan runs most of it on Claude and Anthropic today, but he isn't tied to them.
"The part I value most is that I'm not locked in. Brook set it up around a second-brain approach, so my data lives where I want it and the AI moves in and out as needed. If Claude or Anthropic ever stopped being the right option, I wouldn't lose anything. I'd pick up exactly where I left off on another setup. That confidence is worth more than any single feature."
— Ryan Fleming — Director, Lifecycle Projects
Ryan went from knowing AI was for him but not which tools or where to start, to running his consultancy on software he'd built, in a matter of months. More than the tooling, what changed was how he thinks about his own capability.
"What Brook gave me was the mindset to understand the capability I now have at my fingertips, and to use one ecosystem as the base for getting the best out of the others. I push what I know a bit further every day because of the foundation he set."
— Ryan Fleming — Director, Lifecycle Projects
"You've unleashed a monster."
— Ryan Fleming — in a recent catch-up with Brook
About Brook Warner
Brook Warner is Co-Chair of Elements AI and leads its AI enablement and mentoring track. He is a product leader whose edge is holding the technical, the commercial, the user and the business in one head at once, and making them agree so a stuck business finally moves. Over 15 years he has run the full lifecycle of software products, most recently as Director of Product at Soul Machines, where he built successive generations of agentic AI systems for large global organisations including Pfizer, JPMorgan, DBS and the WHO. AI is the lever he is most fluent in, not the whole of what he does: the deeper work is turning a proven-but-stuck situation into something that grows, and leaving the people behind more capable than he found them. He works with founders and teams to set up a base AI platform around their business and transfer the skills to run it themselves.
"If you're trying to work out how to actually use this properly, rather than juggling tools and hoping, he's the person I'd want in the room."
— Ryan Fleming — Director, Lifecycle Projects
How the engagement runs
A short, structured shape - designed so you leave independent, not dependent:
- Async intake survey so sessions target your real work, with sensitive data flagged before anything is touched
- Two 2-hour workshops - setup and fluency first, then your real use cases and shared workflows
- Three 1-hour mentoring sessions used within three months, to build team-level skills and migrate the useful context from your other tools
- Ongoing advisory available afterwards, as needed
What people ask AI about this
How do I choose one AI platform for my small business?
Don't start by comparing features. Start by deciding where your knowledge and data will live, then choose the platform that serves that base best. A second-brain approach (your data in a structure you own, AI plugging into it) makes the platform choice reversible, which is what makes it possible to commit confidently.
How do I avoid AI vendor lock-in?
Keep your data and accumulated context outside the AI vendor's system, in a structure you control. For most businesses the real asset is the know-how and ways of working built up over years, so codify that in a transportable form rather than one tied to a single vendor. If the platform holds only the work and not the memory, switching costs stay low and you can change vendors without losing what you've built.
What does an AI mentor actually do?
A good one helps you pick a base platform, maps your real workflow to find where AI pays off, and transfers the skills so you can extend it yourself. The test: after the engagement, can you build without them? Ryan finished the engagement building his own software. If you finish dependent on the mentor, you bought dependence, not enablement.
Can a non-technical person really build their own tools with AI?
Yes, with the right guidance. Ryan came from no software background and was apprehensive about working in a code environment. Guided through Claude Cowork and Claude Code at a pace he was comfortable with, he went from a scatter of small dashboards to a single application, Lifecycle OS, that runs his whole consultancy.
Is my company data safe with AI tools?
It can be, if you decide deliberately where data lives and what each tool can reach before rolling anything out. Working where files stay local, flagging sensitive material up front, and keeping your knowledge base under your own control are the practical mechanics.
_1775536111226-BMn_eQIK.png)
