This month I cut hundreds of pounds from my monthly software bill. I didn’t negotiate discounts or downgrade plans. I built my own, using OpenClaw.
Yes, my LLM token costs are up. But that’s temporary, build-phase spending that drops once the systems are running. The SaaS subscriptions were forever.
Renting someone else’s constraints
I’ve spent fifteen years in recruitment technology. I watched as the industry moved from offline processes to cloud software, trading one set of limitations for another. Every SaaS product was built for the median user, which meant workarounds for workflows the product team never considered.
Feature requests sitting in backlogs. Integration gaps. The tax of context-switching between six different dashboards. SaaS businesses became gatekeepers to productivity, especially in recruitment, locking businesses into expensive, multi-tenanted, modular, multi-year agreements.
I built my last recruitment business on Airtable because I was tired of renting someone else’s constraints. My mantra of “build what you need” was true then. It’s truer now than ever. And it just received a massive upgrade.
From plain English to a working system
This isn’t about writing code. I don’t code. It’s about describing what you need in plain English and iterating until it works.
Here’s what I built in the last 10 days:
Sentinel Talent: Tiding started as an experiment to build a fully autonomous recruitment system. It built the entire brand and business first: positioning, website, terms, the lot. Everything agent-first. Then the CRM with agent mail and enrichment pipelines pulling from Apollo, Hunter, Bouncer, Companies House. Ten layers deep with waterfall logic. Sentinel has now commenced outreach and I’m expecting calls in the diary next week. We’ll be able to plug in another human operator, enabled by all these systems.
Update, October 2026: Tiding and Sentinel Talent were experiments at the time, and both have since closed as projects. The Sentinel Talent site is still live, as a record of what the agents built.
Only then did it extend to business development as a service, plugged back into Bolt Search. The BDaaS is less than one week old and already driving leads. It already does more than the off-the-shelf stacks I’ve used, because it carries all the nuance from 15 years in one sector. Trust, relationships and context, written down and put to work.
Bakery business: I’m an operating partner in a bakery business. We now have custom stock management between retail and production, plus recipe costing with PDF ingestion from supplier emails. We’re replacing a £3,000 recipe management tool that sits unused because the UX makes setup too onerous.
None of this required a developer. Just clear thinking, iterative prompts, and willingness to debug when the first attempt failed.
This was just the last 10 days, whilst also running agency processes, calls, board meetings, interviews and more. A lot of it done from my phone.
The infrastructure flip
Watching Juno and Felix build companies designed to run on agents shifted my thinking. They design for agent autonomy rather than human productivity. They inspired our version: Zac at Tiding.
The question changes from “how do I do this faster?” to “how do I remove myself as the bottleneck?”
This changes how you build:
- APIs over dashboards
- Event-driven over manual
- Composable over monolithic
- Natural language interface over rigid UX
The mega-enabled founder
Jack Dorsey’s company Block announced layoffs as they shift to AI-driven operations. I’ve spotted the “hire, train, deploy” model for AI operators growing. AI apprenticeships and training courses are proliferating, mostly thin veneers of content jumping on the trend rather than driving ROI, with some exceptions of course. Everyone has an “AI service” now.
But something else is happening. Founders who understand their workflows intimately can now build exactly what they need. Exact fits, not approximations or workarounds.
The premium shifts to judgment, context and taste, the things that can’t be automated. If you’re selling implementation, you’re racing to the bottom. If you’re applying judgment through personal systems, you’re building leverage. The true goal is ROI.
The real cost
My LLM bills are higher this month. But they’re variable. I pay for what I use, and I’m constantly optimising. The SaaS subscriptions were fixed costs for features I barely touched.
The real cost is clarity. Building forces you to understand your workflows precisely. You can’t describe vague requirements to an AI and expect good results. This is the hidden tax: thinking time, specification, iteration.
For some, this cost is too high. Keep subscribing. But for founder-led businesses where the founder knows exactly what they need? Building is now the better option.
What next?
I’ve achieved more progress in February than in the previous two years. Capability I’ve wanted for years, partly built towards with no-code, now fully realised. The race is on.
The founders who win market share, attention and relationships will be the ones using customised tech, not the ones replaced by it. Personal software is here. Everyone becomes mega-enabled.
I’ll be writing more and sending my newsletter again, now that I can go from a thought to a published piece faster than ever.
If there’s a subscription in your business you’d rather own, like the stock, costing or CRM workflows above, book a 30-minute call and we’ll look at whether it’s worth building.
