Build an AI tool stack of five, not fifteen
Drafted through my n8n + AI pipeline, edited by me.
The typical small business now runs about five AI tools, and most plan to add more. That is how you end up paying for fifteen overlapping subscriptions that each do a slice of the same job. A good AI tool stack is small on purpose.
The five jobs an AI tool stack needs
Almost every useful AI tool falls into one of five jobs. Cover each once, with the best tool for it, and you are done.
- A general assistant for thinking, drafting, and research.
- An automation layer to move work between your tools.
- Content and marketing, if that is where your growth is.
- Customer response, for support or the front desk.
- Notes and capture, so meetings and calls become records.
Before and after: a sprawling stack of overlapping AI subscriptions versus a lean stack of five tools, one per job, feeding one source of truth.
- Three tools that all draft content.
- Two that half-automate the same flow.
- Subscriptions nobody remembers buying.
- Data scattered across all of them.
How to cut the sprawl
- List every AI tool you pay for, and the job it does.
- Where two tools do one job, keep the better one and cancel the other.
- Before adding a tool, name the job it does that nothing you own already covers.
- Wire the survivors into one source of truth, so they share data instead of fragmenting it.
Bring me the AI tools you're paying for, and I'll tell you which five I'd keep and which ten I'd cancel.
Building something this should run inside?
Book a systems callKeep reading
AI
Why your AI pilot never ships (and the fix)
Your AI pilot demos well, then stalls forever. Why most pilots never ship, and the smallest AI pilot you can actually put into production.
automation
The tasks you should never automate
Some tasks you should never automate: the first apology, a real judgment call, a relationship touch, a true one-off. Here's the line to draw.
spreadsheets
Spreadsheet or system: when to graduate
Whether you need a spreadsheet or system comes down to a few honest questions, not ambition. Here's the line, and when to cross it.