Is there a repeatable process at all?
Automation needs something that already happens the same way, again and again. If a task changes every time or happens once a month, it waits.
For early-stage founders
The 5-minute AI-readiness check
You do not need mature operations to think about AI. You do need to know which things are worth automating now, which just need groundwork, and which are honestly too early to touch. Run through this before you spend a dollar on it.
Four questions. Honest answers.
Four questions, 14 boxes. Tick what is true for you today, not what you hope will be true in six months.
Most "AI for startups" advice skips the boring truth: you cannot automate a process that does not exist yet. So this is not a pitch. It is the same first pass we run in our own readiness audit, shrunk down so you can do it yourself over a coffee.
How to use it: go section by section. Tick the boxes that are true for you today. Then read the verdict at the bottom. Be honest. A "not yet" here saves you real money later.
0 of 14 ticked
Is the data actually there?
AI is only as good as what it can read. Information trapped in people's heads or scattered PDFs is the number one reason projects stall.
Is there an owner?
The thing we see kill more automations than bad tech: nobody owns it after launch, so it quietly rots.
Is the pain real and measurable?
If you cannot name the hours or dollars a task costs, it is probably not the first thing to automate.
What you can set up right now.
None of these is AI. All of them make AI cheaper and faster the day you are ready. It is the most useful work you can do here before you even start.
Keep your data in one place.
Clean and centralised from day one. One system of record beats five spreadsheets, and future-you will save weeks of untangling.
Pick tools that have an API.
Check it before you sign up, every time. If a tool cannot connect to anything, it cannot be automated later, however good the interface looks.
Write workflows down as you invent them.
A rough doc while it is fresh is enough. That doc becomes the spec we build against, and writing it usually exposes the steps nobody had actually agreed on.
Name things consistently.
One structure for files, records, and fields. Pick it once and hold the line. Consistency is most of what makes data machine-readable.
Keep a list of what annoys you.
The repetitive tasks your team hates most are your future roadmap. Note them as they happen; nobody remembers them at the start of a project.
Pre-revenue counts. So does a team of two. The point is that none of this needs a budget, and all of it compounds.
Reading your score.
Tick the boxes above and the matching verdict lights up.
You are ready for a real look.
You have a repeatable process, usable data, and an owner. A short audit will find the two or three highest-return things to build first.
Set foundations now, automate soon.
The opportunity is real but a piece is missing (usually the data or the owner). Do the groundwork above and revisit in a few months.
Too early, and that is fine.
Put your energy into the product and into the foundations list above. When you have repeatable operations and volume, that is the moment. We will tell you the same thing on a call if that is where you are.
No deck, no demo
Want a second
pair of eyes?
If you ran this and are not sure how to score yourself, that is exactly what a 30-minute call is for. We will tell you honestly what is worth doing now and what to leave until later, even if the answer is not yet.
Or send us your ticked list and we will react to it