Is my business ready for AI? What readiness means for a small business
Most AI readiness frameworks are written for large companies: data strategy, governance boards, model infrastructure. A small business needs none of that to start. It is ready for AI when an agent can take a recurring task off someone's plate, and that comes down to four things:
- Repeatable processes. The task follows the same steps each time, even when the details change. If nobody could write it down as a checklist, an agent cannot follow it either.
- Work that lives in software. The inputs arrive in email, a spreadsheet, a CRM or an online store, and the result goes back into one. Paper, phone calls and site visits stay out of reach.
- Enough volume. A task that takes ten minutes a month is not worth setting up. A few hours a week of repeat work is where the time saved starts to show.
- A team willing to try. Someone has to own the first agent, check its output for a couple of weeks and say what to change.
This AI automation assessment checks each of the four. It is meant for businesses of one to a few hundred people that want to hand recurring tasks to AI agents, not for companies planning to train their own models.
The four signals and how each is scored
The score adds up five parts. Three come from Jev, a decision model by TypeSafe, which reads your business description and task list and rates each signal from 0 to 2. It returns a probability for each level, and the page uses the weighted average, so a 1.8 counts for more than a 1.2. The other two come straight from your answers.
- Repeatable process, up to 35 points. 0 is ad hoc, different every time; 1 is routine but done from memory; 2 is the same steps every time. It carries the most weight because an agent can only follow steps that exist.
- Work that lives in software, up to 40 points. Up to 30 for how much of the work already happens in software (0 mostly offline, 1 mixed, 2 almost all), plus 2 points for each tool you use, counted up to five tools, for 10 more.
- Volume, up to 20 points. 0 is an hour or two a week, 1 is a few hours, 2 is a day or more. More repeat work means more time back.
- A team willing to try, up to 5 points. Cautious 0, comfortable 3, early adopters 5. It counts for little because a cautious team can still start with an agent that drafts and waits for approval; caution slows the start rather than blocking it.
Score = 35 × process ÷ 2 + 30 × software ÷ 2 + 20 × volume ÷ 2 + 2 × tools (up to 5) + comfort (0, 3 or 5)
Team size goes to the model as context and does not add or remove points. The total puts you in one of four bands:
- Under 40, Getting ready: fix the basics first, usually written-down steps or moving work into software.
- 40 to 64, Ready to start: try one agent on one task, with a person approving each output.
- 65 to 84, Ready now: one or two agents on recurring tasks should save real hours.
- 85 and above, Primed: the question is which task to start with, not whether to start.
AI readiness checklist: 10 things to have in place
Tick these off before you hand a task to an agent. Most small businesses have six or seven already.
- You can name three tasks that come round every week.
- Each of those tasks follows roughly the same steps every time.
- The steps are written down somewhere, even as a rough checklist.
- The inputs arrive in software: email, a form, a spreadsheet, an online store or a CRM.
- The output goes back into software: a sent email, an updated sheet, a posted message.
- You know what starts each task: a time, or an event like a new order or an overdue invoice.
- Together the tasks take at least a few hours a week.
- A mistake would be easy to spot and fix, or a person can approve each output before it goes out.
- One person owns the first automation and will check its output for the first two weeks.
- You know which data must stay private, such as client financials or health records, and who is allowed to see it.
What can I automate first? Common first automations by business type
The first automation that sticks is usually small and dull: a reminder, a report, a sorted inbox. These are the ones that come up most often by type of business:
- Accounting and bookkeeping firms: chasing clients for statements and receipts, categorising transactions, invoice reminders and monthly client reports. More in AI agents for bookkeepers and AI agents for accountants.
- Online stores: order status replies, review replies, stock alerts, returns and a weekly sales summary. See AI agents for Shopify store owners.
- Recruiting firms: screening applications against the brief, scheduling interviews, candidate updates and weekly pipeline reports. See what AI recruiters screen and what they miss.
- Real estate agents: quick replies to new leads, viewing schedules, listing updates and follow-ups after a viewing. See AI agents for real estate agents.
- Agencies: weekly client reports from ad and analytics accounts, status updates and chasing approvals. See AI agents for marketing agencies and the weekly KPI report agent.
- SaaS startups: support ticket triage, trial follow-up, failed payment reminders and weekly metrics in Slack. See AI agents for SaaS.
For published results with the numbers behind them, the AI agent ROI case studies roundup collects them in one place. To see how far other businesses have got, the AI agent adoption survey index grades every adoption survey on its sample, sponsor and how it defines an agent.
The limits of a self-assessment
A self-assessment only knows what you tell it. It cannot see whether your apps let outside tools connect, how tidy your data is, or what your client contracts say about where their data may go. It also scores the tasks you list, not the whole business: leave out a big repeat task and the score will miss it.
Treat the score as a starting point. A high score means it is worth trying an agent on one task now; it does not mean every task should be automated. A low score is a to-do list, and the next step under each bar is where to begin. If you handle health, legal or financial records, check your obligations before any tool touches them.
For a closer look at a single task, the Can AI do this task? checker scores one task in detail and works out what your time on it costs each month.
Questions
How long does the assessment take?
About two minutes. There are five short questions, and the score appears a few seconds after you press the button. The more specific your list of repeating tasks, the more useful the result.
Is it free, and do I need to sign up?
It is free and there is no sign-up. Your result shows on the page straight away and you can copy it as plain text. The email box at the bottom of the page is optional and only for people who want to apply to the Gravity alpha.
How is the readiness score calculated?
Jev, a decision model from TypeSafe, rates how repeatable your tasks are, how much of the work happens in software and how much repeat work there is, each from 0 to 2. Those ratings are worth up to 35, 30 and 20 points. Each tool you use adds 2 points, up to 10, and your team's comfort with new software adds 0, 3 or 5. The total is out of 100.
What if my business scores low?
A low score usually means one thing is missing, most often written-down steps or work that still happens on paper or by phone. Every bar in the result comes with a next step. Fix the lowest one, try a single task with an agent that drafts and waits for your approval, and take the assessment again in a month.
Do you store my answers?
Your answers are sent to the model to score them and are not saved by Gravity, unless you press Share it or ask us to set them up as agents; tasks shared that way stay private and are never published. We count how many times the tool runs each day, nothing more. Leave out client names, account numbers and other private details; the assessment does not need them.
What should I automate first?
Pick a task that happens every week, follows the same steps, lives in software and is easy to check, then keep a person approving each output for the first two weeks. For many small businesses that is invoice reminders, a weekly report or sorting the inbox. The Where to start list in your result ranks the options for your business.