Free tool

Where should AI start in your business?

Tell us your industry, size, teams, data, tools, risk appetite and budget. The finder scores more than 50 practical AI use cases against your answers and returns a ranked shortlist with impact, effort, indicative cost, hours saved and a concrete first step for each.

Finder

Describe your business

Start from a preset or answer the questions below. 54 AI use cases are scored against your answers as you go.

01Industry
02Company size

Number of employees. Larger companies save more hours but pay more to integrate.

03Departments to focus on

Pick the teams where you would most like help. Leave empty to see everything.

04What matters most?
05How ready is your data?

Most AI projects are limited by data quality, not by the model.

06Tools you already use

Existing systems often include AI features or make integration easier.

07Technical capability
08How would it be built?

AI-native teams use coding agents for routine work, with senior engineers reviewing everything.

09Risk appetite
10Budget for a first year of AI work
Your AI opportunities
AI readiness
57 / 100
Developing: pair quick wins with one bigger bet
Strong matches
2
Quick wins
6 of 8
Hours saved / mo
~150
Top 3 build
$37k to $103k
Readiness breakdown
  • Data foundations22 / 40
  • Tools and systems12 / 20
  • Technical capability14 / 20
  • Budget5 / 10
  • Appetite for change4 / 10
Ranked shortlist
  1. 01
    Time capture and billing narratives
    Reconstructs billable time from calendars and documents and drafts invoice narratives.
    68
    Quick winLight build
    Impact4/5
    Effort1.5/5
    ~25 hours saved a month · build $4k to $16k
    First step: Estimate unbilled hours per fee earner by comparing a week of calendars with timesheets.
  2. 02
    Invoice and document data extraction
    Reads invoices, forms and delivery notes and enters the data into your systems.
    61
    Core projectProject
    Impact4/5
    Effort2.5/5
    ~40 hours saved a month · build $16k to $43k
    First step: Count monthly document volume by type and note which fields staff re-key today.
  3. 03
    Automated reconciliation
    Matches bank lines, payments and invoices and surfaces only the exceptions.
    59
    Core projectProject
    Impact4/5
    Effort2.5/5
    ~25 hours saved a month · build $16k to $43k
    First step: Measure month-end close time and the share of transactions matched by hand.
  4. 04
    Shared inbox automation
    Reads shared mailboxes, extracts requests and creates tasks or orders automatically.
    51
    Quick winLight build
    Impact3/5
    Effort1.5/5
    ~30 hours saved a month · build $4k to $16k
    First step: Pick one shared inbox and categorise a week of emails by request type.
  5. 05
    Internal knowledge assistant
    Lets staff ask questions of policies, procedures and manuals and get sourced answers.
    48
    Quick winLight build
    Impact3/5
    Effort1.5/5
    ~30 hours saved a month · build $4k to $16k
    First step: List the 50 questions staff ask most and check the answers exist in writing.
  6. 06
    Management report commentary
    Drafts monthly management accounts commentary and variance explanations.
    48
    Quick winLight build
    Impact3/5
    Effort1.5/5
    ~10 hours saved a month · build $4k to $16k
    First step: Give an AI assistant last quarter's reports and compare its commentary with yours.
  7. 07
    Accounts receivable follow-up
    Sends tailored payment reminders and predicts which invoices will be paid late.
    46
    Quick winLight build
    Impact3/5
    Effort1.5/5
    ~10 hours saved a month · build $4k to $16k
    First step: List overdue invoices by age and customer, and agree the reminder rules.
  8. 08
    Meeting notes and actions
    Records internal meetings, writes minutes and assigns actions in your task tools.
    46
    Quick winConfigure
    Impact2/5
    Effort1/5
    ~30 hours saved a month · build $1k to $6k
    First step: Enable the meeting assistant in your workspace for one team and review privacy settings.
Where to start
  • Start with time capture and billing narratives and shared inbox automation. They need little build work and can show results within weeks.
  • 7 of your 8 shortlisted ideas are held back by data readiness. Consolidating the data they share would lift several at once.
  • With a cautious risk appetite, keep a person approving anything AI sends to customers, and log AI outputs so you can audit them.
AI delivery and value
  • Building your top three with an AI-native team comes to roughly $37k to $103k, about $23k less than a traditional team under these assumptions.
  • Hours saved are valued at $50 per staff hour: roughly $92k a year across your top five if adoption is good. Real savings depend on redeploying that time.
  • AI output still needs clear ownership, human review on anything sensitive, and a check that your data is not used to train third-party models.

Scores combine impact, fit with your goals and departments, effort, data readiness, budget and risk. Hours saved assume good adoption; build costs are indicative first-release ranges and exclude licences and running costs.

Indicative only. We will send the inputs above with your message so a senior engineer can sanity-check them.

How the finder works

Value and feasibility, adjusted for your situation.

Each use case in the library is tagged with the departments and industries it suits, the outcomes it drives, and its typical impact, effort, data needs and risk. Your answers turn those tags into a score. A strong idea for a logistics operator can be a weak one for a ten-person accounting firm, and the ranking reflects that.

Value comes from impact and how well the use case matches the goals you selected. Feasibility comes from effort, the gap between the data it needs and the data you have, your technical capability, your budget, and whether tools you already use make it easier. Risk appetite then discounts ideas that would put more trust in AI than you are comfortable with.

Your AI readiness score is separate from the ranking. It summarises how prepared you are to run AI projects in general, so you can see whether the next investment should be a use case or the foundations under it.

Portfolio

Quick wins, core projects and strategic bets

A healthy set of AI use cases for a business mixes all three. The shortlist labels each idea so you can balance early results with longer-term value.

Quick winCore projectStrategic bet
Typical effortConfigure existing toolsFocused build and integrationMulti-phase programme
Data neededLittle beyond documentsClean data in one or two systemsHistorical, joined, well-governed data
Time to first resultDays to weeksSix to twelve weeksA quarter or more
Indicative first releaseUnder $10k$30k to $80k$80k and up
ExampleMeeting notes, reply draftingDocument extraction, support assistantDemand forecasting, AI product features
How to run itPilot with one teamScoped project with success measuresDiscovery, pilot, then scale

Costs are indicative first-release ranges for a mid-sized company before licences and running costs. AI-native delivery typically lowers the build portion.

Choosing well

How to pick your first AI project

The shortlist is a starting point. Before committing budget, test your top two or three ideas against these questions.

  • Can you measure today's baseline?

    If you do not know how long the task takes now or how often it goes wrong, you will not be able to prove the AI version is better.

  • Is the work frequent and repetitive?

    AI pays back fastest on tasks people do many times a week with a recognisable pattern, not on rare, high-judgement decisions.

  • Does the data exist in usable form?

    Check where the documents, tickets or records live, who owns them, and whether you are allowed to send them to an AI service.

  • Who owns the outcome?

    Every AI use case needs a named business owner who reviews quality, handles exceptions and decides when it is good enough to expand.

  • What happens when the AI is wrong?

    Design the human review step before launch. Internal drafts can tolerate errors; anything sent to customers or regulators needs checks.

  • Buy, configure or build?

    Many quick wins are features in software you already pay for. Build only where your process or data gives you an edge, and ask how any build team uses AI in its own delivery.

The best first AI project is usually the boring one.

Businesses often start with the most impressive idea on the list and stall on data, integration or trust. The teams that make progress pick a narrow, frequent, well-understood task, prove the saving within a couple of months, and use that credibility to fund the harder projects.

Building has also become cheaper. AI-native engineering teams, working with coding agents and senior review, can typically deliver integrations and internal tools with fewer engineer-hours than before, so ideas that once needed a large budget may now fit a focused project. What has not changed is the need for clean data, clear ownership and honest measurement.

FAQ

Questions we often hear

What are the best AI use cases for small businesses?

For most small businesses the best starting points are meeting notes, email and reply drafting, document data extraction, marketing content and an internal assistant over policies and procedures. They need little data preparation, often come built into tools you already use, and save time within weeks. Custom builds make more sense once these are working.

How do I identify AI opportunities in my company?

List the tasks your teams repeat every week, estimate the hours they take, and note where information is read, summarised, classified or re-keyed. Those patterns are where current AI performs well. Then rank the candidates by value, data readiness and risk, which is exactly what this finder does from your answers.

How much does a first AI project cost?

Configuring AI features in existing tools can cost a few thousand dollars, a focused custom build commonly ranges from tens of thousands, and forecasting or AI product features can run into six figures. Company size, integrations and compliance requirements move the number, and AI-native delivery teams typically bring the build portion down.

What does AI readiness mean?

AI readiness describes how prepared an organisation is to use AI well: the quality and accessibility of its data, the systems it runs, its technical capability, budget and appetite for changing how work is done. Low readiness does not rule AI out. It means starting with low-data quick wins while fixing foundations in parallel.

Should we use AI features in existing software or build our own?

Start with what your existing software offers, because it is cheaper and faster to prove value. Build custom AI where your data, workflow or customer experience is genuinely different, or where off-the-shelf features cannot meet your privacy requirements. Many companies end up with a mix of both.

Prioritise with an expert

Turn your shortlist into an AI roadmap.

Send us your shortlist and a little context. We will tell you honestly which ideas are worth doing first, which to leave for later, and what a sensible first project would cost with an AI-native team.

Working with companies globally · Response within 24 hours