Small business
Choosing a first AI project | AI Course London
A useful first AI project is a recurring task with clear inputs, an output you can check and a person responsible for the result. Start with one process, such as drafting a weekly customer update from approved notes. Record preparation, checking and correction time as well as drafting time. Compare that total with your current process before deciding whether to keep it.
AI Course London · First published · Updated
Which task is a good first candidate?
Choose work that happens often enough to learn from, has an identifiable source and can be reviewed before it affects a customer. A draft update, an internal checklist or an action list is easier to assess than an open-ended promise to automate the business.
Write down the current process first. Include the person who gathers the input, the person who checks the result and where the final version is used. If nobody owns the source or review, resolve that problem before introducing another tool.
- The input is complete enough for the task.
- A responsible person can recognise a good result.
- An incorrect draft can be caught before it is sent or acted on.
- The task is permitted under the organisation’s data and tool policies.
How can I practise without using customer information?
Create a fictional source that contains the same kinds of decisions as the real task. Preserve missing information and exceptions so the exercise tests your checking process. Do not present a practice result as a real customer outcome.
Fictional weekly-update exercise
Source material
Workshop supplies arrived on Monday. Two sample colours need a final decision. The display installation date is not agreed. Morgan will check the colours by Wednesday.
Prompt to try
Draft an internal weekly update using only these notes. Use three headings: Completed, Next action and Decisions needed. Keep the unconfirmed installation date explicit. Do not claim a launch date, customer demand or time saved.
Human reference for checking
Completed: workshop supplies arrived on Monday. Next action: Morgan checks the two sample colours by Wednesday. Decisions needed: final colour choice and the display installation date. No launch date is confirmed.
This is a fictional practice exercise. The reference is written for checking your result; it is not an attendee outcome or a claimed model response.
How do I measure whether the task improves?
Compare the whole job with your existing method. Measure source preparation, drafting, review, correction and final acceptance. A shorter generation time alone does not show that the process is better.
| Measure | Current method | AI-assisted attempt |
|---|---|---|
| Input preparation | Record minutes | Record minutes |
| Drafting | Record minutes | Record minutes |
| Checking and correction | Record minutes and errors | Record minutes and errors |
| Final acceptance | Accepted or rejected, with reason | Accepted or rejected, with reason |
| Total effort | Add the actual observations | Add the actual observations |
What if the draft looks good but contains errors?
Keep it out of the live workflow until the review has passed. Identify whether the problem came from an incomplete source, an unclear instruction or the tool adding unsupported content. Correct the cause, then test the same case again.
Retain rejected cases in your internal evaluation. They show where supervision is needed. If checking consistently takes more effort than the original task, choose another candidate instead of hiding that cost in a headline about time saved.
When should I connect the task to other systems?
Connect a workflow only after its input, output, review owner and failure behaviour are understood. A draft that a person checks is a different commitment from a system that sends messages, changes records or spends money automatically.
Start with one clearly bounded handoff and a way to stop or correct it. Confirm permissions before using customer or employee records. Our public workshop is a place to practise on a permitted work problem; the current agenda describes the day and support offered.
Before you try it
Will AI save my business ten hours every week?
This guide does not establish a time-saving figure. Measure the total effort on your own tasks, including review and correction, before making a business case.
Do I need to automate the whole process?
No. A useful first result may be a draft that a person reviews. Keep consequential actions under an appropriate review process while you learn.
Can I bring a real business problem to the workshop?
Yes. Bring your laptop, complete the preparation and choose examples you are permitted to share. Use fictional material if your organisation has not approved a real-data exercise.
Check the source
Product features, plans and settings change. Check the provider and your organisation before using work material.
Provider information
Check the actual plan before choosing
Claude is the starting point in our public workshop. Check the current plan for your task.
Compare the product and plan with your required workflow and workplace permissions.
Check the current Notion AI features and plan terms if your team already uses Notion.
Claude and ChatGPT links go directly to the providers. The Notion link is an affiliate link; we may receive a commission if you make an eligible purchase.
Practise on your own work problem
Join Andrew for a one-day London workshop for non-technical professionals. Work in a group of at most eight, with guided practice from 9am to 4pm. The £599 fee includes lunch, materials and 30 days of support.
See workshop dates and details