As of 13 August 2026, AI can choose a profitable freelance niche.

What to actually do

1. Use a tool built for this

   The route this page recommends

2. Do it yourself

   Second choice

   A chat interface, chat-fluent skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open a document and record your skills, strongest work examples, industries you understand, available hours, income target, location, preferred clients and work you will not accept.
   2. Search UK-focused job boards, LinkedIn, freelance marketplaces and relevant business communities for recent requests matching those skills, then paste links or short excerpts and the stated requirements into the prompt.
   3. Collect comparable UK freelancer or agency offers and prices for each possible service, keeping the source link and separating advertised prices from your own assumptions.
   4. Paste your background and evidence into the prompt, run the analysis, and ask the chatbot to rank five niches using only the supplied evidence and clearly labelled inferences.
   5. Take the top two recommended niches and write down the proposed buyer, deliverable, scope and price test; compare each claim against the source links and remove anything the evidence does not support.
   6. Contact five potential buyers for each shortlisted niche using the generated questions and outreach message, then record replies, objections, requests for examples and any paid or serious trial interest.
   7. Apply the prompt's decision rule to those responses and choose one niche to test further, or gather more evidence instead of treating the chatbot's ranking as proof of profitability.

   Prompt

   Help me choose and test a profitable freelance niche in the UK. Do not promise profitability and do not invent market figures, client demand, prices or competitor facts. Separate evidence from inference, and say when information is missing or needs checking.
   
   My situation:
   - Skills I can sell: [list your skills]
   - Relevant work, results or portfolio examples: [describe them]
   - Industries or subjects I know: [list them]
   - Work I do not want to do: [list them]
   - UK location and whether I can work remotely: [details]
   - Hours available each week: [number]
   - Monthly income target: [amount in pounds]
   - Earliest date I want to start: [date]
   - Existing contacts, audiences or channels: [details]
   - Constraints, including equipment, qualifications or caring responsibilities: [details]
   
   Evidence I have gathered:
   - UK job, project or service listings: [paste links or excerpts]
   - Prices shown by comparable freelancers or agencies: [paste links or excerpts]
   - Search, community or enquiry evidence: [paste links or excerpts]
   - Any conversations with potential clients: [paste anonymised notes]
   
   Produce:
   1. Five plausible niches, each with the buyer, problem solved, deliverable, likely route to first clients, relevant proof from my background, and the evidence supporting demand.
   2. A score from 1 to 5 for fit, reachable demand, ability to show proof, competition risk and speed to first sale. Explain every score and mark unsupported scores as unknown rather than guessing.
   3. A ranked recommendation, with the strongest objections and what would change the ranking.
   4. A narrow first offer for the top two niches, including a deliverable, scope boundary, target UK buyer and a price-testing approach. Do not state a market price unless it appears in my evidence.
   5. A seven-day validation plan using actions I can complete without buying a course or assuming I have an audience. Include a pass or fail condition for each test.
   6. The exact questions I should ask five potential buyers, and a short outreach message that makes no unsupported claims.
   7. A final decision rule: tell me whether to test niche one, test niche two, test both, or gather more evidence. Do not call either niche profitable until real buyers have shown interest or paid.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

3. Hand it to a person

   The distant third

   A person who owns the outcome does this end to end, worth it when the failure is dear.