YOU ARE HEREAIm→03 Apply→Earn responsibly
← AImBuild with AI

BUILD YOUR AI FUTURE

Don't sell hype. Build useful value.

Turn skills you already have into useful, evidence-based AI-assisted work, services or business.

It is not too late. You do not have to become a programmer. Combine useful AI knowledge with work, people or an industry you already understand—and build evidence one small result at a time.

BEFORE ANY INCOME CLAIM

The AIm opportunity test.

If an idea cannot pass these four questions, AIm does not present it as a credible earning path.

  1. 01Real problemCan you name the person, repeated pain and current alternative?
  2. 02Human capabilityCan you judge the work without blindly trusting AI?
  3. 03Small proofCan you demonstrate a safe result before making a large claim?
  4. 04Honest economicsWill someone pay after costs, review time, risk and support are counted?
01 · GROW IN YOUR ROLE

Use AI to save time, communicate clearly, research carefully and strengthen the work you already know.

Member pathway · practical evidence · responsible use

Explore this pathway
What to learn
AI literacy, verification, privacy, prompt iteration and measurement.
First project
Improve one low-risk weekly task with sample data and compare the before and after.
Proof to build
A one-page case study showing time saved, quality checks, limits and human approval.
Possible buyer or employer
Your current employer, team or professional network.
Start here
List five repetitive tasks. Choose one where mistakes are easy to detect and reverse.
Watch out for
Uploading confidential information or claiming productivity gains you did not measure.
02 · START SMALL

Find a real customer problem, shape one simple offer and test demand before spending heavily.

Member pathway · practical evidence · responsible use

Explore this pathway
What to learn
Customer interviews, problem definition, simple prototypes, pricing and consent.
First project
Interview five people in one niche and manually deliver one useful outcome.
Proof to build
Interview notes, a sample deliverable and evidence that someone values the result.
Possible buyer or employer
A specific local or online niche you can reach and understand.
Start here
Ask what wastes time, causes errors or gets postponed—without mentioning AI first.
Watch out for
Building an app, buying tools or registering a company before testing demand.

WHERE REAL DEMAND CAN EXIST

Earn from a useful outcome—not the word “AI”.

These are credible directions to investigate, not vacancies or income guarantees. Demand varies by country, industry and your evidence.

01AI adoption & workflow support
Why it can be valuable
Organisations need people who can map work, test tools, train users and measure results.
Skills to develop
Process mapping, facilitation, privacy, evaluation and change support.
Proof before promotion
A supervised workflow pilot with a before-and-after measure.
02AI-assisted specialist services
Why it can be valuable
Research, analysis, marketing, bookkeeping, design and documentation still require domain judgement.
Skills to develop
A real professional skill plus AI literacy, checking and client communication.
Proof before promotion
A reviewed deliverable showing where AI helped and where a person decided.

Evidence base: ILO task-exposure research, the World Economic Forum’s 2025 skills outlook, OECD small-business research and completed-work marketplace data. AIm treats these as signals—not promises.

NEW · CAREER COMPASS

Which roles are growing—and which tasks will change?

Explore ten AI career paths, the experience employers need, and a visual 40-role outlook for AI exposure and resilience.

Open the career outlook →

CONSULTANT PATHWAY

From useful experience to credible evidence.

You become useful by understanding people, processes and consequences—not by adopting a title.

01Choose a field where you can judge good work
RESULT

One narrow industry, buyer and workflow—not a promise to transform every business.

DO THIS
  1. List work you understand from employment, study or lived experience.
  2. Speak to five people who perform or manage it.
  3. Choose a repeated problem with visible cost, delay, error or frustration.
PROOF TO DELIVER

A one-page niche brief naming the buyer, user, current process and reason to act.

02Sell discovery before technology
RESULT

A paid or tightly scoped workflow review with no promise that AI is the answer.

DO THIS
  1. Map inputs, decisions, handoffs, exceptions and accountability.
  2. Record baseline time, rework, volume and quality.
  3. Identify data rights, sensitive information and non-negotiable human decisions.
PROOF TO DELIVER

A current-state map, baseline measure and ranked opportunity list.

03Build the smallest supervised demonstration
RESULT

One reversible improvement tested with synthetic, public or permissioned data.

DO THIS
  1. Compare AI with rules, forms, search and existing features.
  2. Create normal, difficult and malicious test cases.
  3. Agree success, failure and stop conditions before the demonstration.
PROOF TO DELIVER

A working demonstration, test set, error log, cost estimate and human review point.

04Run a measured pilot and document the truth
RESULT

Evidence showing whether the workflow became faster, better, safer or not worth continuing.

DO THIS
  1. Choose a small user group, owner and fixed pilot period.
  2. Track corrections, overrides, incidents, time and total cost.
  3. Interview users about added work as well as saved time.
PROOF TO DELIVER

A before-and-after case study including failures, limitations and the customer's approval.

05Offer an outcome with clear limits
RESULT

A fixed-scope service the client can understand, verify and safely end.

DO THIS
  1. State deliverables, exclusions, data rules, review ownership and support.
  2. Price discovery and the pilot before promising a transformation.
  3. Define change requests, acceptance, cancellation and incident response.
PROOF TO DELIVER

A proposal, statement of work, risk register, acceptance criteria and honest price.

SIMPLE OFFER TEMPLATE

“I help [specific team] improve [one workflow] by [measurable outcome]. We begin with discovery and a supervised pilot. You retain approval of [important decisions]. The pilot includes [deliverables], excludes [boundaries] and succeeds when [acceptance measure].”

Start with a fixed discovery fee and pilot scope. Price broader implementation only after evidence exists.

AIm STARTUP LAB

Build an AI business people genuinely need.

You do not need funding, a technical co-founder or a finished app to begin. You need access to a real problem, honest customer conversations and a small test. Use AI only where it improves the outcome enough to justify the cost and risk.

Eight decisions · evidence before expansion · no income promises
01PROBLEMFind pain before an AI idea
THE OUTCOME

A specific, frequent and costly problem experienced by people you can reach.

DO THIS
  1. Choose a field, community or workflow you understand.
  2. Interview at least five potential users about recent behaviour.
  3. Record current alternatives, cost, delay, risk and who decides to buy.
EVIDENCE TO EARN

Several people describe the same problem without being led—and at least one agrees to test a solution.

FOUNDER QUESTION

Would this still be worth solving if AI disappeared tomorrow?

02CUSTOMERChoose one first customer
THE OUTCOME

One narrow customer group, buyer, user and situation—not “everyone”.

DO THIS
  1. Separate the person using the solution from the person paying.
  2. Identify where these people already gather and how they buy.
  3. Write one sentence: We help [person] achieve [outcome] when [situation].
EVIDENCE TO EARN

You can name ten reachable prospects and explain why they would trust you.

FOUNDER QUESTION

Who feels this problem strongly enough to act now?

03VALUEDefine the result and business model
THE OUTCOME

A measurable customer outcome and a believable way the business earns revenue.

DO THIS
  1. Describe the result in time, cost, quality, access or reduced risk.
  2. Compare direct and indirect alternatives, including doing nothing.
  3. Test a simple price and calculate delivery cost per customer.
EVIDENCE TO EARN

A customer commits time, data, a pilot fee, pre-order or another meaningful action.

FOUNDER QUESTION

What valuable result is the customer buying—not what technology?

YOU NEED

A problem, customer access and learning speed.

Useful domain knowledge, time to talk to users, a way to deliver a first result, basic financial records and the willingness to discover that your first idea is wrong.

YOU MAY NOT NEED YET

Investment, employees or custom AI models.

Begin manually or with established tools. Add software, automation and funding only when evidence reveals a real bottleneck.

BEFORE LAUNCH

Know your costs, data, duties and stop conditions.

Model usage, hosting, support and human review must fit the price. Check local tax, privacy, consumer and sector rules with a qualified adviser.

10 PRACTICAL FIELD GUIDES · WEEKLY SCOUT

Choose one field. Open only what is useful.

These are research directions—not recommendations to enter a regulated field without experience. Begin with administration and decision support; keep qualified people responsible for safety, money, rights and professional judgement.

01Professional servicesLegal, consulting, recruitment and advisory firms
Process to examine
Client intake, meeting preparation, document search, first drafts and follow-up.
How AI may improve it
Structure information once, retrieve approved knowledge and leave judgement with the professional.
IT security & trust
Confidentiality, privilege, client consent, source accuracy and strict separation between clients.
Safe first pilot
Summarise ten invented intake records into a standard review form.
Measure
Preparation time, missing facts, corrections and professional approval.
Skills to learn
Information architecture, source checking, confidentiality, prompt design and professional quality review.
Tool categories
Secure document search, meeting transcription and approved office copilots.
Proper example
A consultant turns fictional discovery notes into a cited briefing, then corrects it before a client meeting.
POSSIBLE PAID OFFER

A fixed-scope knowledge and client-intake improvement: map the current process, build one cited assistant and train staff to review it.

30-DAY EVIDENCE PLAN
  1. 1Interview two professionals and map one repeated delay.
  2. 2Create a redacted knowledge set and ten test questions.
  3. 3Run a supervised pilot and record every correction.
  4. 4Deliver the process map, test result, risks and recommendation.
DO NOT AUTOMATE

Do not provide legal, recruitment or professional advice you are not qualified to judge, and never mix one client's information with another's.

Evidence: NIST AI Risk Management Framework ↗
02Accounting & finance operationsBookkeepers, finance teams and small accounting practices
Process to examine
Invoice capture, transaction explanation, reconciliation preparation and exception routing.
How AI may improve it
Classify routine documents and surface anomalies while a qualified person approves records and advice.
IT security & trust
Financial data, fraud, access control, audit trails and no autonomous payments.
Safe first pilot
Use synthetic invoices to prepare an exception list for human review.
Measure
Exceptions found, false alerts, review time and audit completeness.
Skills to learn
Bookkeeping fundamentals, reconciliations, fraud indicators, access control and audit evidence.
Tool categories
Accounting-platform assistants, document extraction and spreadsheet copilots with audit logs.
Proper example
A bookkeeper uses synthetic invoices to draft an exception queue; a qualified person posts every final entry.
POSSIBLE PAID OFFER

A human-approved exception workflow for invoice intake, reconciliation preparation or management-report commentary.

30-DAY EVIDENCE PLAN
  1. 1Choose one repetitive finance queue—not payments or final posting.
  2. 2Build synthetic examples including duplicates and unusual items.
  3. 3Compare AI flags with an experienced finance reviewer.
  4. 4Report accuracy, false alerts, time saved and audit evidence.
DO NOT AUTOMATE

Never let AI move money, approve suppliers, post final entries or give regulated financial or tax advice without authorised review.

Evidence: NIST AI RMF Playbook ↗
03Healthcare & care administrationClinics, allied health practices and care organisations
Process to examine
Scheduling, approved information, referral administration and note organisation.
How AI may improve it
Reduce administrative friction without replacing diagnosis, safeguarding or clinical decisions.
IT security & trust
Highly sensitive health data, consent, bias, clinical safety and sector regulation.
Safe first pilot
Test appointment-question routing with fictional patients and a staff escalation path.
Measure
Correct routing, waiting time, unsafe answers and staff overrides.
Skills to learn
Healthcare administration, privacy, consent, safeguarding, escalation and clinical-AI limits.
Tool categories
Approved scheduling, knowledge-search and clinical-administration tools—not general consumer chatbots.
Proper example
A clinic tests fictional appointment questions against approved public guidance and routes uncertainty to staff.
POSSIBLE PAID OFFER

A low-risk administrative pilot for appointment routing, approved information retrieval or referral-document preparation.

30-DAY EVIDENCE PLAN
  1. 1Select an administrative task with a clear escalation route.
  2. 2Use fictional patients and approved public guidance.
  3. 3Test unsafe, incomplete and urgent scenarios with staff.
  4. 4Document failures, overrides, consent and a safe stop condition.
DO NOT AUTOMATE

Do not automate diagnosis, treatment, safeguarding, urgency decisions or patient communication outside approved clinical governance.

Evidence: WHO ethics and governance of AI for health ↗

VERIFIED FOUNDER RESOURCES

Start with evidence you can trace.

These sources provide guidance—not endorsement, funding or a guarantee of success. The weekly Startup Scout checks them and proposes updates for administrator approval.

Y Combinator Startup SchoolFree founder fundamentals: users, MVP, customers, pricing and launch+

AIm summary: Use this as evidence and guidance—not as a promise of funding, customers or income. Apply it to one small experiment and record what happened.

Optional: open the official resource ↗
SBA business guideMarket research, competitive analysis, planning and startup costs+

AIm summary: Use this as evidence and guidance—not as a promise of funding, customers or income. Apply it to one small experiment and record what happened.

Optional: open the official resource ↗

WEEKLY CAREER & BUSINESS 10

Watch for understanding—not promises.

Ten verified videos that play inside AIm. The weekly Scout proposes replacements; an administrator approves them.

01 · Google Cloud Tech

Introduction to Generative AI

Foundation for explaining AI to a client

02 · IBM Technology

What are Generative AI models?

Understand capabilities before recommending tools

03 · CGP Grey

AI Doesn't Know Anything. It Just Passes Tests.

Learn why verification belongs in every service

TRUSTED STARTING SOURCES

Learn from organisations with something to verify.

If a link moves or a course is unavailable, the AIm mini-lessons remain available.

Grow with GoogleAI training for small businesses+

Why it is here: This established provider offers a useful starting point. AIm gives you the purpose first so you can decide whether the full source is worth leaving the page for.

Optional: open the official source ↗
Microsoft LearnAI fluency and business-user pathways+

Why it is here: This established provider offers a useful starting point. AIm gives you the purpose first so you can decide whether the full source is worth leaving the page for.

Optional: open the official source ↗
AWS Skill BuilderCareer and role-based AI learning+

Why it is here: This established provider offers a useful starting point. AIm gives you the purpose first so you can decide whether the full source is worth leaving the page for.

Optional: open the official source ↗

AIm SCAM SHIELD

Never pay to get paid.

Honest opportunities explain the work, the organisation and how value is created. Scammers create urgency, hide the employer and ask you to risk money or identity information.

Stop when you see

  • Guaranteed income, effortless profit or “secret systems”.
  • Unexpected WhatsApp or text offers for liking, rating, “optimising” or clicking.
  • Crypto deposits, recharge payments or fees to unlock work or withdraw earnings.
  • Fake cheques, reshipping, celebrity deepfakes or unverifiable testimonials.
  • Requests for passwords, banking details, identity documents or unknown software too early.

Check before you continue

  1. Find the vacancy on the employer’s official website.
  2. Verify the recruiter’s name, domain and contact details.
  3. Search the company name with “scam” and “complaint”.
  4. Ask a trusted person to review the offer and contract.
  5. Keep screenshots; report fraud to the platform, bank and your national reporting authority.
FTC job-scam guide ↗FBI warning about fake task jobs ↗