How to get a job in AI in 2026: an honest checklist from people who hire

Every week someone asks some version of the same question: "I've done the courses. Why can't I get an interview?"

We put that question to the hiring managers, founders and recruiters in our Hiring circle. Their answers were remarkably consistent — and a little uncomfortable. So here's the checklist, without the sugar-coating.

1. Stop collecting certificates

Courses are useful for learning. They are almost useless as a hiring signal, because everyone applying has the same ones. One founder told us she skips straight past the certifications section of a CV. "It tells me you can finish a course. I need to know you can finish a project."

Keep learning. Just stop presenting learning as proof.

2. Ship one real thing — not five tutorials

The strongest signal for any AI role, technical or not, is a project that:

• Solves a specific problem for a specific person. Not "a chatbot." A tool that helps a local clinic summarise intake notes, or helps a friend's sales team research accounts.
• Was used by someone other than you. Even three real users teaches you more than a polished demo nobody touched.
• Shows where it failed. A short write-up of what broke and what you changed is worth more than the project itself. It proves judgement.

One real project with honest notes beats a GitHub full of cloned tutorials every time.

3. Pick a lane — AI plus something

"I want to work in AI" is too broad to hire for. The people breaking in fastest are combining AI with something they already know:

• AI plus your old industry (healthcare, logistics, law, finance)
• AI plus a function (sales, support, operations, recruiting, marketing)
• AI plus a technical specialty (data engineering, evaluation, security, infrastructure)

Your previous career isn't something to hide. It's your advantage. A former nurse who can build a reliable clinical workflow tool is far more hireable than a generalist with the same technical skills.

4. Learn the unglamorous parts

Hiring managers kept coming back to the same gap: candidates can build a demo, but they don't understand what it takes to make AI work in production. If you want to stand out, get comfortable with:

• Evaluation. How do you know your system is actually good? How do you measure it over time?
• Cost and latency. What does it cost per request? What happens at a thousand users?
• Failure modes. Hallucinations, prompt injection, bad data, edge cases. How do you catch them?
• The humans involved. Who uses this, who's affected by it, and who needs to trust it?

You don't need to be an expert. You need to show you know these questions exist.

5. Write in public (a little)

You don't need to become an influencer. But a handful of clear, honest posts about what you're building and learning does two things: it proves you can think and communicate, and it gives people a reason to remember your name. Short and specific beats long and generic. "What I learned trying to get an agent to reliably fill out insurance forms" is more interesting than "10 AI trends for 2026."

6. Get into smaller rooms

This is the part most job seekers skip. The best AI roles often aren't posted publicly for long — or at all. They get filled through referrals and warm introductions. That means the real job search happens in places where people know your work:

• Small communities where hiring managers are actually present
• Build squads and hackathons where people see you work
• Direct relationships with a few practitioners who can vouch for you

Applying to 300 listings is exhausting and mostly invisible. Being known by 15 of the right people is usually more effective.

7. Prepare for AI-era interviews

Interviews are changing too. Expect to be asked:

• To walk through a project and defend your decisions
• How you'd evaluate whether an AI output is good enough
• What you'd do when a system gives a confident, wrong answer
• How you use AI tools in your own work — and when you choose not to

Some companies now let you use AI during take-home tasks, then ask you to explain every choice. Be ready to show your judgement, not just your output.

A word on the fear

If you're changing careers into AI partly because you're worried about your current job, you're not alone — and that's a legitimate reason. But go in with a plan, not a panic. Pick the lane. Ship the project. Find the room. It usually takes months, not weeks, and that's normal.

The people who make it aren't always the most technical. They're the ones who kept building, kept talking to real practitioners, and didn't try to do it alone.

---

Our AI Job Seekers circle runs a structured eight-week path — portfolio reviews, mock interviews with hiring managers and warm introductions for people who ship. If you're serious about moving into AI, request an invitation.