Switching careers in the AI era: an honest 6-month plan.
I've helped a lot of people switch into AI-adjacent roles. The ones who succeeded followed roughly the same pattern. Here's the honest version that fits in six months — without quitting your day job.
Every person I know who successfully switched into an AI-related role from a different career followed roughly the same six-month pattern. None of them did a bootcamp. Most of them didn't get a new degree. They did something more focused, and harder, and more useful: real work, in public, in a direction they cared about.
Here's the honest plan, condensed.
Month 1: Pick the smallest, most specific destination you can
Most career-switch attempts fail at this step. People say "I want to work in AI". That's not specific enough to plan around.
A specific destination looks like: "I want to be a product manager on an AI feature team at a Series B or later company, in a vertical I care about, within 12 months." That's a planable target. You can identify the companies. You can identify the skills. You can identify who currently has that job and what they did to get it.
If you can't get this specific, you'll burn six months learning vaguely useful things. The specificity is the half of the work nobody talks about.
| Vague target (doesn't work) | Specific target (does work) |
|---|---|
| "I want to work in AI" | "AI product manager at a Series B-D fintech, within 12 months" |
| "I want to be a data person" | "Data analyst on a marketing team at a mid-size SaaS, within 9 months" |
| "I want to learn to code" | "Backend engineer at an early-stage AI startup, within 14 months" |
| "I want to do something more meaningful" | "Health-tech designer at a company building patient-facing AI tools" |
| "I want to use AI in my job" | "Marketing ops lead who owns the AI-tool stack at a 50-200 person company" |
Month 2: Talk to ten people doing the job
Reach out to ten people on LinkedIn or Twitter who currently have the job you want. Ask each one for 20 minutes. Most will say no. Two or three will say yes.
In each conversation, ask three questions:
- What does your typical day look like? (You need to know if you actually want this job.)
- How did you get this role from where you were before? (Their actual path, not the LinkedIn version.)
- What's the one skill you wish you'd built earlier? (This is the bonus answer that compresses your learning plan.)
Pattern-match across the answers. The plan you'll follow is hiding in the overlap between three or four of their answers.
Month 3: Build the smallest portfolio piece that's relevant
Now you know the destination and the rough path. Build one specific thing that someone in your target role would actually do at work.
Going for PM? Write a teardown of an AI feature you wish existed, with a real spec.
Going for AI engineer? Build a small RAG (retrieval-augmented generation) app that does one specific task, deployed somewhere people can use it.
Going for AI ops or data analyst on an AI team? Pull a real dataset, build a small dashboard, write up what you found.
The piece doesn't have to be impressive. It has to be specific to the role and shippable in a month. The point isn't the piece. The point is having a concrete artifact to point to when someone asks "what have you done?"
Month 4: Write about it publicly
Take the piece you built and write 1,500 words about it. What problem it solved. What was hard. What you learned. Publish it somewhere — a blog, LinkedIn, Substack, even just a long Twitter thread.
Most people skip this step because it feels self-promotional. It is. That's the point. Hiring managers in 2026 google candidates. If they find writing that suggests you can think clearly about the work, you're suddenly a different kind of candidate.
Month 5: Apply, while doing more work
Start applying. But here's the trick: don't just apply through job boards. Apply to specific people. Find the hiring manager or someone on the team. Send them your portfolio piece. Reference the conversations from month 2.
Job boards are a 1-in-100 funnel. Targeted outreach with a real artifact is a 1-in-10 funnel. The applications happen in parallel with continued work building your second portfolio piece. Don't stop creating just because you started applying.
Month 6: Be patient with the close
The last month is the hardest. You've done the work. You're getting some interviews. You haven't gotten an offer yet. Most people quit at this exact point, conclude "this doesn't work for me", and go back to their old job.
It does work. The conversion from "started getting interviews" to "got an offer" is usually 6 to 12 weeks. The first offer is almost always at month 7 or 8 for people who started in month 1. The plan compresses the timeline. It doesn't eliminate the wait.
What you didn't do
Notice what's not in this plan: no bootcamp, no second degree, no certification grind. Those things have a place. They're not the bottleneck for most career switchers in 2026.
The bottleneck is specificity and proof. The specificity comes from talking to people doing the job. The proof comes from actually doing the work, in public, in a way the hiring manager can verify in five minutes.
What goes wrong
Three failure modes I see most:
The "I'll start applying when I'm ready" trap. You're never ready. Start applying once you have one portfolio piece. The market teaches you what's missing faster than you can guess.
The "studying for the sake of studying" trap. Endless courses, no shipping. If you've been in a course for more than two months and haven't built anything, you're learning the wrong thing.
The "I'm too late" trap. Six months feels long when you start. It's not. Most people will take three years to do what this plan does in six. The discount on time is real, but only if you actually start.
Start tomorrow
Not next Monday. Not after the holidays. Tomorrow. The plan doesn't work if month one starts in two months.
Pick the destination. Send the first three messages to people doing the job. The rest of the plan unfolds from there.