"Do I need a technical background?" and "How long until I can do this on my own?" each came up in about six of my client calls recently. The people asking were managers, founders, HR and finance professionals, teachers and agency owners. This is the roadmap I wish someone had handed them on day one.
Before you start: 3 rules
- Use your own work, not toy examples. Your emails, your reports, your Excel sheets.
- 20 to 30 minutes a day beats a weekend binge.
- One paid tool at most. Not sure which? Read which AI tool to pay for.
Month 1 (days 1 to 30): Use AI every day
Goal: AI becomes a habit, not a novelty.
Week 1: Learn the words
Read five pages in my AI for Non-IT Library: model, large language model, prompt, tokens and AI hallucination.
Week 2: Write better prompts
Give context, the role, the format you want and an example. A question I hear a lot from senior people: "How do I get executive-level output, not generic text?" The answer is almost always more context: who it is for, what decision they need to make, and one example of a good past document.
Week 3: Daily tasks
Pick three tasks you do every week: summarising long emails, drafting a report, comparing two documents, cleaning an Excel sheet. Do each one with AI.
Week 4: Check the work
AI makes confident mistakes. Practise checking numbers, names and dates every time.
The "aha" moment usually comes when someone tries AI on the job they dread most. One of my learners connected Codex to a design tool he uses and finished an animation that normally takes him hours in one sitting. His message to me afterwards: "This is the first time in my life I've done the homework." That is the feeling month one is for.
Month 2 (days 31 to 60): Connect AI to your work
Goal: stop copy-pasting.
- Projects and files: keep one project per area of work, with your documents and instructions saved.
- Skills: save instructions you repeat (your writing style, your report format) as a reusable skill so AI follows them every time. Skills and scheduled tasks came up in about ten of my recent client calls. People love them once they see them.
- Scheduled tasks: for example, a Monday morning summary of last week's emails. As of October 2026, ChatGPT offers scheduled tasks even on Free and Go (up to three active tasks, at most daily), with more frequent runs on paid plans. Claude offers scheduled tasks on its paid plans (Pro and above).
- Connectors: connect email, calendar or Drive. Read is it safe to let an AI agent use your computer first, and start read-only.
Month 3 (days 61 to 90): Build one thing
Goal: one small tool that saves you time every week.
Kinds of things my learners have built with no coding background:
- A simple app for a real daily task, built with Claude Code and put online
- A scheduled update that replaced a manual Excel tracker, sent over WhatsApp and email
- An automation that sends a summary every morning
- A checker that verifies a supplier's test certificate against the required standard
Steps:
- Write one paragraph: the problem, who uses it, what "done" looks like.
- Ask a coding agent (Claude Code or Codex) to build the first version.
- Test it with real data. Fix one thing at a time.
- Put it online with GitHub plus Netlify or Vercel, if others need to use it.
- Keep your API keys out of your code and out of GitHub.
A month-3 build I love: a product manager with zero technical background installed Claude Code on our first call and, within about five days, had a working attendance app for a real daily problem, hosted on Netlify through GitHub. She told me the word "API" used to scare her. Now she explains it to others. If your build feels too small, it is probably the right size.
Will this get me a job?
Honestly: no course or roadmap can guarantee a job or a certification, and I don't promise either. What 90 days does give you is proof: real things you built from real work, which you can show in interviews. For the bigger picture, read my take on AI and jobs in India.