In January, Jason bought business AI licences for all 60 staff at his interior design firm. By March, three people were using them every day. Most of the others had logged in once.

Jason's experience is common. Buying tools doesn't create adoption. People adopt AI when they know how it fits their own work, and when they have time and permission to learn. IMDA's latest Singapore Digital Economy Report found that more than two-thirds of workers feel they need AI training, but only just over a third had attended any in the past year. The most common reason they gave was that their employer hadn't sent them.

Here's a practical 90-day plan to close that gap. It follows four stages: Learn, Design, Build and Launch. Your timing may differ, but the order matters.

I. Days 1–30: Learn and spot opportunities

Find out where people are. A five-minute survey is enough: which AI tools staff use, how often, and how confident they feel.

Teach the basics using real work. Cover how to give AI clear instructions, how to work with documents, and how to check outputs. Use each team's own tasks, not generic demos.

Set ground rules early. Agree which tools are approved, what information must never go into them, and that every output gets checked before use.

Collect use cases. Ask each team to list three tasks that take too long. Jason's designers listed summarising client briefs, drafting quotations and writing up site-visit notes.

II. Days 31–60: Design and build

Map the workflow. For the most promising two or three use cases, write down each step as it happens today.

Prioritise. Choose problems that are frequent and low-risk, where the output is easy to check.

Design the AI-assisted version. Decide where AI drafts or sorts, and where a person reviews.

Build simple prototypes. These are rarely software. They might be a shared prompt template, an AI assistant set up with your SOPs, or a simple no-code workflow. Jason's team built a template that turns a client's WhatsApp messages and photos into a structured design brief.

Test with real examples. Use past cases, with personal details removed.

III. Days 61–90: Launch and improve

Test in live work. Use the new workflow on real jobs, alongside the old way at first.

Appoint champions. One person per team who helps colleagues and collects feedback.

Set up review and ownership. Agree who checks outputs, who maintains each template or tool, and how problems get reported.

Make it part of daily work. Add the new steps to your SOPs and onboarding.

Measure and decide. Look at usage, time saved, quality and staff feedback. Then choose what to scale, fix or drop next quarter.

By day 90, a realistic goal is at least one workflow in regular use, and a clear plan for the next. Some solutions will need longer, and that's fine.

IV. Your 90-day checklist

Days 1–30

  • Run a short AI skills survey
  • Hold hands-on training using real tasks
  • Publish simple AI usage guidelines
  • Collect three use cases per team

Days 31–60

  • Map two or three workflows
  • Pick one or two to build
  • Create prototypes and test them on past cases

Days 61–90

  • Pilot in live work
  • Name a champion in each team
  • Assign review and ownership
  • Measure, update SOPs and plan the next quarter

V. Start with your people, not your tools

An AI-ready workforce isn't built by one workshop or one software purchase. It's built by giving people the skills, the time and a clear path from learning to using.

AI Workforce's 90-Day AI Transformation for SMEs follows this same Learn, Design, Build, Launch journey, combining practical training modules with guided prototyping and optional implementation consultation. New to AI? Start with our practical guide for SMEs.

Jason is an illustrative example based on a typical SME.