"AI training for everyone" is the worst kind of learning program. It's generic enough to be irrelevant to everyone, and specific enough to be useful to no one.

The teams that are actually building AI fluency aren't doing it through company-wide sessions. They're doing it role by role — matching the skills people learn to the work they actually do every day.

Here's how to build those paths, starting from scratch, without a curriculum team or an L&D budget.

The Core Principle: Start with the Job, Not the Tool

Most AI learning content starts with the tool. "Here's what ChatGPT can do. Here are the features. Here's how to log in."

Role-specific learning paths start with the job. "You're a customer support rep. Here are the three workflows where AI will save you the most time this month. Here's how to use it for each one. Here's how to tell if it's working."

The difference in engagement is dramatic. People learn when they can immediately apply what they're learning. Generic AI content doesn't clear that bar.

Four Roles, Four Starting Points

Customer Support

Support teams have the fastest ROI from AI — and the most obvious starting point.

The core skill: AI-assisted response drafting for common ticket types.

Learning path: 1. Week 1: Identify the 5 most common ticket types your team handles. For each, write one detailed prompt template that captures: the situation, the tone you want, any constraints (refund policies, response time commitments), and the format you need. 2. Week 2: Use AI to draft responses for those 5 types. Compare AI drafts to human drafts. Note where AI gets it right and where it consistently misses context. 3. Week 3: Build a shared prompt library — one template per common ticket type. Make it collaborative so the team improves it together. 4. Week 4: Add one new use case: using AI to summarize long email threads before responding.

Milestone to aim for: AI drafts are usable for at least 3 of the 5 ticket types with minor edits only.

What to avoid: Using AI for emotionally sensitive tickets early. Build the skill on factual, process-driven tickets first.

Sales

Sales AI learning takes longer to show results — the skill is more nuanced — but the ceiling is high.

The core skill: AI-assisted research and personalized outreach.

Learning path: 1. Week 1: Pre-call research. Before every discovery call, use AI to summarize what's publicly known about the prospect's company (industry, recent news, size, likely pain points). Compare the quality of AI research to your current pre-call prep. 2. Week 2: Follow-up email drafting. Use AI to draft follow-up emails after calls — include a brief about what was discussed, what the prospect cared about, and what the next step is. Let AI produce a draft; edit it to your voice. 3. Week 3: Proposal summaries. If you're writing proposals, use AI to draft the executive summary section based on notes from the discovery call. 4. Week 4: Objection prep. Use AI to help you anticipate objections and draft responses before big calls.

Milestone to aim for: Pre-call research time cut by 50%. Follow-up emails drafted and edited in under 5 minutes.

What to avoid: Using AI-generated outreach without personalizing it. Generic AI emails perform worse than no email at all.

Operations

Ops people are often skeptical of AI — and appropriately cautious. Start with read-only use cases before moving to any AI-assisted output.

The core skill: AI-assisted summarization and analysis of documents, reports, and long threads.

Learning path: 1. Week 1: Vendor contract review. Paste a vendor contract into Claude or ChatGPT and ask: "Summarize the 5 key terms. Flag any unusual provisions I should review carefully." Compare to doing it manually. 2. Week 2: Report summarization. Take a long internal report or spreadsheet summary and use AI to pull out the key findings and action items. 3. Week 3: Process documentation. Use AI to help you write or clean up a standard operating procedure. Give it the raw notes; let it produce a structured draft. 4. Week 4: Meeting follow-up. Start using AI to convert messy meeting notes into structured summaries with action items and owners.

Milestone to aim for: Document review time reduced. At least one internal process documented faster with AI than without.

What to avoid: Using AI output directly in anything that needs legal review or external sharing without human verification first.

Marketing

Marketing is often the first team to experiment with AI — and the first to plateau because they only use it for first drafts.

The core skill: Full content workflow with AI as a research and drafting partner, not just a text generator.

Learning path: 1. Week 1: Reframe the workflow. Stop asking AI to "write a blog post." Start asking it to help you think through the post first: "What are the best angles for a piece about [topic] targeting [audience]? What objections might readers have?" Use AI for ideation before drafting. 2. Week 2: Research synthesis. Give AI a brief and ask it to summarize what's already been said on this topic and where there are gaps. Build your differentiated angle from that. 3. Week 3: Structured drafting. Draft with AI in stages — hook first, then structure, then section by section — rather than generating an entire post at once. Edit each stage before proceeding. 4. Week 4: Repurposing pipeline. Take one finished piece of content and use AI to repurpose it: LinkedIn post, email newsletter blurb, short-form tweet thread. Build a reusable prompt for this.

Milestone to aim for: Content production time per piece reduced by 40%. Quality maintained or improved.

What to avoid: Publishing AI-generated content without meaningful editing. It's obvious and it underperforms.

How to Build Your Own Role Path (Template)

For any role not covered above, use this framework:

  1. List the 5-7 most time-consuming or repetitive tasks in this role
  2. For each task, ask: is this task primarily about (a) producing text, (b) analyzing information, (c) researching something, or (d) communicating? AI is good at all of these — the approach differs.
  3. Pick the 2 tasks with the highest time cost and clearest AI fit. Those are your starting points.
  4. Design a 4-week progression: Week 1 is observation (try AI on this task and see what happens), Weeks 2-3 are skill-building (improve through iteration), Week 4 is integration (AI is part of the normal workflow for this task).
  5. Define one measurable milestone — time saved, quality benchmark, or output volume.

Making It Stick: The Role-Path Review

Once you've built paths for each role, check in monthly:

  • Is the person actually using AI for the tasks in their path?
  • What's working? What keeps getting skipped?
  • What new use cases have they discovered on their own?

The answers reshape the path for the next month. This is how learning becomes continuous instead of episodic.

OpenSkills AI builds these role-specific paths automatically — based on industry, job function, and individual skill gaps. Each employee gets a personalized learning path that adapts as they progress, with AI coaching to help them get unstuck.

See how it works for your team or start for free.

For more on the foundation this sits on, read what a learning culture looks like at a 12-person company.