How to Choose Your First AI Project

Every company wants the benefits of AI, but most teams stumble on the very first step:

‍What should we automate first?

The instinct is to think big – end-to-end transformation, every process redesigned, AI “everywhere.” But starting that way adds risk, slows adoption, and often leads to stalled or abandoned projects.

There’s a far more reliable path: start small, prove value fast, and pick the right workflow. And the right workflow follows a simple logic.

The Four Filters of a High-Success AI Workflow

Not every process is a good candidate for your first AI project. The goal isn’t ambition – it’s momentum. Early wins matter because they build confidence, unblock scepticism, and demonstrate that AI can actually deliver outcomes.

Here’s the practical filter that works across industries:

1. Repeatable
   The work happens daily or weekly. This ensures the impact is immediately felt and easy to measure.
   
   
2. Rule-based
   The workflow follows predictable steps or decision rules, even if they’re informal. AI works best when you can describe the logic behind the task, not just the task itself. Rule-based tasks are ideal for deterministic or structured LLM-driven workflows.
   ‍
   
3. High-friction
   Slow, manual, error-prone, admin-heavy – anything that drags teams down. These workflows produce fast, visible improvements when automated, which is exactly what you need at the start.
   ‍
   
4. Measurable
   You must be able to compare “before vs. after” in a way that leaders trust: cycle time, error rate, cost, throughput, or even quality. The measurement doesn’t need to be perfect; it just needs to be real.
   ‍

These four filters reduce the noise and narrow the universe of possible AI projects into a small circle of high-likelihood wins.

But there’s one more layer that separates good candidates from great ones.

Why ‘Hero Workflows’ Should Be Your Starting Point

Inside every business, there’s a small set of workflows that rely on a few people — but have outsized importance. A sales coordinator who manages bids. An operations lead who reconciles invoices. A finance analyst who handles reporting.

The work isn’t always glamorous, but it’s essential. And when these workflows improve, the whole business feels it.

This is what we call hero workflows:

‍High-impact, high-dependency, small-team processes that keep the organisation running.

Hero workflows are perfect for early AI adoption because:

• They affect revenue, margin, or risk directly
• They’re handled by 1–3 people, so change management is minimal
• Improvements are immediately visible to leadership
• They generate disproportionate value relative to their size

When you automate or augment hero workflows, you aren’t just saving time — you’re unlocking leverage. AI becomes a multiplier.

Why 30 Days Is the Sweet Spot

Your first AI project should not run for months. If it does, it becomes a research project — not an operational win.

A 30-day window forces clarity:

• What exactly is the workflow?
• What is the desired outcome?
• What’s the baseline?
• What’s the expected measurable change?

If you can’t define these in under an hour, you shouldn’t pick that workflow.

If you can define them clearly, you’ve got a great candidate — and your AI implementation moves faster.

Examples Across the Business

You don’t need to reinvent the business to find high-value candidates. Most companies have dozens of them:

• Sales: lead qualification, pipeline cleanup, outbound drafting
• Marketing: segmentation, briefs, research
• Finance: invoice processing, reconciliations, monthly reporting
• HR: onboarding steps, policy checks, documentation
• Operations: SOP adherence, ticket triage, customer updates

But again — don’t automate everything. Start narrow. Start where value is obvious. Start with one hero workflow.

The Bottom Line

Successful AI adoption isn’t about scale — it’s about sequence.

When you pick a workflow that is repeatable, rule-based, high-friction, and measurable, and when you layer on the logic of hero work, you create a 30-day proof point that gives your organisation belief that AI can drive outcomes and for your team can adopt it safely and in control.

Once you have that, the next project becomes easier. And the one after that. And the one after that.

🎧 Listen to the full episode on Spotify | Apple Podcasts | YouTube

‍