All articles
- AI automation
- Strategy
How to pick your first AI automation (and skip the science project)
A simple scoring method to find the one workflow where AI pays for itself in weeks, not quarters.
Hassan Mujtaba · Oct 20, 2026 · 1 min read
Most teams start their AI journey in the wrong place: with the most impressive idea instead of the most profitable one.
Score every candidate workflow
List the repetitive workflows in your business and score each from 1–5 on:
- Volume: how often does it happen? Daily beats monthly.
- Time per run: minutes of skilled human time each time.
- Rules vs judgement: clear rules and examples automate well.
- Data access: can a system read the inputs today (email, PDFs, a database, an API)?
- Cost of a mistake: low-risk first; keep a human review step for the rest.
Multiply volume × time to get hours per month, then favour the highest score with low risk.
Good first projects
- Reading invoices, contracts or reports and filling a system of record
- Qualifying inbound leads and drafting the first reply
- Turning meeting notes or voice memos into structured summaries
- Internal search over policies, SOPs and past tickets
What to avoid first
- Anything customer-facing with no human in the loop
- Workflows whose data lives in people's heads
- "Replace a whole team" projects. Start with one step of the process.
Next step
If you want a second opinion, our $499 AI & Automation Audit maps your workflows and hands you a build-ready plan, credited back in full if you build with us.
