- AI Automation
- Strategy
Where to start with AI automation (hint: not with the flashiest demo)
· Compublue Team
The highest-ROI automation in your company is probably boring. Here is a practical framework for finding it, pricing it, and shipping it in weeks.
Every week we talk to companies that want "to do something with AI." The instinct is usually to start with the most impressive demo — a talking assistant, a content generator, something board-ready. The better starting point is almost always duller: the workflow your team complains about most.
Start with cost, not capability
Make a list of tasks your team repeats every week. For each one, estimate three numbers: how many times it happens per month, how long it takes, and what an error costs when it happens. Multiply. Sort. The top of that list is your AI roadmap — and it rarely matches the demo you saw on LinkedIn.
Document processing, inbox triage, data entry between systems, and first-draft responses are the usual winners. None of them are glamorous. All of them pay back in months.
The 70% rule
A useful automation does not need to handle every case — it needs to handle the common cases perfectly and route the rest to a human with full context. We aim for systems that fully automate around 70% of volume at launch. The review queue for the remaining 30% does two jobs: it protects quality, and it generates training data that shrinks the queue every month.
Teams that insist on 100% automation on day one usually ship nothing. Teams that accept a human-in-the-loop ship in weeks and improve from there.
Measure from day one
Before the automation goes live, capture the baseline: hours spent, error rate, turnaround time. Then instrument the automated system to report the same numbers. This is not bureaucracy — it is how you know whether to expand the program, and it is how the project survives its first budget review.
What a good first project looks like
Small enough to ship in four to eight weeks. Painful enough that the team cheers when it works. Measurable enough that finance believes the result. If you have a candidate workflow in mind and want a second opinion on whether it qualifies, we are happy to take a look — the workflow audit is the same first step we take with every client.