Automation discovery (also called workflow discovery) is the process of finding which tasks are worth automating and capturing how they work — before you build anything. It answers "what should we automate, and what does it involve?" Recording a task once makes discovery instant: you capture the exact steps as you do the work, instead of guessing or documenting from memory.
Most automation advice starts too late — at "pick a tool and build." But the reason automation projects fail usually isn't the tool. It's that no one ever pinned down which workflows were worth automating, or captured precisely how they run. Automation discovery is the step that fixes both, and it's where the leverage is.
What is automation discovery?
Automation discovery is the practice of finding and assessing automation opportunities, then capturing how the chosen workflows actually work, so they can be automated reliably. It's the bridge between "we should automate some of this" and "here's the exact workflow to build." Skip it, and you're automating a guess.
Automation discovery vs. workflow discovery
The two terms are used almost interchangeably, with a subtle difference of scope:
| Term | Scope | Question it answers |
|---|---|---|
| Automation discovery | Broad — across teams/org | Which processes should we automate? |
| Workflow discovery | Specific — one task/workflow | How exactly does this task run? |
In practice you need both: automation discovery to prioritise, and workflow discovery to capture each chosen workflow precisely enough to build. For the deep dive on the capture side, see what is workflow automation discovery.
How to do automation discovery
1. Inventory the repetitive work
List the recurring, rule-based tasks people do — the weekly reports, the data entry, the onboarding steps, the triage. A quick filter: if it's done the same way more than once a week and a checklist could describe it, it's a candidate.
2. Capture how each candidate actually runs
This is where discovery is usually lost. Documenting from memory bakes in an out-of-date, edge-case-free version. Capturing from a real run keeps the truth. The fastest way is to record the task once as it's actually done.
3. Prioritise by ROI
Score candidates by frequency × time saved, and favour the most rule-based (they automate cleanest). Rank, then start at the top. The full prioritisation framework is in our workflow automation discovery guide.
4. Build the highest-value ones
Export the captured workflow to your automation platform and ship it. Because discovery already produced a precise spec, building is a copy-paste rather than a research project.
Discovery methods compared
- Manual audit / interviews — thorough but slow and lossy; the workflow is reconstructed from memory.
- Process & task mining — analyses system logs or activity to surface candidates at scale; powerful but heavy and observational. (See discovery vs process mining vs task mining.)
- Recording-based discovery — capture one task by doing it once; the output is a buildable workflow, not a chart. Fastest path for most teams.
You can't automate a workflow you haven't discovered. The tool was never the hard part — knowing exactly what to build is.
How Spion does automation discovery
Spion is a free Chrome extension built for discovery-by-recording. You record a task once; Spion reconstructs the steps, inputs and decisions into a clean, editable workflow — then exports it as a ready-to-run automation (Claude, Make, Zapier, n8n) or a step-by-step PDF. Discovery and capture, collapsed into a single recording.