How to Identify Processes That Can Be Automated with AI: The 6 Signs of a Good Candidate
By Pedro Sonda, dooop
A process is a good candidate for AI automation when it meets six signs: it occurs frequently; involves reading, interpreting, or producing text and documents; has known and manageable exceptions; uses accessible data and systems; has errors that are easy to detect and correct; and is stable without frequent rule changes. Processes with high volume and fully fixed rules can be automated without AI, using traditional automation. Processes with high variation and little standardization need to be organized before any automation.
Updated October 2026.
Where Companies Are Already Automating
In finance, accounts payable automation was one of the most cited AI uses in 2025, by 37% of finance functions using AI, alongside knowledge management (49%) and error and anomaly detection (34%) (Gartner, 2025). In Brazilian industry, administration is the area where companies using AI apply it most (87.9%) (IBGE, 2025).
Interpretation: Administrative processes involving documents and volume are the most common candidates. But the same signs can be found in operations, customer service, and sales.
dooop Framework: The 6 Signs
| # | Sign | Question | Positive Sign |
|---|---|---|---|
| 1 | Volume | How many times per month? | Hundreds or thousands |
| 2 | Text and Documents | Does the work involve reading, interpreting, or writing? | Invoices, contracts, emails, forms, tickets |
| 3 | Known Exceptions | Do we know what to do when something deviates from the standard? | Documented or documentable exceptions |
| 4 | Accessible Data | Is the information in accessible systems? | Systems with integration or export capability |
| 5 | Detectable Error | Would an error be noticed and corrected in time? | Possible review before impact |
| 6 | Stability | Do the rules change frequently? | Rules stable for months |
The 4 Groups of Processes
| Group | Characteristics | Recommended Action |
|---|---|---|
| A. Rule-Based Automation | High volume, fixed rules, structured inputs | Traditional automation; AI not necessary |
| B. AI Automation | High volume, text or documents, known exceptions | Good candidate; start with human review |
| C. Decision Support | Complex decision, high error cost | AI prepares, person decides; see How to Choose Between Automation and Decision Support |
| D. Organize First | Many exceptions without rules, scattered data, unstable process | Standardize and organize before automating |
Differences between rule-based automation, generative AI, and agents are explained in Generative AI, Agents, and Automation: What’s the Difference?.
Generic Examples by Group
| Process | Likely Group | Why |
|---|---|---|
| Copying data between two systems with the same format | A | Fixed rules, no interpretation |
| Reading invoices from different suppliers and posting in ERP | B | Varied documents, known verification rules |
| Approving credit for new customers | C | Impactful decision, requires judgment |
| Handling complaints resolved differently by each area | D | No defined standard |
How to Conduct the Survey
- List the processes in an area with approximate volume and time.
- Apply the 6 signs to each, scoring from 1 to 3.
- Classify into the 4 groups.
- Select candidates from group B with the highest volume and pain points.
- Verify if the process should really be automated; see When a Process Should Not Be Automated with AI.
For the level of individual tasks rather than processes, see How to Identify Repetitive Tasks That Can Use AI.
What to Do Now
- Choose an area and list ten processes.
- Apply the 6 signs with the manager and the executors.
- Map in detail the two best candidates; see How to Map Processes Before Applying Artificial Intelligence.
To find and prioritize processes methodically, explore dooop’s discovery at dooop.com.br/discovery-de-oportunidades-de-ia.
Frequently Asked Questions
Does AI replace the process automation we already have? Not necessarily. Rule-based automations remain useful for predictable steps. AI expands automation to steps involving text, documents, and variation.
Can a process with few occurrences be automated? It can, if the solution is simple and inexpensive. Projects involving integration and development usually require higher volume to justify the investment.
Also Read
About dooop
dooop is a consultancy based in Caxias do Sul (RS) that prepares leaders and organizations to transform artificial intelligence into business capability through strategy, governance, and training.