Update case record using ai agents

[Apaul]

Is it possible to configure an AI Agent to automatically update the priority field based on the category?

For example, the agent should be able to read rules from the case description.

If the description states: “If Category is Billing, set Priority to 1”, then the AI Agent should interpret this rule and update the record accordingly.

[GlideFather]

Hi @Apaul,

first of all it's crucial to mention what table - incident?

Then this can be easily done by business rule, flow or client scripts. Depending if you want to have the Priority changed after the Category is selected on the form or after submission etc.

There's no need to use AI Agent for that. Because most of the current clients do not use it, maybe they will in future.

Is this a learning opportunity or real requirement?

See this no code BR - I don't have a Billing for category so I used different but it works for an example:

It would work if you would submit an incident. 

If you want to have it dynamically, you need to set onChange (catalog) client script.

[Dinesh Patel]

Yes, that’s possible 👍

You can configure an AI Agent (or automation layer) to read natural-language rules and then apply them to records. Essentially, you’d be combining NLP (Natural Language Processing) with a rules engine / workflow automation. Here’s how it could work:

---

1. Define the Input

Case description: e.g. “If Category is Billing, set Priority to 1.”

The AI Agent needs to parse such rules and map them to system fields.

---

2. Parsing & Interpretation

Use an LLM or NLP model to extract structured logic from text:

Condition: Category = Billing

Action: Priority = 1

This can be done by setting up a prompt or parser to always output JSON or a structured rule format.

Example JSON:

{
"condition": {
"field": "Category",
"operator": "equals",
"value": "Billing"
},
"action": {
"field": "Priority",
"value": "1"
}
}

---

3. Rule Storage

Store these parsed rules in a database or rules table.

This allows the agent to consistently apply them without re-parsing the same description every time.

---

4. Execution Layer

An automation engine / workflow engine checks incoming or updated cases.

When a case has Category = Billing, the engine executes the matching rule and updates Priority = 1.

---

5. Error Handling & Governance

If the agent finds ambiguous or conflicting rules, it should either:

Ask a human to confirm, or

Apply a default fallback (e.g. Priority = 3).

---

✅ So yes, the AI Agent can be configured to read rules in natural language and update the case priority automatically.
Depending on your platform (Salesforce, ServiceNow, Jira, or custom CRM), you could build this with:

LLM-powered parser + workflow automation (Flow, Rules Engine, or custom scripts).

Or integrate with tools like LangChain + RPA + database triggers.

[Apaul]

I used a subflow to achieve my goal but i wanted to use this approach as it seems interesting to me. Lets see if it works. Our main usecase is not the one I posted, it was just basic. Now to implement the real one this might help me.