AI Nodes
Two kinds of AI you can combine in one workflow: Decision AI picks from outcomes you define, and Generative AI creates and interprets content.
Kovca has two kinds of AI nodes. Both sit under AI Nodes in the Studio node library, and a workflow can use both alongside logic, actions, and triggers.
Decision AI
Make decisions from defined possibilities. Fast, focused evaluations that classify, score, or determine outcomes using criteria you define.
- Classify: Choose from predefined categories.
- Score: Evaluate against defined criteria.
- Decide: Answer yes/no or another bounded outcome.
- Custom Decision AI: Create reusable decision capabilities.
Defined, predictable output fields · Built for lightweight, high-volume evaluations · Clear decision criteria · Fixed price per evaluation.
Generative AI
Generate and interpret content with flexible instructions. Create content, summarize information, extract data, transform text, and reason through open-ended problems.
- Generate Text: Write emails, replies, messages, and descriptions.
- Summarize: Condense documents, conversations, and earlier results.
- Extract Data: Pull fields like names, dates, and amounts out of text.
- Transform: Rewrite, translate, or reformat text.
- Reason & Analyze: Explain risks, causes, or trade-offs.
- Custom Generative AI: Create reusable generation capabilities.
Flexible natural-language instructions · Text or structured output · Content generation and interpretation · Metered by usage (tokens).
Which should I choose?
Ask what the step produces. If the answer is one of a set of outcomes you can list in advance (a category, a priority, a score, yes or no), use Decision AI. If the step creates or interprets content (a reply, a summary, extracted details, a rewrite, a written analysis), use Generative AI. Generative AI can return structured fields too; the difference is what each is built for, not a hard limit.
| Task | Choose | Why |
|---|---|---|
| Classify incoming emails by topic | Decision AI | It chooses from categories you define. |
| Decide whether a return meets your policy | Decision AI | The answer is yes, no, or another outcome you define. |
| Score a lead against your criteria | Decision AI | It rates against levels you describe. |
| Write a personalized reply | Generative AI | It creates new content. |
| Summarize a call transcript | Generative AI | It condenses content into new text. |
| Extract order details from an email | Generative AI | It pulls open-ended values (names, dates, amounts) out of text. |
| Translate or rewrite a message | Generative AI | It transforms content. |
Combine them
Many workflows use both. For a support request: Decision AI classifies the issue and its urgency, a condition routes it, an HTTP action looks up the customer, Generative AI writes a personalized reply on the path that needs one, and an HTTP action sends it. The Studio template Support auto-reply does exactly this, and the Studio assistant builds workflows the same way: Decision AI for decisions, Generative AI only where content has to be written or interpreted.
Speed and pricing
- Decision AI answers in about a tenth of a second. Each successful run is one evaluation, at a fixed price per evaluation. See Billing.
- Generative AI usually takes a few seconds. It never uses evaluations; it's metered by tokens against a monthly Generative AI allowance and isn't billed during the beta. See Generative AI.