How to turn every support conversation into a living source of new, report‑ready insights.
1. Why Generative?
Traditional fixed categorization asks your AI to pick from a pre‑set list. That works for stable, well‑defined buckets but the content in real customer conversations changes as the business changes. Generative AI workflows let the model answer open‑ended questions in its own words, then store that answer as a re‑usable code. The result is a system that, like an experienced researcher, can:
Spot a brand‑new trend the first time it appears.
Re‑use that same phrase for every matching ticket that follows.
Keep your dataset both structured and forward‑looking.
2. What is a Qvasa Generative AI Workflow?
A generative workflow is a set of rules you write once, telling Qvasa’s AI how to extract a concise phrase from each ticket. It contains:
Component | Purpose |
Prompt Question | The exact question posed to the AI (e.g. “What was the product deficiency?”). |
Business Process Context | Extra background so the AI understands your product, policies, and vocabulary. |
Word‑length Bounds | Min/max words for the answer. Shorter bounds force tighter, reusable codes; wider bounds allow creativity. |
Re‑use Strategy | Qvasa first tries to match an existing code in its database; if none fits, it invents a new one and saves it. |
Every time a ticket flows through the workflow, the AI responds with one phrase that meets these rules. That phrase is immediately written to Qvasa’s database as a Generative Code.
3. Example: “Product Deficiency” Workflow
Question: What is the product deficiency?
Rules: 3–7 words, describe the issue succinctly, prefer an existing code if it partially matches.
How it behaves
First occurrence – A user reports that clicking Sign In returns an error. The AI sees no match in its list, creates “Signin button error”, and stores it.
Later tickets – Every time a similar complaint appears, the AI picks that same phrase instead of creating variants like “login error” or “can’t sign in.”
4. Reportability: Turning Codes into Dimensions
Because generative codes are stored just like any other attribute, they become:
Base objects in all Qvasa reports – sum, count, pivot, or trend them over time.
Global filters – slice any dataset by Product Deficiency = “Sign‑in button error” or combine multiple codes.
Drill‑paths – click a bar in a chart to open the underlying tickets that share the same generated code.
No extra ETL, no SQL. Just create the workflow and the data appears everywhere you analyze.
5. Designing Your Own Workflow
Define the question. Think of the single line you wish every ticket could answer (e.g. What feature is missing?).
Provide context. Tell the AI what counts as a “feature”, what product areas exist, and any vocabulary to avoid.
Set word bounds. 2–5 words for terse codes, 5–10 for richer phrases.
Connect to a categorization workflow. Data starts getting generated instantly
Tip: Keep questions narrow. A workflow asking “Why is the customer unhappy?” can work, but you’ll get cleaner analytics by splitting that into focused questions like “Which feature caused dissatisfaction?” or “What outcome was expected?”
6. Frequently Asked Questions
Q: What makes a generative code re‑usable?
Qvasa looks for a % token match with existing codes before inventing a new one, ensuring small wording differences don’t splinter your data. We also re-prompt the AI with similar codes before we save new ones to encourage code re-use.
Q: Can I edit or merge codes later?
Soon. Merging and managing codes manually is coming.
Q: How many workflows can I run?
As many as you like. Each runs independently and adds another dimension to your dataset.
7. Key Takeaways
Unlimited questions. Ask anything that can be answered in a short phrase.
Discovery‑ready. Capture issues you didn’t know existed and watch them trend.
Instantly analyzable. Codes appear in dashboards the moment the ticket is saved.
With generative workflows, Qvasa gives you a tireless analyst that reads every conversation, answers your questions in clear, consistent language, and turns those answers into first‑class data. The result: faster insights, stronger decisions, and a support operation that keeps pace with your customers.