Can AI classify and summarise end-user answers?
Yes. Configurable AI classification and summarisation can process information within end-user journeys, with the service builder choosing the source, purpose and output.
These features extend AI assistance into the journey itself. The builder defines the purpose, source information and intended result so the output fits the particular service.
Two configurable capabilities
| Feature | Builder configuration | Example use |
|---|---|---|
| Classification | Define buckets and describe what belongs in each, then choose the content to classify. | Suggest a category for a reported problem, for confirmation or later workflow. |
| Summarisation | Choose source values and summary instructions, such as desired length, format and facts to preserve. | Pre-populate a short description for review or produce a derived summary for later use. |
Configure AI for a clear service purpose
A facilities-reporting service might classify a description into “Heating and ventilation”, “Water and plumbing” or “Needs review”, using descriptions written by its builder. A later page could offer a short summary for the person to check and edit.
For summarisation, an instruction could ask for three short sentences preserving the supplied location and start date without adding facts or promising a resolution. Destinations include pre-population and derived or calculated fields.
Plan review and fallback behaviour
Keep the original information available, decide whether the person should confirm the generated result, and provide a usable route when content is ambiguous or processing is unsuccessful. Exact arithmetic and fixed eligibility rules remain suited to ordinary calculations and conditions.
These examples illustrate service-design choices. Configure and test the source information, output and handling of unsuccessful results for your service; confirm feature access for your account.
Read the guides for AI classification and AI summarisation, or explore AI capabilities for Builder assistance and related features.