AI Survey Insights for a Research Platform
AI summaries, sentiment and a database-restricted chatbot for open-ended survey answers, plus panel tools.
Key Details
| Challenge | Open-ended survey answers were only processed up to 300 responses. The research company also needed to recruit and manage a respondent panel. |
|---|---|
| Solution | Three workstreams: AI summarisation and sentiment for open-ended answers, a chatbot limited to the client's own database, and tools to recruit and manage the panel. |
| Technologies | Vue, React, Node.js, NestJS, PostgreSQL, Python, FastAPI, Azure OpenAI, Llama 3.1, Mixtral, Qdrant, Pinecone, Azure, Mailchimp |
Technologies used
VueReactNode.jsNestJSPostgreSQLPythonFastAPIAzure OpenAILlama 3.1MixtralQdrantPineconeAzureMailchimp
Client
The client is an Australian civic-research company working with survey responses and a respondent panel. The client is not named here. The engagement was won. A completion date is not stated, so it is described as ongoing.
Product screenshots
What we built
- AI summarisation of open-ended answers beyond the previous 300-response limit
- Sentiment classification
- A data-agent chatbot restricted to the client's database
- Panel recruitment
- Panel management, including mailing-list integration
Engagement
A total headcount and an elapsed duration are not stated. Proposed effort ranges and a fee from the estimate are not published as delivery results.
Outcome
The scope is the survey-insight and panel work above. No measured outcome is stated.