ListenLocal — approved public-service answers people can read or hear
The problem
Application and appointment instructions can be difficult to locate in agency documents. WeKnora can retrieve document-backed answers; VoiceStudio can turn approved text into local audio.
The idea
Why now
WeKnora supplies a self-hosted document knowledge stack and VoiceStudio supplies local speech generation, enabling an agency-controlled text-and-audio workflow.
What it combines
WeKnora retrieves passages from approved documents; VoiceStudio generates speech from the final staff-approved answer. The combination offers listening without maintaining separate audio scripts.
How to build it
- Scope. Choose an application or appointment with an existing next-step page. (1 day)
- Draft. Retrieve source passages and prepare answers for staff approval. (1 day)
- Publish. Generate and review audio, then publish approved answers and links. (1 day)
- Measure. Compare next-step arrival rates with the existing page. (1 day)
Who benefits
- Residents
- Read or hear reviewed instructions and follow the next step.
- Agency staff
- Control published wording, sources, and audio.
How it makes money
- Model
- Agency subscription
- Pricing
- $500 per agency per month for one service workflow
- Wedge
- One frequently visited application or appointment page
- Channels
- Local government digital-service teams · Public-service website contractors
Distribution
A local government agency pays and places the service on an existing application or appointment page. Public-service website contractors can bring it to additional agencies.
Why it wins
Unlike an open-ended voice chatbot, it publishes only reviewed answers with visible sources and next steps. The MVP does not clone voices or publish automatically.
Risks
An outdated document could produce incorrect instructions, and short audio may sound unclear. Assign each answer an agency owner, flag cited-document changes for reapproval, and have staff listen before publication.
Strongest objection left standing
The MVP still must establish that generated short-form audio is clear enough for public instructions.
Build it with
- Tencent WeKnora: open-source LLM knowledge platform (RAG + agent + self-maintaining Wiki)Retrieve passages from the agency's approved documents.
- VoiceStudio — open-source, fully-local ElevenLabs alternativeGenerate local speech from staff-approved answer text.
Repo to start from
listen-local - reviewed public-service answers with audio
Evidence
Debated by openai/gpt-6-sol, anthropic/claude-sonnet-5.5, qwen/qwen3.8-max-0902, moonshotai/kimi-k2.5. Judged by JEV.
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.