Voice AI Agent
Budget: £20 – £250 GBP
I am building a call-based assistant that proactively phones users on iOS and Android, brick phone holds check-ins, and then remembers what was shared so the next conversation feels truly personal.
Here’s the flow I have in mind: the agent dials the user from a mobile app, greets them naturally, explores how they are feeling, and, when appropriate, offers coping strategies or simply listens with empathy. After each call it should securely log the dialogue, extract key wellness markers, and store both the raw transcript and structured data so those insights can be surfaced in later sessions. When it starts the next call, it must be able to reference previous discussions (“Last week you mentioned difficulty sleeping—how was your rest since then?”).
I’m comfortable with whichever stack you prefer as long as:
• Calls originate on iOS and Android (native or cross-platform).
• Speech recognition and TTS are high quality and low-latency.
• Memory includes historical conversation data and health & well-being fields that I can query through an API or dashboard.
• Privacy is baked in (data encryption at rest and in transit).
The most important is the voice has to be human (think personal plex level)
Deliverables
1. Working mobile prototype (APK + TestFlight build).
2. Server or on-device pipeline that handles speech → intent → response, plus memory retrieval and update.
3. Simple admin panel or documented endpoints to review conversation history.
4. Deployment instructions so I can replicate the environment.
Acceptance criteria: the agent completes a five-minute scripted check-in, references at least one past user detail, and saves the session to the database without errors.
If this sounds like a challenge you’re excited about, please outline the tech stack you’d use and any similar voice or wellness projects you’ve shipped.
Here’s the flow I have in mind: the agent dials the user from a mobile app, greets them naturally, explores how they are feeling, and, when appropriate, offers coping strategies or simply listens with empathy. After each call it should securely log the dialogue, extract key wellness markers, and store both the raw transcript and structured data so those insights can be surfaced in later sessions. When it starts the next call, it must be able to reference previous discussions (“Last week you mentioned difficulty sleeping—how was your rest since then?”).
I’m comfortable with whichever stack you prefer as long as:
• Calls originate on iOS and Android (native or cross-platform).
• Speech recognition and TTS are high quality and low-latency.
• Memory includes historical conversation data and health & well-being fields that I can query through an API or dashboard.
• Privacy is baked in (data encryption at rest and in transit).
The most important is the voice has to be human (think personal plex level)
Deliverables
1. Working mobile prototype (APK + TestFlight build).
2. Server or on-device pipeline that handles speech → intent → response, plus memory retrieval and update.
3. Simple admin panel or documented endpoints to review conversation history.
4. Deployment instructions so I can replicate the environment.
Acceptance criteria: the agent completes a five-minute scripted check-in, references at least one past user detail, and saves the session to the database without errors.
If this sounds like a challenge you’re excited about, please outline the tech stack you’d use and any similar voice or wellness projects you’ve shipped.