← Work Case study · Top of Mind · Belgium
Agentic capture for
voice, text and photos.
Agent workflows that pull contacts, tasks and relationship context out of raw inputs, on iOS and Android.
Raw inputs
- Voice note
- Text message
- Photo
Extraction agent
edge functions
Structured output
- Contacts
- Tasks
- Relationship context
(01) The work
What I
built.
- 01
Agent workflows
Built AI-driven agent workflows using OpenAI Codex that automatically extract contacts, tasks, and relationship context from voice, text, and photo inputs.
- 02
Extraction pipelines
Designed and tuned prompt logic for custom edge functions and Supabase-backed extraction pipelines, turning raw capture inputs into structured actionables with minimal user friction.
- 03
Cross-platform mobile
Developed cross-platform mobile (iOS and Android) features using React Native and Codex, integrating AI inference into mobile and backend flows to deliver a seamless agentic capture experience with reduced manual data entry.
(02) At a glance
Inputs
Voice · Text · Photo
Outputs
Contacts · Tasks · Context
Platforms
iOS · Android
Start a project