Resona

Inspiration

We realized that some of the most valuable information in life exists inside conversations, yet most of it disappears immediately after meetings, networking events, lectures, and discussions. Existing AI tools focused heavily on transcription and summaries, but very few systems actually understood context, relationships, opportunities, or unresolved actions. We wanted to build a calmer, more intelligent memory layer that helps people remember what truly matters instead of overwhelming them with information.

What it does

Resona is an agentic AI memory system that transforms conversations into structured intelligence. It captures audio conversations, extracts people, opportunities, tasks, emotional signals, priorities, and follow-ups, then connects them into a long-term contextual memory graph. Instead of showing raw transcripts, Resona compresses intelligence into clarity by surfacing only what deserves the user’s attention.

How we built it

We built Resona using a modern TypeScript-based stack with scalable cloud infrastructure and AI-powered processing pipelines. The platform combines conversational understanding, contextual memory systems, and intelligent workflow automation to transform raw conversations into actionable insights. We also integrated productivity workflows and scheduling capabilities to help users seamlessly act on important follow-ups and opportunities surfaced by the system.

Challenges we ran into

One of the biggest challenges was avoiding cognitive overload. Early outputs were technically powerful but surfaced too much information at once. We had to rethink the system around “intelligence compression” rather than raw AI extraction. Another challenge was building reliable contextual memory linking between people, topics, and opportunities without making the product feel invasive or overwhelming. Designing calm frontend experiences for highly intelligent backend systems was also surprisingly difficult.

Accomplishments that we're proud of

We successfully built an end-to-end agentic memory pipeline that goes far beyond meeting summaries. Resona can identify opportunities, unresolved loops, emotional signals, relationship context, action items, and recurring themes from real conversations. We also created a working memory graph system capable of linking conversations, people, topics, and opportunities together. Most importantly, we built a product experience focused on cognitive relief rather than information overload.

What we learned

We learned that building AI products is less about exposing intelligence and more about compressing complexity into calm clarity. The hardest part was not transcription or extraction, but deciding what actually matters to the user. We also learned how important emotional UX, prioritization systems, and trust design become when building products around memory and personal conversations.

What's next for Resona

We also plan to explore an open ecosystem approach around conversational memory infrastructure by open-sourcing selected developer tools, MCP integrations, memory frameworks, and contextual AI building blocks while continuing to evolve Resona’s core intelligence, prioritization, and understanding engines privately. Future iterations of Resona will focus on much deeper semantic understanding, relationship intelligence, realtime contextual memory, and proactive AI workflows that can reason across conversations, people, tools, and long-term user context. Long term, we want Resona to become not just a product, but a foundational platform for contextual AI memory experiences. .

Built With

• api
• calendar
• css
• database
• gemini
• google
• google-cloud
• next.js
• shadcn/ui
• tailwind
• trigger.dev
• typescript
• vercel