Talkio

Talkio helps sales reps keep context across multiple calls by organizing post-call insights into a clear client history, so every conversation builds on the last and moves deals forward.

Comment 4

Inspiration

We looked into government consulting companies and noticed that they would only have 1 agent at a time helping. This made us want to build this project

What it does

Post-call AI analysis platform that transforms your sales conversations into actionable intelligence.

How we built it

We built a post-call sales intelligence platform that captures sales calls, analyzes them after they end, and turns conversations into structured, actionable insights.

Sales calls are handled through Twilio, which captures call metadata, recordings, and transcripts once each call ends. This data is securely delivered to our backend via webhooks, where it is analyzed post-call to generate structured summaries, client context, and clear next steps.

Our backend processes each call using AI post-call analysis, generating structured outputs such as summaries, key discussion points, outcomes, and clear next-action recommendations. These insights are saved per organization and per client, preserving context across future calls.

The application is built with Next.js and TypeScript, allowing us to combine a fast, modern frontend with server-side logic in a single codebase. MongoDB is used for semantic vector search and also for storing organizations, clients, calls, and historical context in a flexible schema that evolves as conversations grow.

Authentication is handled through Clerk. For UI and design, we used Tailwind CSS and Radix UI to build a clean, accessible, enterprise-grade interface.

We deployed the application on DigitalOcean, which hosts the platform and powers two AI Agents: “Max,” an AI onboarding agent on the landing page that helps users explore the product and an in-product B2B sales supervisor providing help and insights to the sales rep.

For sales training and demo scenarios, we integrated ElevenLabs to simulate realistic buyer behaviour AI Agents, allowing reps to practice conversations and improve their pitch based on feedback.

Challenges we ran into

Some challenges were integrating multiple AI agents, having a common knowledge base around them, and also frontend components.

Accomplishments that we're proud of

We are proud of the variety of features and the vast ability to integrate more and scale it in the future

What we learned

We've learned that patience is key and to be organized by using the whiteboard to organize our needs/wants.

What's next for Talkio

We are wanting to have real-time conversations and analytics. Integrating Microsoft Teams, Zoom for reachability of clients.

Built With

Share this project:

Updates