Inspiration
Sales teams waste countless hours manually researching prospects—digging through LinkedIn, company websites, news articles, and financial reports. We saw an opportunity to transform this tedious process into something instant and intelligent. The inspiration came from watching sales reps spend 60% of their time on research instead of actually selling. We asked: what if AI could be the research analyst on every sales team?
What it does
Clariq is an AI-powered sales research partner that automates the entire prospect research workflow. Users submit a research job (company name, context, specific questions), and Clariq:
- Orchestrates multi-source data collection via Exa's neural search
- Synthesizes findings using Cerebras's lightning-fast LLM (llama-4-scout)
- Delivers decision-ready briefs with actionable insights
- Enables voice-powered research sessions through LiveKit integration
- Tracks job status asynchronously via REST API
Sales teams get from "who is this prospect?" to "here's how to win them" in minutes instead of hours.
How we built it
Backend: FastAPI powers our async research orchestration engine. We integrated Cerebras Cloud SDK for blazing-fast AI summarization, Exa API for intelligent web search, and LiveKit Agents for voice workflows. Appwrite handles authentication and data persistence.
Frontend: Next.js 15 + TypeScript delivers a responsive dashboard with real-time job tracking. We used shadcn/ui components styled with Tailwind for a polished, professional interface.
Architecture: Designed for scale from day one—async job queues, stateless API design, JWT-based auth, and cloud-ready deployment on Railway (backend) and Vercel (frontend).
Challenges we ran into
Async orchestration complexity: Coordinating multiple API calls (search, scraping, summarization) while maintaining job state was tricky. We implemented a robust status polling system and error handling.
LLM prompt engineering: Getting consistently high-quality, actionable insights required iterating on prompt templates and few-shot examples for Cerebras models.
Voice integration: LiveKit's real-time audio pipeline required careful configuration of VAD, transcription, and TTS components to feel natural and responsive.
Rate limiting & costs: Balancing API quotas across Exa, Cerebras, and LiveKit while keeping the service performant and cost-effective.
Accomplishments that we're proud of
✅ End-to-end working product deployed in production with both web and API access
✅ Sub-10-second research reports thanks to Cerebras's incredible inference speed
✅ Voice-enabled workflows that let users interact naturally with research data
✅ Clean, documented API that developers can integrate into existing sales tools
✅ Real-world validation: Successfully generated actionable briefs for 50+ test companies
What we learned
- Speed matters: Using Cerebras cut our report generation time from 45 seconds to under 10 seconds—a game changer for user experience
- Context is king: Neural search (Exa) dramatically outperformed keyword search for finding relevant, recent company information
- Voice UX is hard: Natural conversational flows require thoughtful design of interruptions, context switching, and error recovery
- Async patterns scale: Building with queues and webhooks from the start made it easy to handle concurrent research jobs
What's next for Clariq
🎯 AI Voice Agent (Q2 2025): Launch public voice rooms where AI agents conduct discovery calls using research context. Includes RAG-powered continuous learning from conversation transcripts.
📊 CRM Integrations: Native plugins for Salesforce, HubSpot, and Pipedrive to trigger research automatically when deals enter pipeline stages.
🧠 Multi-agent workflows: Orchestrate specialized AI agents (competitor analyst, financial researcher, news monitor) that collaborate on complex research tasks.
📈 Analytics dashboard: Track research effectiveness, identify winning patterns, and correlate insights with deal outcomes.
🌍 Multi-language support: Expand beyond English to serve global sales teams.
Our vision: Make every salesperson feel like they have a team of research analysts working 24/7 to help them win deals with confidence.
Built With
- appwrite
- cerebras-cloud-sdk
- deepgram
- exa-api
- fastapi
- livekit
- next.js-15
- openai-tts
- python
- railway
- react
- shadcn/ui
- silero-vad
- tailwind-css
- typescript
- uvicorn
- vercel

Log in or sign up for Devpost to join the conversation.