Applied AI research at the frontier of customer experience
We're building the AI systems that power 1 billion+ customer conversations. Our research doesn't live in papers — it lives in production, at enterprise scale, every day.
We build what ships
Every research project at Ada Labs targets a production deployment. Our GARAGe retrieval system processes 1.5 trillion tokens monthly. Our Reasoning Engine powers 350+ enterprise deployments. Research here means impact.
We solve CX-specific problems
General-purpose AI is impressive. Domain-specific AI is transformative. We focus on the unique challenges of customer experience: real-time voice reasoning, multi-turn conversation management, hallucination prevention at scale, and agent-to-agent collaboration.
We’re hiring
The best AI Practitioners want hard problems with real-world impact. We have both. Join a team building at the cutting-edge of what's possible in agentic CX.
Inside the lab
The research, ideas, architectures, and findings powering Ada's work — documented.
Unified reasoner evaluation science Ada's Unified Reasoner replaced its Modular Reasoner with a re-baselined eval harness and Legitimacy Classifier. Adversarial pass rate: 88% to 97%.
Ada's AI agents detect and suppress automated emails, preventing infinite loops (~8,000 per week) and enabling earlier, configurable survey delivery.
How Ada built a per-conversation quality-scoring judge that scales to millions of conversations and stays calibrated to each customer's definition of "good."
DE replays production conversations through modified prompts & models, using an LLM-as-judge. Verdicts aggregate into win-rate metrics; traces into themes.
Replacing Ada's fastText language detector with an LLM tool call lifted recall from 78.6% to 97.9%: call the tool first and condition on script, not confidence.
Migrated knowledge and coaching retrieval to turbopuffer and Cohere Embed v4, improving latency, recall, and headroom to scale per-tenant namespaces.
Research in motion
Future research thrusts that ensure Ada’s platform not only keeps pace, but stays ahead of the curve.
-
Agent-agent collaboration
Specialized AI agents that work together on complex customer issues, collaborating to share context, negotiate resolution strategies, and learn from each other's outcomes.
-
Voice experiences
Real-time voice AI that reasons, not just transcribes. Dual reasoning architecture for latency optimization. Playbook-driven voice experiences that work across channels. The future: seamless voice-to-text-to-action transitions within a single conversation.
-
Long-running tasks
Customer issues that span hours or days — order tracking, insurance claims, multi-step returns. Agents that maintain state, check in proactively, and resolve asynchronously.
-
AI agents working behind the scenes
Proactive CX that analyzes patterns, preps responses, identifies at-risk customers, and recommends process improvements —all without waiting for a customer to initiate contact.
The Ada Labs blog
Ada engineers get 48 hours and the freedom to build. The latest Hack Days produced five standout projects, most of which are already on the roadmap.
Read
Discover how Ada's engineers used Claude Code to break through the testing barrier on legacy code refactoring, and what it really takes to make AI reliable.
Ada's GARAGe improves AI agent accuracy by enhancing retrieval, not just generation. See how two-hop indexing doubled semantic relevance and boosted precision.
Solve hard problems. Ship real things.
Breakthrough ideas. Real-world application. Be part of the team building what comes next.
- Senior Machine Learning Scientist Remote - Canada