Blog - Engineering
The Data Comes First: Mining Real Conversations for Test Coverage
Learn how Cresta mines historical data, synthetic customers, knowledge-base Q&A pairs, and post-launch feedback to build test coverage that reflects how customers actually behave.
Evaluating Speech-to-Text Quality: Beyond Word Error Rate
A look at why Word Error Rate alone isn't enough to evaluate speech-to-text quality, and which metrics actually predict downstream AI performance in contact center environments.
The Three Pillars of Voice Integration: Building Hybrid AI Contact Centers That Work With Your Existing Infrastructure
Learn the three pillars of voice integration—and how to connect AI agents to your current systems with speed, flexibility, and confidence.
Designing the AI Agent Supervision Experience
Learn how Cresta's Product Design team translated five UX imperatives into the Agent Operations Center, giving supervisors real-time control to guide both AI agents and human agents.
Cresta Crew: Anthony Mein, Forward Deployed Engineer
Meet Anthony Mein, one of our pioneering Forward Deployed Engineers, and learn about his journey from biology lab AI to building enterprise AI agents at the forefront of customer experience.
.webp)
Why AI Agent Evaluations Fail — and How the Swiss-Cheese Model Prevails
Learn about Cresta's forward-deployed team and how they approach building and iterating on AI agents.
Automation Discovery: Designing Systems to Extract Blueprints from Conversation Data
See how Cresta’s Automation Discovery turns unstructured conversation data into clear workflow blueprints, helping teams identify repeatable patterns, uncover edge cases, and assess automation readiness with more confidence.
Beyond Audits: How We Actually Test Our Security
Learn about Cresta's approach to testing our security from all angles in this post on red teaming exercises
When the Call Runs Too Long: Modeling Outcomes for Long Conversations
In this post, we discuss four practical approaches for modeling outcomes on long conversations, each suited to different use cases and constraints.
How Cresta Scales Data Annotation With a Human-Supervised Multi-Agent System (MAS)
How do you scale high-quality data annotation without sacrificing rigor? In this post, we explore our multi-agent approach—combining multiple LLM annotators, structured deliberation, and human oversight—to replicate the discipline of expert human workflows at enterprise scale.
Crafting a Natural-Sounding AI Voice
Go behind the scenes of Signal, Cresta’s AI concierge, and explore the technical foundations that power its conversational experience.
Evaluating AI Voices – What Does It Mean to Sound “Good”?
"Your AI Agent doesn't sound good"—this critique is perhaps the deepest dread for anyone building a voice agent.
How We Built a State-of-the-Art Research Agent for Call Center Conversation Analytics
Take a deep dive into how we built Cresta AI Analyst, and all of the lessons learned along the way.
Building and Deploying Production‑Grade AI Agents: Cresta’s End‑to‑End Approach
In this post, we walk through Cresta’s full lifecycle for building and deploying production-grade AI agents.
Why You Can’t Trust Out-of-the-Box Evaluators
Generic AI evaluators promise plug-and-play accuracy—but fall short where nuance and domain context matter most. This post breaks down why out-of-the-box LLM evaluators mislead enterprise teams, how misalignment erodes trust, and how Cresta’s expert-aligned, transparent evaluation framework turns measurement into a foundation for reliable AI performance.
No Blog posts match your search
Try changing your search terms.
Ready to see the results for yourself?
See the unified platform for human and AI agents in action with a personalized demo
