Cresta
AI for the enterprise contact centerEnterprise contact-center AI platform combining autonomous voice and chat agents with real-time coaching for human agents.
- Maker
- Cresta
- Founded
- 2017
- Pricing
- No free tier · from Custom
- Last checked
- September 2026
Verdict
Enterprise contact-center AI platform combining autonomous voice and chat agents with real-time coaching for human agents.
Score breakdown
vs Quickchat AI, the category leaderLarge enterprise contact centers with high call volume and dedicated CX operations staff
Key features
- AI Agent: autonomous voice and digital agent that handles conversations end to end
- Agent Assist: real-time suggestions, next-best-action and knowledge surfacing during live calls
- Conversation Intelligence: analysis across all conversations to surface trends and revenue opportunities
- Quality Management: automated scoring across 100% of conversations instead of sampled reviews
- Models fine-tuned on the customer's own conversation history
- Real-time coaching and behaviour tracking for supervisors
- Integrations with major CCaaS, telephony and CRM systems
- Enterprise security and compliance controls for regulated industries
Strengths
- +Covers the whole contact center in one platform: autonomous AI agents, live agent assist, conversation intelligence and automated QA
- +Models are trained on a customer's own transcripts rather than generic data, so guidance reflects what actually works on that team's calls
- +Proven at large scale with United Airlines, Alaska Airlines, Hilton, CVS and Intuit among named customers
- +Real-time coaching gives supervisors a measurable path to lifting the performance of the whole agent pool, not just deflecting tickets
- +Strong enterprise footing — roughly $282M raised, a reported $1.6B valuation and 100% QA coverage in place of transcript sampling
Weaknesses
- −No published pricing and no free tier or self-serve trial; AWS Marketplace listings put Agent Assist contracts around $150,000 for 12 months
- −Annual seat-based contracts and enterprise procurement put it out of reach for SMBs and most mid-market teams
- −Implementation is a project, not a signup — expect telephony/CRM integration work and a model-training period before value appears
- −Heavily weighted toward high-volume voice contact centers; a poor fit for teams whose support is mostly async email or in-product chat
Overview
Overview
Cresta is an enterprise contact-center AI company founded in 2017 by Zayd Enam, Sebastian Thrun and Tim Shi, and now led by CEO Ping Wu from Sunnyvale, California. It has raised roughly $282 million, most recently a $125 million Series D in November 2024 co-led by World Innovation Lab and the Qatar Investment Authority, at a reported valuation near $1.6 billion. Sacra estimates the company passed $100 million in annual recurring revenue during 2026.
The platform spans four products: an autonomous AI Agent for voice and digital channels, Agent Assist for real-time guidance to human agents, Conversation Intelligence for analytics across every conversation, and Quality Management for automated QA. The differentiator versus general-purpose support agents is that Cresta trains on a customer’s own transcripts, so its suggestions are grounded in the phrasing and outcomes that already work for that specific team — and that it treats the human agent as a first-class user rather than only a fallback for deflection.
The cost is the obvious trade-off. Cresta publishes no rate card, sells annual seat-based contracts through enterprise procurement, and its AWS Marketplace listings imply six-figure yearly commitments for a single module. Deployment involves telephony and CRM integration plus a training period before the models are useful. For a Fortune 500 contact center this is normal; for anyone smaller, tools like Intercom Fin, Decagon or Sierra will land value far faster and cheaper.
Key Benefits
- One platform, whole funnel: Automation, live agent coaching, analytics and QA share the same conversation data instead of living in four vendors.
- Customer-specific models: Training on your own transcripts produces guidance that matches your product, policies and top performers.
- 100% QA coverage: Automated scoring replaces sampling, so quality trends reflect all conversations rather than a handful.
- Enterprise-scale proof: Airlines, hotel groups, financial services and healthcare brands run it in production.
Use Cases
- Autonomous call handling — Route routine inbound calls to the AI Agent for end-to-end resolution, escalating complex cases to humans.
- Live agent coaching — Surface next-best-action prompts and knowledge articles mid-call to shorten handle time and lift resolution rates.
- Contact center QA — Score every conversation automatically and target coaching at the specific behaviours that correlate with outcomes.
- Revenue and retention plays — Use conversation intelligence to spot save opportunities and upsell moments across the whole call corpus.
Pricing
Checked September 2026 · USD- Seat-based annual subscription, quoted per deployment
- Modules priced separately (AI Agent, Agent Assist, Conversation Intelligence, Quality Management)
- AWS Marketplace listings show ~$150,000/year for Agent Assist for Voice (up to 100,000 calls) or Chat (up to 125,000 chats)
- Implementation, model training and dedicated customer success included