A Y Combinator S21 voice AI agent platform with roughly $125K raised through YC plus a seed round from Global Founders Capital, Soma Capital, and Freeflow Capital. It builds phone-based voice agents through a drag-and-drop Flow Builder with IVR detection and navigation, running on self-developed infrastructure that claims 200–400ms latency end to end and stays under 800ms even when mixing third-party models like GPT or ElevenLabs. In October 2026 it shipped voice rebuilt on Sesame CSM-1B, adding natural “um” and “uh” pauses and reporting customer hang-up rates down more than 60%. Auto-design and self-improvement let teams train agents from call recordings or scripts to mimic strong reps, then evaluate failures and retrain after deployment. Developer access is via REST and GraphQL APIs plus a JavaScript SDK. Pricing is per second: Standard Voices at $0.09/minute and Premium Voices at $0.14/minute, with Voicelab (standalone TTS) at Free (180 minutes), Starter $20/month (500 minutes), and Pro $150/month (5,000 minutes, HIPAA workspace), and Enterprise volume discounts reported as low as $0.04–$0.05/minute all-in. PitchBook records a completed merger/acquisition on 6 May 2026, though the company continues to operate.
**Pros & Cons**:
Pros: Self-developed infrastructure with claimed 200–400ms latency, staying under 800ms even with third-party models, addresses the single biggest failure mode in phone voice agents; Sesame CSM-1B rebuild adds natural disfluencies and reports a 60%+ drop in hang-ups, which is a concrete outcome rather than a tone claim; Flow Builder with IVR detection handles the real-world telephony plumbing that simple chat widget builders skip; Auto-design trains agents from existing call recordings or scripts, reducing cold-start effort; self-improvement evaluates failures post-deployment and retrains, which is closer to a production loop than one-shot configuration; per-second billing aligns cost with actual call volume, and Enterprise pricing reportedly reaches $0.04–$0.05/minute all-in.
Cons: **SOC 2 Type II status is contradictory**—Vogent’s own comparison articles describe it as compliant, while EveryDev, G2, and cAImpare describe it as “audit pending” or “in preparation”, so it should not be stated as certified; **PitchBook records a completed merger/acquisition on 6 May 2026**, and while operations continue, post-acquisition independence and long-term direction need monitoring; **independent review volume is small and concentrated**—Krain shows 4.8/5 from 75 reviews, but G2 shows zero reviews and Trustpilot has no data, and several Krain entries read like pitch feedback rather than usage experience; **per-minute pricing at $0.09–$0.14 scales linearly** with call volume, so high-volume deployments need careful cost modelling against flat-rate competitors; **positioned for phone voice agents**, not website chat widgets, so it is a different channel from most tools in this category.
**Who is it for?**:Clinics, logistics, travel, and customer support teams that handle inbound and outbound phone calls; teams that need IVR navigation and telephony plumbing rather than a website chat widget; organisations that want to train agents from existing call recordings or scripts; enterprises requiring HIPAA workspaces and volume pricing.