The sales call that never gets tired, never needs a coffee break, and remembers every detail from minute one through minute forty: That's the pitch from Origa, an India-based startup betting that the future of enterprise sales automation isn't in quick exchanges, but in conversations that sprawl.
Most voice AI agents—the kind handling customer service queries or capturing basic lead information—flame out after a few minutes of back-and-forth. Origa's founders claim their system can hold its own through sales calls stretching to three-quarters of an hour, juggling objections, pricing negotiations, and qualification criteria for deals in the $50,000 to $5 million range. Without, as they put it, breaking character.
It's an audacious technical challenge. And judging by the company's latest funding round, at least some investors think they're onto something.
A modest raise, but the numbers tell a story
Origa just secured $450,000 in seed funding led by Antler Singapore, pushing total capital raised to around $1 million. Not the sort of headline-grabbing round that dominates tech Twitter, but the traction metrics paint a more intriguing picture.
The startup reports 33 paying customers scattered across India, the UAE, Malaysia, and the United States. Zero churn. Monthly conversation volume has hit 570,000 minutes—roughly 9,500 hours of AI-mediated sales talk. Revenue, according to the company, grew twelvefold over nine months. The platform now handles more than 100,000 concurrent calls, a figure that would have seemed fantastical for voice AI just a few years ago.
For a company founded in 2024 tackling one of conversational AI's thorniest problems—sustaining coherent, context-rich dialogue over extended periods—those numbers hint at early product-market fit. Narrow, perhaps. But real.
The fresh capital will fuel expansion in India and the UAE, plus engineering hires aimed at deepening what the founders describe as the platform's "technical scaffolding." Translation: they want the AI to get better at being human.
Sales calls that actually sound like sales calls
Here's what Origa is not trying to do: replace the chatbot on your bank's website or automate "press one for billing" phone trees. The company positions itself as an "intelligent qualification layer" for high-ticket B2B transactions—the messy, consultative sales cycles where a human sales development rep might spend half an hour probing a prospect's budget, timeline, and pain points before passing them to a closer.
The platform integrates with the usual suspects: Salesforce, HubSpot, Zoho. When the AI decides a prospect is primed and ready, it can execute a live transfer to a human account executive. Use cases span lead qualification, customer re-engagement, and—interestingly—screening interviews, though the company has quietly dialed back that positioning in favor of pure sales automation.
Origa claims its system sustains conversations between five and 45 minutes "naturally," maintaining context across more than 50 conversational turns. That duration significantly outpaces most voice agents designed for transactional customer service or lightweight lead capture.
Industries currently relying on the platform include real estate, financial services, healthcare, and education. In one real estate deployment cited by the company, qualified leads jumped 40% while turnaround time shrunk by a factor of eight compared to manual outreach. Impressive, if it holds at scale.
Under the hood: Training beats prompting
Origa's technical differentiation—at least according to its founders—rests on training depth rather than clever prompt engineering. The platform was fine-tuned on more than 1 million real-world sales conversations, a corpus designed to teach the AI how to navigate the chaotic, non-linear reality of actual selling. Tangents. Mid-call pricing pivots. Objections that surface out of nowhere.
The system claims sub-800-millisecond end-to-end latency, a hallucination rate below 0.1%, and 99.9% accuracy on real-time API integrations that pull CRM data during live calls. It also incorporates compliance guardrails for regulated verticals—PII and PHI protections, post-LLM processing that adds natural pauses and prosody to avoid the robotic cadence that still haunts many voice agents.
Those specs, if accurate, put Origa in the same conversation as platforms like Retell AI (which advertises latency around 600 milliseconds) and Deepgram's Voice Agent API. Retell emphasizes its proprietary turn-taking engine; Deepgram markets a unified voice-to-voice interface with barge-in handling. PolyAI targets enterprise customer service with a hybrid of retrieval and generative models.
The bet Origa's founders are making—Himanshu Geed (IIT Delhi, IIM Calcutta, ex-Uber city launcher), Shubham Garg, and Sunil Jain—is that specialization in extended sales conversations carves out defensible territory in a crowded field. Generalized customer service agents are table stakes now. The real value, they argue, lives in automating the labor-intensive middle of the funnel, where calls run long and context is everything.
Origa operates under the legal entity Anson Careers Private Limited, incorporated in India in 2018—a detail that suggests the founders may have pivoted from an earlier venture, though the company hasn't publicly elaborated on that history.
The bigger question: Platform or feature?
Gartner recently forecast that agentic AI will autonomously resolve 80% of common customer service issues by 2029. Separate industry projections suggest 40% of enterprise applications will embed task-specific AI agents by 2026. But those predictions mostly envision short, transactional exchanges—password resets, order tracking, basic troubleshooting.
Origa is wagering on a different future. One where the AI doesn't just answer questions, but qualifies, persuades, and occasionally endures the silence of a prospect who's mentally running numbers mid-call.
If the early traction metrics hold—and that's always the caveat with startups at this stage—the company has at least demonstrated that a slice of the enterprise sales world will pay for an AI capable of navigating a 40-minute conversation without losing the thread.
The harder question, the one that will determine whether Origa becomes a venture-backable platform or gets absorbed as a feature in someone else's stack, is this: Can 1 million training conversations scale across industries with wildly divergent vocabularies, objection patterns, and deal structures? Or does this remain a point solution for a few high-ticket verticals with predictable sales motions?
Real estate and financial services share some structural similarities—long sales cycles, compliance-heavy environments, high per-deal values. But healthcare and education? Those markets present entirely different conversational dynamics. The company's ability to generalize without sacrificing performance will dictate its ceiling.
For now, Origa has bought itself runway and validation. The next few quarters will clarify whether it's built the future of B2B sales automation—or just a very sophisticated tool for a very specific problem.
Either way, someone's going to be on the phone for a very long time.
