Isaiah Granet heard the same answer from roughly 180 investors: phone calls were finished, so Bland AI was building for the wrong market. The $40 million Series B says the opposite, and so does the money now rushing into voice AI.
There is a particular kind of investor rejection that stings worse than a simple no. It's the one that says your entire market doesn't exist. Isaiah Granet, co-founder and CEO of Bland AI, heard that verdict roughly 180 times before anyone wrote a check. The pitch was straightforward: let businesses automate phone calls using AI voices that could handle real conversations at scale. The response, again and again, was that voice was a dead medium. Text would take over. Phone calls were on their way out. Don't bother.
Granet and his co-founder kept building anyway. Bland AI went from pre-seed to a $40 million Series B in ten months, bringing total funding to $65 million. Emergence Capital led the Series B, with continued participation from Scale Venture Partners and Y Combinator. That is a blunt correction from the market. The category investors dismissed as obsolete is now pulling serious capital because enterprise phone work never disappeared, it just became expensive enough for software to attack.
Bland's pitch is narrow by design. It builds AI phone agents for enterprise customers: inbound and outbound call handling, appointment scheduling, customer service calls and the old call-center workflows that still sit inside healthcare, insurance, financial services and logistics. The platform runs its own speech and reasoning models rather than simply routing everything through a third-party provider, and it offers self-hosted deployment for organizations with stricter data requirements. You can see why that matters. A healthcare company sending patient calls through a loose stack of outside AI vendors is asking for a compliance headache. A bank that can run the system on its own infrastructure gets a cleaner conversation.
Granet has described the rejection period as a specific investor misread: because text-based interfaces were improving, voice would shrink away. That sounded plausible if you were thinking like a consumer who ignores unknown numbers and answers friends in messages. It was weaker if you had spent any time looking at how businesses actually operate. Customers still call clinics, insurers, banks, airlines and repair companies when the problem is urgent, confusing, or just too annoying to type out.
The real question was never whether phone calls would vanish. It was whether a machine could handle enough of them without making customers furious.
Bland is betting the answer is yes. Its published pricing starts at $0.14 per minute for connected calls, with enterprise deals negotiated directly. That model gets sharper as volume rises. A company handling a million customer calls a month has a very different spreadsheet from one handling ten thousand, and Bland is clearly built for the former. Small teams may test voice AI because it feels new. Large operators buy it because hold times, staffing costs and missed calls show up in the numbers every week.
The Bet Looks Less Strange Now
The voice AI funding market has moved hard since those early rejections. Vapi, a direct competitor, closed a $50 million Series B in May 2026 led by Peak XV, with participation from Kleiner Perkins, Bessemer Venture Partners and Microsoft's M12 venture fund. According to TechCrunch, Vapi said it had processed more than one billion calls and increased enterprise annual recurring revenue tenfold before the raise, at a valuation of roughly $500 million. Amazon Ring chose Vapi's platform over forty competing products, TechCrunch reported.
ElevenLabs sits next to the same boom rather than directly inside Bland's call-center lane, but it points in the same direction. Voice is no longer being treated as a side feature for demos and voice notes. Investors are funding the infrastructure, the agents and the audio models around it. According to data from AssemblyAI, voice AI venture investment jumped from roughly $315 million in 2022 to $2.1 billion in 2024. The category that 180 investors told Granet was dying attracted more than two billion dollars in a single year.
Frankly, the 180 rejections story is useful because it isn't just another founder perseverance tale. Perseverance is cheap to praise after the round closes. The more interesting point is that the investors who passed were pattern-matching from a real consumer trend and applying it to the wrong market. People may prefer messages for casual communication. Enterprises still have phone lines because customers use them, regulators recognize them, legacy systems depend on them and not every problem fits neatly into a chat window.
You see this mistake often when investors treat their own behavior as the market. A venture partner in San Francisco may avoid calls all week. That doesn't tell you much about a patient trying to reschedule an appointment, a driver checking an insurance claim, or a customer whose card stopped working at the worst possible moment. Those are not abstract use cases. They are the dull, repeatable calls that make up a large part of service operations.
Bland still has hard work ahead. Voice agents have to handle accents, interruptions, angry customers, bad audio, compliance scripts and the strange way people change their mind halfway through a sentence. A bad chatbot wastes a minute. A bad phone agent can make a customer hang up and call a competitor. That risk is exactly why Bland's self-hosted option and proprietary model stack matter. Enterprise buyers don't just want a voice that sounds human. They want control, logs, privacy and predictable behavior when the call stops following the script.
So the lesson here is not that investors are foolish and founders are always right. That's too easy. The lesson is that a market can look dead from the outside while still being painfully alive inside the companies paying for it. Bland AI found one of those markets in the ordinary business phone call. The investors who passed were betting on how people like to communicate. Granet was betting on how companies still have to operate.
Also read: Bond markets are absorbing $570 billion in AI debt and nobody knows if the revenue will arrive • Databricks is closing in on Snowflake and the numbers make that case plainly • SpaceX pays $60 billion in stock for Cursor, betting the AI coding toolchain is worth owning outright
