What are the most valuable conversational AI startups?

Top startups in the Conversational AI market ranked by valuation

Here is an updated table that ranks the top startups in the Conversational AI market based on their latest reported or estimated valuations.

If you need to dig deeper and get more detailed data, please check our report covering the Conversational AI market.

What are the most valuable Conversational AI startups right now?

Perplexity, ElevenLabs and Sierra currently sit in a different valuation league from the rest of the Conversational AI market.

Our dataset puts Perplexity at roughly $19.5–$21 billion, ElevenLabs at $16.7–$22 billion and Sierra at $15.8 billion. After those three, the ranking drops sharply. Abridge sits around $6.6–$7.4 billion in our current estimate, Glean around $4.8–$6.2 billion, and Decagon at $4.5 billion.

Using the midpoint of each range, the top three alone are worth roughly $55 billion combined. Companies ranked fourth through tenth add about another $30 billion. That means three startups represent close to two-thirds of the value of the top ten.

These leaders barely resemble one another. Perplexity is building a conversational alternative to traditional search. ElevenLabs started in synthetic voice and now spans voice models, audio generation and conversational agents. Sierra builds AI agents for enterprise customer service.

The biggest Conversational AI companies today are no longer competing for exactly the same customer or budget. Conversation has become the interface for several much larger markets.

Rank	Company	Current valuation	Main business
1	Perplexity	$19.5B–$21.0B	Conversational search
2	ElevenLabs	$16.7B–$22.0B	Voice and conversational AI
3	Sierra	$15.8B	Customer-service agents
4	Abridge	$6.6B–$7.4B	Clinical conversations
5	Glean	$4.8B–$6.2B	Enterprise search and agents
6	Decagon	$4.5B	Customer-service agents
7	EliseAI	$3.6B–$3.8B	Housing and healthcare agents
8	Hippocratic AI	$3.5B	Healthcare agents
9	Parloa	$3.0B–$3.4B	Customer-service agents
10	Uniphore	$2.4B–$2.7B	Enterprise conversational AI

Is Conversational AI really producing $20 billion startups now?

Yes. Conversational AI has moved into the $20 billion valuation range, although only a tiny number of companies have made it that far.

Perplexity is the clearest example. TechCrunch reported that the company raised another $200 million at a $20 billion valuation after being valued at $18 billion only a couple of months earlier. Perplexity had been worth about $9 billion at the end of 2024, so its valuation roughly doubled again within a year.

ElevenLabs has moved almost as quickly. The company announced a $500 million Series D at an $11 billion valuation in early 2026 after finishing 2025 with more than $330 million in ARR. Our wider current range also reflects later private-market evidence pointing toward a materially higher value.

Sierra has climbed even faster. The company announced a $10 billion valuation in 2025, then raised another $950 million at a $15.8 billion post-money valuation in 2026.

Still, $20 billion is nowhere near normal for Conversational AI. By the fourth company in our ranking, the valuation is below $7.5 billion. Around the twelfth position, valuations are already closer to $1.5 billion.

The market can clearly create $20 billion companies now. It has created only a few of them.

How many Conversational AI unicorns are there today?

At least 18 Conversational AI companies in our dataset are comfortably worth $1 billion or more, with several more sitting right around the unicorn line.

Eighteen companies have valuation ranges whose lower bound is at least $1 billion. Ada, Character.AI, Hume AI and Cartesia then sit around the boundary, with ranges that cross or touch $1 billion. Cognigy was acquired for just under $1 billion.

That already makes Conversational AI a fairly deep unicorn market. But counting unicorns hides how uneven the group has become.

Cresta, Deepgram, Observe.AI, LangChain, Inflection AI and Invoca are roughly $1–$2 billion companies. Decagon, EliseAI, Hippocratic AI and Parloa are around $3–$4.5 billion. Abridge sits higher again. Then Perplexity, ElevenLabs and Sierra suddenly jump into the mid-teens or around $20 billion.

A $1 billion Conversational AI company is now much closer to the middle of the serious private market than to its top.

Why are Perplexity, ElevenLabs and Sierra worth so much more than everyone else?

Perplexity, ElevenLabs and Sierra have reached much higher valuations because each one is attacking a market far bigger than chatbot software.

Perplexity is trying to change how people search for information. Its valuation moved from roughly $9 billion to around $20 billion while the company expanded beyond consumer search into enterprise products, research tools and its Comet browser. Search is an enormous existing market, so even a relatively small shift in user behavior can support a much bigger outcome than a narrow software category.

ElevenLabs has followed a different path. It began with synthetic speech, but the company now sells voice generation, dubbing, conversational models and enterprise voice agents. When it announced its $11 billion Series D, ElevenLabs also disclosed more than $330 million in ARR for 2025. That gives investors both a large technical platform and a real revenue base to value.

Sierra sells enterprise agents that can handle customer-service tasks rather than simply generate text. The company said it had hundreds of customers only 18 months after launch when it announced its $10 billion round. Its valuation later rose to $15.8 billion.

All three companies have attached conversation to something companies or users already spend heavily on: search, audio and enterprise service operations. That is a much stronger position than selling another generic conversational interface.

Is voice AI becoming the hottest part of Conversational AI?

Voice AI is currently one of the strongest parts of Conversational AI, with valuable companies appearing at almost every layer of the stack.

ElevenLabs is the obvious giant, with a current valuation range that reaches roughly $22 billion in our dataset. But the pattern runs much deeper.

Deepgram raised $130 million at a $1.3 billion valuation and positions itself as real-time voice infrastructure. Retell AI is estimated around $1.5–$1.7 billion. Hume AI reaches roughly $1.2 billion at the top of our range. Cartesia sits around $750 million–$1 billion, PolyAI around $750–$825 million, Sanas around $730–$780 million, Vapi around half a billion dollars and Bland AI around $450 million.

A huge amount of business still happens over the phone. Healthcare providers book appointments by voice. Insurers manage claims through call centers. Restaurants take orders. Banks verify customers. Sales teams qualify leads. Customer-support departments handle millions of calls.

TechCrunch recently described the market in similar terms when covering new voice startups: large companies are already using providers such as ElevenLabs and Deepgram, infrastructure platforms such as Vapi and Retell, and customer-support systems such as Sierra and Decagon.

Voice AI is developing into a full market stack rather than one isolated feature.

Voice AI company	Current valuation	Total funding in dataset
ElevenLabs	$16.7B–$22.0B	$843M
Retell AI	$1.5B–$1.7B	$5M
Deepgram	$1.3B–$1.5B	$234M
Hume AI	$800M–$1.2B	$74M
Cartesia	$750M–$1.0B	$191M
PolyAI	$750M–$825M	$205M
Sanas	$730M–$780M	$125M
Vapi	$480M–$520M	$73M
Bland AI	$450M–$470M	$111M

Are AI customer-service agents becoming more valuable than old-school chatbots?

Yes. Investors are currently paying much more for AI systems that can complete customer-service work than for the older generation of chatbot platforms.

Sierra is worth $15.8 billion. Decagon reached $4.5 billion after tripling its valuation in under six months. Parloa went from a $1 billion valuation to $3 billion in roughly seven months.

The commercial traction behind those increases is also becoming easier to see. Decagon said more than 100 new global enterprise customers joined during its previous fiscal year. Parloa finished 2025 above $50 million in ARR with 150% net revenue retention. Sierra said it already had hundreds of customers when it announced its $10 billion financing.

Older Conversational AI vendors generally sit much lower in our ranking. Kore.ai is estimated around $200–$300 million, boost.ai around $140–$200 million, Avaamo around $40–$80 million and Clinc around $80–$150 million.

The product has changed. A traditional bot might identify an intent, provide an answer and hand the customer to a human when things became complicated. Modern customer-service agents are increasingly expected to query internal systems, understand account context and complete tasks.

That broader job is attracting much larger valuations.

Which Conversational AI startups are increasing their valuations the fastest?

Sierra, Decagon and Parloa have produced some of the sharpest confirmed valuation jumps in Conversational AI.

Sierra provides the biggest example. The company was valued around $1 billion in early 2024, then roughly $4.5 billion later that year. Sierra announced a $10 billion valuation in 2025, followed by a $15.8 billion valuation in its next major financing. That is nearly a sixteenfold increase from its first billion-dollar valuation.

Parloa's change happened over a shorter period. The company announced a $1 billion Series C valuation in 2025 after quadrupling revenue over the previous year. By the end of that year, Parloa had passed $50 million in ARR. Its next $350 million financing valued the company at $3 billion.

Decagon tripled its valuation to $4.5 billion in under six months when it raised $250 million. The company tied that jump to a fiscal year in which it added more than 100 enterprise customers.

ElevenLabs also belongs in this discussion. Its $11 billion Series D valuation was more than three times its valuation one year earlier, according to the company.

These increases are unusually fast even for successful software startups. Investors are repricing the small group of companies that appear to be turning AI agents into large commercial businesses very quickly.

Is healthcare becoming one of the biggest Conversational AI markets?

Healthcare now has one of the deepest groups of highly valued Conversational AI startups.

Abridge leads the group. Its Series E valued the company at $5.3 billion, almost double its previous valuation only a few months earlier. Our current estimate is higher because we also consider more recent private-market and comparable evidence.

Hippocratic AI raised $126 million at a $3.5 billion valuation and said the financing brought its total capital raised to $404 million. The company focuses on safety-oriented healthcare agents that interact with patients.

Below those two sit several substantial companies. Suki is estimated around $500–$650 million, Infinitus around $550–$700 million, Syllable around $350–$475 million, Corti around $220–$320 million, DeepScribe around $180–$280 million and Hyro around $170–$200 million.

Healthcare fits conversational AI unusually well because so much expensive work begins with speech. A doctor talks with a patient. A patient calls to schedule a visit. A provider calls an insurer. A nurse follows up after treatment. Each conversation produces documentation or another workflow afterward.

Abridge illustrates how far that can go. The company has expanded beyond basic note-taking into inpatient care, orders and other clinical workflows integrated with systems such as Epic.

Healthcare is now large enough inside this ranking to be treated as one of the main Conversational AI categories rather than a small vertical niche.

Why is Abridge worth more than almost every other healthcare AI startup?

Abridge has reached a multi-billion-dollar valuation because it sits directly inside a painful, high-frequency hospital workflow and is expanding from documentation into the work that follows the conversation.

The company initially became known for turning doctor-patient conversations into clinical notes. That sounds narrow until we look at how often clinicians have to do it and how expensive their time is.

Abridge then started pushing beyond the note itself. The company has announced inpatient integrations, outpatient orders and deeper connections with Epic. Its own description of the strategy is revealing: clinical conversations can feed several parts of a health system rather than ending as one generated document.

The valuation followed that broader role. Abridge reached a $5.3 billion valuation in its 2025 Series E after being valued at $2.75 billion only months earlier.

That trajectory puts Abridge well above most healthcare Conversational AI peers. Hippocratic AI is worth $3.5 billion, while Suki, Infinitus, Syllable and DeepScribe remain below $1 billion in our dataset.

Abridge is currently being valued less like a medical transcription tool and more like a platform built around one of healthcare's richest sources of workflow data: the clinical conversation.

Does enterprise search really count as Conversational AI now?

Yes. Enterprise search increasingly belongs in Conversational AI because products such as Glean and Moveworks now let employees ask questions, reason over company information and trigger work through the same interface.

Glean originally became known for finding information scattered across workplace applications. It then added assistants and agents.

The business has grown rapidly alongside that product shift. Glean announced $100 million in ARR in early 2025, doubled to $200 million nine months later, and then reached $300 million only six months after that. The company also said its Fortune 500 customer count had nearly doubled year over year when it announced the latest milestone.

Glean's latest disclosed primary funding round valued it at $7.2 billion. Our current range is more conservative because the ranking aims to estimate current value rather than automatically keep every company's last financing price forever.

Moveworks followed a similar route from employee assistant to enterprise search and agents. ServiceNow completed its acquisition of Moveworks, and its latest regulatory filings put final purchase consideration at about $2.4 billion.

In practice, the boundary is fairly simple. When natural-language conversation becomes the main way an employee searches company knowledge and asks software to do something, enterprise search overlaps directly with Conversational AI.

Are Conversational AI apps worth more than the companies building the infrastructure underneath them?

Usually. The highest valuations currently go to companies that own a customer workflow, although ElevenLabs shows that infrastructure companies can break through when they move closer to the end user.

Deepgram provides real-time voice APIs and raised at a $1.3 billion valuation. LangChain is around $1.3 billion in our dataset. Braintrust, which focuses on AI evaluation and observability, is worth about $800 million. Cartesia, a real-time voice-model company, sits around $750 million–$1 billion.

Those are large businesses, but application companies can now reach several times those values. Sierra is worth $15.8 billion, Abridge is around $7 billion in our estimate, Decagon is $4.5 billion, Hippocratic AI $3.5 billion and Parloa at least $3 billion.

ElevenLabs is the main exception because it increasingly spans both layers. It still builds core audio and voice technology, but it also sells ElevenAgents and other products directly into enterprise workflows.

The valuation gap is fairly intuitive. Infrastructure providers earn money from enabling other companies. Application companies can potentially capture more of the final budget because they sell the finished outcome to the customer.

Infrastructure can still be a very strong business. Owning the workflow is simply producing the biggest valuation upside right now.

Does raising more money actually lead to a higher Conversational AI valuation?

No. Some Conversational AI companies have raised enormous amounts of money without retaining equally large valuations, while a few much leaner companies have created far more value per dollar raised.

Inflection AI is the clearest extreme in one direction. The company raised roughly $1.5 billion before much of its original team and strategy moved toward Microsoft. Our current proxy value is around $1–$1.4 billion.

ASAPP has reportedly raised more than $500 million, yet our estimate is only about $180–$300 million. Gupshup has raised roughly $544 million and sits around $450–$650 million in the dataset.

At the other end, Retell AI had raised only about $4.5 million when TechCrunch profiled it in 2024, very close to the roughly $5 million figure in our dataset. Its current estimated valuation is far higher.

Deepgram raised $130 million in its latest round and reached a $1.3 billion valuation. ElevenLabs has raised $843 million in our dataset against a valuation reaching well into the teens of billions. Sierra has raised far more capital, but its $15.8 billion valuation still stands at roughly ten times total reported funding.

Funding measures how much capital investors have put into a company. The current valuation reflects what they think the entire company is worth. The two can diverge dramatically.

Company	Approx. current valuation	Funding in dataset	Approx. valuation / funding
Retell AI	$1.6B midpoint	$5M	~320x
ElevenLabs	$19.35B midpoint	$843M	~23x
Sierra	$15.8B	~$1.6B	~10x
Decagon	$4.5B	~$481M	~9x
Hippocratic AI	$3.5B	$404M	~9x
Parloa	$3.2B midpoint	$562M	~6x
Deepgram	$1.4B midpoint	$234M	~6x
Inflection AI	$1.2B midpoint	$1.53B	<1x

Are newer AI-agent startups overtaking the old Conversational AI companies?

Yes. The current valuation ranking shows a clear generational change, with several startups founded during the generative-AI boom already worth much more than companies that spent years building the previous generation of conversational software.

Sierra was founded in 2023 and is already worth $15.8 billion. Decagon was also founded in 2023 and reached $4.5 billion. Hippocratic AI, another 2023 company, is worth $3.5 billion.

Meanwhile, several earlier Conversational AI companies remain in the hundreds of millions. Kore.ai is around $200–$300 million in our dataset, boost.ai around $140–$200 million, Clinc around $80–$150 million and Avaamo around $40–$80 million.

Modern language models changed what a conversational product can do. Older systems often depended much more heavily on intent classification, predefined dialogue paths and manually designed flows. The newer companies can start with much more capable foundation models and spend their engineering effort on integrations, reliability, workflow execution and enterprise deployment.

The older vendors can adapt, and many are already doing so. But the valuation ranking shows that the new generation did not need years to catch up financially.

Is enterprise Conversational AI winning over consumer AI?

Outside Perplexity, the billion-dollar Conversational AI market is overwhelmingly enterprise-focused today.

Character.AI is the most useful consumer comparison. The company became one of the breakout consumer AI products of the early generative-AI wave, yet our current proxy valuation is around $900 million–$1.1 billion after Google's licensing agreement and the departure of key founders.

Most of the other highly valued companies sell to businesses or healthcare organizations. Sierra, Abridge, Glean, Decagon, EliseAI, Hippocratic AI, Parloa, Uniphore, Cresta, Deepgram and Observe.AI all make most sense through an enterprise lens.

Perplexity is the important exception. It has a major consumer product, but the company is also expanding into enterprise search, research products and other paid tools. Its roughly $20 billion valuation therefore reflects a much broader opportunity than consumer chat alone.

The economics help explain the gap. An enterprise can justify paying a lot for software that reduces contact-center costs, saves doctors hours of documentation, helps employees find information or automates business processes. Consumer conversational products need huge audiences before relatively small per-user spending produces the same revenue.

For now, enterprise workflows are where most of the high valuations are clustering.

Are big acquisitions proving that Conversational AI companies have real strategic value?

Recent acquisitions show that large software companies will pay serious money for Conversational AI assets, although the exit values vary enormously.

ServiceNow agreed to acquire Moveworks in a deal originally announced at $2.85 billion. After the transaction closed, ServiceNow's filings recorded approximately $2.4 billion of final purchase consideration.

NiCE acquired Cognigy, one of the better-known enterprise Conversational AI platforms, in a transaction valued at about $955 million. NiCE explicitly described Cognigy as an enterprise-grade conversational and agentic AI company when announcing the deal.

Lower down the ranking, acquisitions provide useful reference points for companies such as Lakera, Cogito, Haptik, Espressive and Loris. Some values are directly disclosed while others remain estimates, so we do not treat all of them with the same confidence.

The spread itself is useful. A successful company in Conversational AI can sell for tens of millions, hundreds of millions or multiple billions depending on its growth, technology, customers and strategic importance to the buyer.

That gives us much better evidence than assuming every AI acquisition automatically validates a unicorn valuation.

Company	Transaction value used in dataset	Transaction context
Moveworks	~$2.4B final consideration	Acquired by ServiceNow
Cognigy	~$955M	Acquired by NiCE
Lakera	$202M	Acquisition value
Haptik	$90M–$110M	Acquisition value
Espressive	$80M–$130M	Acquisition estimate
Loris	$50M–$70M	Acquisition estimate
Cogito	$50M	Acquisition value
Wysdom	$25M–$55M	Acquisition estimate

What types of Conversational AI startups get the highest valuations?

The most valuable Conversational AI startups currently control an important workflow, distribution point or technical bottleneck rather than simply offering a good chat experience.

Perplexity owns a direct relationship with users searching for information. ElevenLabs owns valuable voice technology and increasingly the applications built on top of it. Sierra and Decagon sit directly inside customer service. Abridge sits inside clinical workflows. Glean connects employees with internal company knowledge and business systems.

The second tier tells the same story. Hippocratic AI focuses on healthcare interactions. Parloa manages customer-service agents. Deepgram supplies voice infrastructure. Retell gives developers infrastructure for automated phone calls.

Generic conversational platforms are much more common farther down the ranking.

Generating fluent language is widely available now. Companies become harder to replace when they also own proprietary context, enterprise integrations, customer distribution, regulated workflows, specialized models or the systems required to complete an action.

Is Conversational AI basically becoming an AI-agent market?

Increasingly, yes. The companies attracting the largest valuations treat conversation as the way users talk to an agent that can actually do work.

Sierra sells customer-service agents. Decagon describes its product as an AI concierge. Parloa built its platform around AI agents for customer experience. Glean has expanded into enterprise agents. ElevenLabs is investing heavily in voice agents through ElevenAgents.

Even companies that existed before the current agent boom have changed their positioning and products. Uniphore has expanded toward broader business AI. Kore.ai now emphasizes agentic AI. Druid describes itself as an enterprise agentic-AI platform.

From the customer's point of view, the difference is straightforward. Asking “Where is my order?” and receiving a generated answer is useful. Asking the same question and having the software check the order, identify the problem, reschedule the delivery and update the account is far more valuable.

Conversation remains the interface. The money is moving toward the software capable of doing something after the conversation starts.

How concentrated is the Conversational AI startup market?

Conversational AI valuations are extremely concentrated at the top, with three companies accounting for more than half the implied value of the top 25 in our dataset.

Using midpoint valuations, the top 25 companies are worth roughly $104 billion combined. The top ten represent about $85 billion of that amount, or a little over 80%.

Perplexity, ElevenLabs and Sierra alone contribute roughly $55 billion. That is more than half of the top-25 total.

The drop becomes clearer as we move through the ranking. Around rank ten, companies are worth about $2.5 billion. Around rank twenty, they are roughly $1 billion. By rank fifty, the typical range is closer to a few hundred million dollars.

This is a large market with a very long tail, rather than one where dozens of similarly sized leaders share the value.

Group	Approx. combined midpoint value	Share of top-25 value
Top 3	~$55B	~53%
Top 5	~$67B	~64%
Top 10	~$85B	~82%
Top 25	~$104B	100%
Companies around rank 10	~$2B–$3B each	—
Companies around rank 20	~$1B each	—
Companies around rank 50	~$250M–$400M each	—

Are today's Conversational AI valuations mostly hype?

No. Several of the biggest Conversational AI valuations now have substantial revenue or customer growth behind them, but the highest prices still assume years of exceptional execution.

ElevenLabs offers one of the clearest examples. The company finished 2025 with more than $330 million in ARR before raising at an $11 billion valuation. That is a real business at meaningful scale.

Glean has produced an even clearer growth sequence. It reached $100 million in ARR, doubled to $200 million nine months later, and then announced $300 million roughly six months after that. A company growing at that speed deserves to be analyzed differently from an AI startup with no disclosed commercial traction.

Parloa passed $50 million in ARR after quadrupling revenue around its earlier funding period and reported 150% net revenue retention. Decagon said it added more than 100 new enterprise customers over its previous fiscal year.

Still, the valuation increases leave little room for ordinary growth. Sierra's rise from roughly $1 billion to $15.8 billion in a little over two years assumes a very large future company. Parloa's move from $1 billion to $3 billion in roughly seven months does the same. Decagon tripled in less than six months.

The market is supported by more real revenue than the word “hype” suggests. The aggressive part is how much future growth is already embedded in the biggest valuations.

What does the Conversational AI valuation ranking tell us about who is actually winning?

The companies winning Conversational AI today are the ones turning a conversation into access to a much larger economic activity.

Perplexity connects conversation with search and information discovery. ElevenLabs connects it with voice, audio and enterprise communications. Sierra, Decagon and Parloa connect it with customer-service work. Abridge connects it with clinical documentation and healthcare workflows. Glean connects it with enterprise knowledge and internal actions.

The companies farther down the list reinforce the pattern. Deepgram and Retell provide infrastructure for voice interactions. Hippocratic AI handles healthcare conversations. EliseAI focuses on operational workflows in housing and healthcare.

There is much less evidence that chat itself is the valuable asset. Fluent conversation has become easier to build as foundation models improve.

The companies capturing the most value own what surrounds that conversation: the user relationship, proprietary context, workflow integrations, specialized infrastructure or the ability to complete the next action.

So how big and important is the Conversational AI startup market right now?

Conversational AI has become a major startup market, but most of its value now comes from companies that have moved well beyond the old chatbot model.

Our ranking contains roughly 20 companies at or around unicorn status, several $3–$7 billion businesses and three companies in the roughly $15–$20 billion tier. Using midpoint values, the top 25 alone represent around $104 billion of implied enterprise value.

The strongest companies are also spread across several markets. Perplexity is increasingly a search company. ElevenLabs spans voice infrastructure and applications. Sierra and Decagon automate customer service. Abridge turns clinical conversations into healthcare workflows. Glean connects natural-language interfaces with enterprise knowledge and agents.

That fragmentation shows what Conversational AI has become. It is less a standalone software category these days and more an interface layer spreading across search, healthcare, customer service, enterprise software and voice.

Our final judgment is straightforward: Conversational AI is currently much bigger and more commercially important than the old chatbot market ever became. The startups creating the most value are the ones using conversation as the entry point to a valuable workflow, rather than treating conversation as the whole product.