Who are the top conversational AI startups by revenue today?
In our conversational AI market deck, you will find everything you need to understand the market
SUMMARY
Sierra is the top pure-play conversational AI startup by revenue today, at roughly $200 million ARR. Cresta, Kore.ai, Fin and probably Decagon form the next tier around $100 million, while larger mixed businesses such as ElevenLabs and Gupshup complicate any simple leaderboard.
The biggest trap is comparing company-wide revenue with conversational-agent revenue. ElevenLabs has passed $500 million ARR and Intercom is above $400 million, but neither figure represents a pure conversational-agent business in the way Sierra's roughly $200 million does.
The market has already developed a surprisingly clear revenue hierarchy. Sierra sits around $200 million, several established companies cluster near $100 million, Parloa has moved beyond $50 million, and PolyAI and Respond.io occupy the roughly $35 million to $40 million range.
Sierra's lead is more impressive because it was built almost entirely during the generative-AI era. The company went from roughly $100 million to $200 million ARR in about two quarters after it had already reached meaningful enterprise scale.
The $100 million tier is less precise than it looks. Cresta has directly disclosed passing $100 million ARR, Kore.ai has been reported around $110 million, Fin is estimated around $100 million to $110 million, and Decagon's latest roughly $100 million figure remains an outside estimate rather than a fresh company disclosure.
Older conversational-AI companies have not disappeared just because newer agent startups get more attention. Kore.ai and Cresta already have substantial enterprise revenue bases, customer relationships and contact-center distribution that the newer companies still have to build.
Usage-based pricing is making startup revenue figures harder to compare. Sierra, Fin and Decagon can charge for conversations, successful resolutions or completed outcomes, so a sharp rise in customer usage can push annualized revenue higher much faster than traditional seat-based SaaS.
Voice is becoming the most important category boundary. PolyAI helped establish enterprise voice agents, but Sierra, Fin, Decagon, Parloa and especially ElevenLabs are increasingly competing for the same phone calls, customer-service workflows and sales conversations.
The consumer companion market has enormous engagement but weaker monetization than enterprise customer service. A company such as Sierra can justify large contracts against human support costs, while companion products usually depend on relatively small subscriptions and in-app purchases.
The most important competitive change may come from ElevenLabs and Decagon. ElevenLabs is already much larger than Sierra at the company level and says voice agents are driving enterprise growth, while Decagon could already be around the $100 million tier if outside estimates are accurate. For the moment, though, Sierra has the cleanest claim to the pure-play revenue lead.
This market map, featured in our conversational AI market deck, highlights top companies and startups in the conversational AI market
Who are the top conversational AI startups by revenue today?
Sierra is currently the clearest revenue leader among pure-play conversational AI startups, at roughly $200 million in ARR, but ElevenLabs, Fin/Intercom and Gupshup become larger if we use a broader definition of the market.
The ranking gets confusing because those companies sell different things. ElevenLabs has passed $500 million ARR, but voice agents are only part of its audio platform. Intercom is above $400 million ARR, while its Fin AI agent accounts for roughly $100 million. Gupshup generates more than $350 million a year, but management says AI-related products contribute about 25% to 30%. Sierra is different: customer-facing AI agents are essentially the business.
Once we separate company-wide revenue from conversational-agent revenue, the market becomes much clearer. Sierra leads the pure-play specialists. Cresta, Kore.ai, Fin and probably Decagon sit around the $100 million level. Parloa is above $50 million, PolyAI is around $40 million based on its latest public trajectory, and Respond.io has reached $35 million while remaining profitable.
| Company | Latest useful revenue figure | What that figure covers | Where we place it |
|---|---|---|---|
| Sierra | ~$200M ARR | Enterprise AI agents | Pure-play leader |
| Cresta | $100M+ ARR | Human + AI contact-center platform | Top tier |
| Kore.ai | ~$110M ARR | Conversational and enterprise AI | Top tier |
| Decagon | ~$100M annualized, estimated | Customer-service AI agents | Likely top tier |
| Fin / Intercom | $400M+ total; Fin ~$100M | Legacy support software + AI agent | Bigger company, smaller AI business |
| Gupshup | $350M+ revenue; 25–30% AI | Messaging + conversational AI | Bigger company, mixed revenue |
| Parloa | $50M+ ARR | Customer-service AI agents | Fast challenger |
| PolyAI | ~$40M trajectory | Voice AI agents | Voice specialist |
| Respond.io | $35M ARR | Messaging + AI conversations | Profitable challenger |
Why is conversational AI revenue so hard to rank today?
Conversational AI revenue is harder to compare than normal SaaS revenue because companies are calling several different financial metrics “ARR.”
A traditional SaaS company might sign an annual contract for a fixed number of seats. Sierra can instead charge according to conversations or successful outcomes. Fin charges by successful resolution. Decagon offers conversation- and resolution-based pricing. If usage jumps in one quarter, annualizing that period can produce a much larger figure than trailing twelve-month accounting revenue.
Investors have started paying closer attention to this distinction. In recent reporting on AI startup metrics, venture investors described founders using contracted ARR, committed ARR and annualized run-rate revenue in ways that would have been treated much more carefully in the traditional SaaS market.
Decagon shows how large the difference can become. Earlier reporting put the company significantly above $30 million in annualized revenue. Sacra now estimates roughly $100 million based on its more recent growth. That estimate is useful for judging Decagon's likely size, but it does not carry the same weight as Cresta publicly announcing that it crossed $100 million ARR.
We can confidently distinguish a $200 million business from a $35 million one. Deciding whether two companies around $100 million are really five or ten million dollars apart is much shakier.
As this chart shows, and as featured in our conversational AI market deck, search interest in conversational AI has increased sharply
What should actually count as a conversational AI startup?
For this ranking, a conversational AI startup needs to make a large part of its money from software that talks with customers or employees through natural language and completes useful work during those conversations.
That comfortably includes Sierra, Decagon, Parloa, PolyAI, Kore.ai and Cresta. Fin also belongs here even though it grew out of Intercom's older customer-support business.
The definition becomes much less useful if we include every AI product with a chat interface. Glean lets employees ask questions conversationally, but the core product is enterprise search and knowledge. Lovable lets users build software through conversation. Cohere sells models and AI infrastructure. Those businesses compete for some of the same budgets, yet customers are buying something different.
ElevenLabs sits directly on the boundary. Its voice agents increasingly handle sales and customer-support conversations, while the company still makes substantial revenue from speech generation, dubbing, audio models and other products. We therefore include ElevenLabs in the analysis without assigning its entire company ARR to conversational AI.
If you want more recent data on this point, please see our latest conversational AI market report.
Should OpenAI and Anthropic count as conversational AI startups?
OpenAI and Anthropic would crush everyone in this ranking by revenue, but counting them would turn the article into a ranking of foundation-model companies.
ChatGPT and Claude are obviously conversational AI products. OpenAI and Anthropic, however, also make money from APIs, enterprise model access, coding products and other workloads. Their businesses now operate at annualized revenue levels in the tens of billions of dollars, hundreds of times larger than most conversational-AI specialists.
The interesting part is what that scale difference says about Sierra, Fin and Decagon. These companies can build very large businesses without owning the underlying frontier model. An enterprise may use OpenAI or Anthropic underneath the agent while paying the application company for integrations, workflows, evaluation, security and the completed customer interaction.
For this article, OpenAI and Anthropic sit outside the core ranking because conversational AI is only one way their models are monetized.
This chart, included in our conversational AI market deck, shows annual VC investment in conversational AI startups
Is Sierra the biggest pure-play conversational AI startup right now?
Yes. Sierra is currently the strongest candidate for the largest independent pure-play conversational AI startup, with roughly $200 million in ARR.
The speed is unusually important here. Sierra announced $100 million ARR after seven quarters in market. It then recorded its first $50 million quarter and entered its third year above $150 million ARR. By May, co-founder Bret Taylor said the company had added another $100 million in only two quarters, taking the annualized figure to about $200 million.
The customer mix makes that growth harder to dismiss as small-company experimentation. Sierra says one-quarter of its customers generate more than $10 billion in annual revenue and half generate more than $1 billion. Publicly named customers include ADT, Cigna, Deliveroo, Discord, Ramp, Rivian, SiriusXM, Sonos and Wayfair.
These agents also do more than answer basic support questions. Sierra sells systems that process returns, handle account changes, work through insurance interactions and carry out other customer-service tasks tied directly to operating costs.
That is why Sierra currently deserves the pure-play revenue crown. We can debate exactly how outcome-based ARR should be measured, but the gap between roughly $200 million and most specialists below $100 million is too large to be explained by accounting terminology alone.
If you want more recent data on this point, please see our latest conversational AI market report.
How fast is Sierra actually growing?
Sierra's revenue growth is extreme even by today's AI standards: the company doubled from roughly $100 million to $200 million ARR in about two quarters.
That second $100 million is the most telling part. Plenty of AI startups grow quickly from $2 million to $10 million because the starting point is tiny. Sierra accelerated after it had already reached a scale that takes many enterprise-software companies years to build.
Compare the increase with the current size of competitors. Parloa's entire disclosed business is above $50 million ARR. Respond.io is at $35 million. PolyAI's latest public trajectory was around $40 million. Sierra added roughly the combined scale of several established challengers in a matter of months.
Its February company update also showed how the customer profile was changing. Sierra said adoption had jumped particularly quickly among Fortune 20 companies and that one large healthcare customer went live in seven weeks. That short implementation cycle matters because enterprise AI deployments often get stuck between successful demos and real production usage.
The $200 million headline matters less than the second $100 million: it arrived after Sierra had already reached scale.
| Sierra milestone | Approximate ARR | What changed |
|---|---|---|
| First major milestone | $100M | Reached after seven quarters |
| Start of year three | $150M+ | First $50M quarter |
| Latest disclosed run rate | ~$200M | Another $100M added in roughly two quarters |
This chart, included in our conversational AI market deck, breaks down Cognigy's playbook in conversational AI
Is Cresta quietly one of the biggest conversational AI startups?
Yes. Cresta has already crossed $100 million ARR, putting it firmly among the largest independent conversational-AI companies today.
Cresta confirmed the milestone when it announced board changes in April. The company said it had surpassed $100 million in annual recurring revenue while expanding across Fortune 500 customers such as United Airlines, Cox Communications and Marriott.
Its business is broader than Sierra's. Cresta originally became known for software that helped human contact-center employees during conversations through coaching, recommendations and analytics. It has since added autonomous AI agents and now sells one platform covering both humans and AI.
That mix probably makes Cresta's $100 million less directly comparable with Sierra's agent revenue, but it also gives the company something the newer startups lack: an existing position inside large contact centers. Cresta can sell automation into customers already using its human-agent software rather than having to win every deployment from scratch.
Revenue rankings tend to understate Cresta because the company attracts fewer dramatic funding headlines than Sierra or Decagon. The actual numbers put it near the front of the market.
Is Kore.ai still bigger than the newer AI-agent startups?
Kore.ai is still one of the biggest independent conversational-AI businesses, with roughly $110 million in ARR, even though newer names get much more attention.
Kore.ai has been building enterprise conversational systems since well before the generative-AI boom. Recent financial reporting around the company put ARR at roughly $110 million, with customer service accounting for around 70% of the business. That implies roughly $75 million to $80 million tied to customer-service use cases alone.
The older product history explains why comparisons with Sierra or Decagon are imperfect. Kore.ai sells a broader platform spanning customer interactions, employee assistants and enterprise AI applications. Its revenue base was already established before today's autonomous-agent wave.
Still, $110 million is too large to ignore. Kore.ai also indicated that it expected ARR to move toward $200 million during its next growth phase. We treat that as a forecast rather than current revenue, but reaching even part of it would keep Kore.ai among the biggest companies in the category.
The market narrative has moved faster than the revenue table. Kore.ai may feel like an older conversational-AI company, yet financially it remains ahead of most of the startups launched during the ChatGPT era.
This chart, included in our conversational AI market deck, shows annual funding in conversational AI startups
Is Fin bigger than Sierra today?
Fin is currently smaller than Sierra as an AI-agent business, although Fin's parent company is more than twice Sierra's size by total ARR.
Intercom moved past $400 million in company ARR, while Dealroom and Sacra estimates put Fin itself around $100 million to $110 million and growing quickly. That means roughly one-quarter of the company has already shifted toward the AI agent even though Intercom spent more than a decade building traditional support software.
The distinction became even more important after Salesforce agreed to acquire Fin for approximately $3.6 billion. Salesforce specifically highlighted Fin's ability to resolve customer queries across live chat, email, WhatsApp, SMS, phone and Slack, which shows what it is actually buying: the fast-growing agent layer rather than another conventional helpdesk product.
Dealroom's reconstruction of Intercom's revenue is revealing. The older Intercom business appears broadly flat while Fin accounts for essentially all recent growth. Fin has therefore become the economic engine of a company that existed long before generative AI.
Sierra still leads on conversational-agent revenue at roughly $200 million versus Fin around $100 million to $110 million. Intercom only wins the company-wide comparison because it brings roughly $300 million of older support-software revenue with it.
| Metric | Sierra | Fin / Intercom |
|---|---|---|
| Conversational-agent ARR | ~$200M | ~$100M–$110M |
| Total company ARR | ~$200M | $400M+ |
| Large legacy software business | No | Yes |
| Current status | Independent | Agreed acquisition by Salesforce |
If you want more recent data on this point, please see our latest conversational AI market report.
Does Gupshup's $350 million revenue make it the conversational AI leader?
No. Gupshup is a $350 million-plus company, but management says only around 25% to 30% of revenue currently comes from AI-related products.
CEO Beerud Sheth gave that breakdown while discussing how Gupshup is shifting from traditional business messaging toward AI-powered conversations. Applying the company's own percentage to $350 million gives roughly $88 million to $105 million of AI-related revenue.
That puts Gupshup's newer AI business surprisingly close to Fin, Cresta and Kore.ai rather than anywhere near a $350 million conversational-AI lead.
The company still has a distribution advantage that the younger startups would love to own. Gupshup handles messaging across WhatsApp, SMS, RCS, Instagram, Telegram and voice and operates at enormous message volume across more than 100 countries. It can introduce AI into customer relationships that already exist.
The transition also appears to be improving economics. Sheth said AI products helped gross margins rise by three to four percentage points, while internal AI automation cut some support response times by 90% to 95%. More recently he put AI-related revenue at 25% to 30% and said he expects it eventually to exceed half of the business.
So Gupshup belongs near the top of the competitive map. Its $350 million headline simply overstates how much conversational AI revenue it generates today.
This chart, included in our conversational AI market deck, compares the main business model options for conversational AI enterprise platforms
Has Decagon already caught the $100 million conversational AI tier?
Probably, but we rank Decagon's current revenue with lower confidence than Cresta or Sierra because the newest $100 million figure is an outside estimate rather than a fresh company disclosure.
Earlier reporting from The Information said Decagon was generating significantly more than $30 million in annualized revenue after being around $10 million roughly a year earlier. Sacra now estimates that Decagon reached approximately $100 million in annualized revenue by July, up from an estimated $44 million at the end of 2025.
The commercial evidence makes a move toward $100 million plausible. Decagon added more than 100 global enterprise customers during 2025, including Avis Budget Group, Block and Deutsche Telekom, alongside companies such as Duolingo, Notion, Oura, Rippling and Substack. The company charges per conversation or successful resolution, so expanding usage inside a large customer can push revenue up quickly.
There is still an evidence gap. Decagon announced its $250 million Series D and $4.5 billion valuation without publishing a new revenue number. We therefore should not present $100 million with the same confidence as Cresta's publicly announced $100 million milestone.
Our best current ranking puts Decagon around the $100 million group, with an asterisk. Among the major challengers, it is also the company most likely to move sharply when the next hard revenue disclosure arrives.
Can Parloa catch Sierra and the US conversational AI leaders?
Parloa has become Europe's strongest pure-play challenger at more than $50 million ARR, but it still needs to roughly double before reaching the second revenue tier and quadruple to match Sierra.
The company crossed $50 million ARR after reporting 150% net revenue retention, meaning existing customers were increasing their spending fast enough to add substantial growth before new customer wins were counted. Customers include Allianz, Booking.com, HealthEquity, SAP and Swiss Life.
Parloa has also won a multimillion-dollar contract with TP, one of the world's largest customer-experience outsourcing groups. That kind of deployment has more upside than collecting dozens of small chatbot customers because a successful agent can eventually spread across huge call volumes.
Its financing gives Parloa room to push harder in the US. The company raised $350 million after reaching a $3 billion valuation and has been expanding partnerships and its American operation.
The revenue gap is still large today. Parloa has crossed $50 million while Sierra is around $200 million. The interesting part is that 150% net retention gives Parloa a credible way to close some of that gap without needing customer acquisition alone to do all the work.
This chart, featured in our conversational AI market deck, illustrates revenue distribution by customer segment in the conversational AI market
How big is PolyAI now, and is voice becoming the bigger opportunity?
PolyAI appears to be around a $40 million ARR business based on its latest disclosed trajectory, while the broader voice-AI market around it is growing much faster.
PolyAI's UK accounts provide a useful reality check. Statutory revenue rose from roughly $8.9 million to around $15 million in the year ending January 2025. CEO Nikola Mrkšić later said the company expected ARR to double toward roughly $40 million, with US revenue nearly tripling.
That gap between $15 million of historical recognized revenue and roughly $40 million of ARR shows why we keep separating accounting revenue from annualized startup metrics. PolyAI can genuinely be growing toward $40 million while its latest completed financial accounts show far less.
The bigger change is competition. ElevenLabs ended 2025 above $330 million ARR and then officially said it had passed $500 million ARR during the first four months of 2026. The company says enterprise voice-agent deployments across customer support, sales, hiring and marketing are now driving much of that acceleration. We still cannot isolate how much of the $500 million comes specifically from agents, because ElevenLabs also sells speech generation, dubbing and other audio products.
Meanwhile Sierra, Fin, Decagon and Parloa are all moving deeper into voice. Large customers increasingly want one agent that can work through phone calls, chat, email and messaging apps while accessing the same customer data.
PolyAI helped prove that enterprises would pay serious money for AI phone conversations. It now has to compete in a market where almost every major conversational-AI company wants the same calls.
Is Respond.io really big enough to matter in conversational AI?
Yes. Respond.io has reached $35 million ARR with 169% year-over-year growth and a 30% profit margin, making it one of the most financially interesting companies below the $50 million tier.
Those numbers came directly from the company when it raised a $62.5 million Series B in June. The growth rate is striking, but the profit margin may be even more unusual. Many AI-agent startups are spending heavily to grab enterprise customers; Respond.io says it is already profitable while growing triple digits.
Its product grew out of customer messaging rather than autonomous AI. Businesses use Respond.io to manage conversations across WhatsApp and other channels, and AI agents are taking over more of the work inside those conversations.
That resembles Gupshup's evolution on a smaller scale: both companies already owned the communication layer before generative AI made automation much more capable.
At $35 million, Respond.io remains far below Sierra and the $100 million group. But it is already close enough to PolyAI and Parloa that leaving it out of a current conversational-AI revenue ranking would give a distorted picture of the field.
This chart, included in our conversational AI market deck, shows how AI chatbot platform technology has evolved over time
Where does ElevenLabs belong in the conversational AI revenue ranking?
ElevenLabs is now a $500 million-plus ARR company and could eventually become the largest conversational-AI business in this article, but we still cannot attribute that full amount to conversational agents.
This is the biggest freshness change in the ranking. ElevenLabs ended 2025 around $350 million ARR and said in May that it had already surpassed $500 million during the first four months of 2026. Management explicitly said enterprise deployment of voice agents across customer support, sales, hiring and marketing was driving the acceleration.
That is much stronger evidence than we had when ElevenLabs was primarily known for text-to-speech. Deutsche Telekom, Square, Revolut and other large organizations are using its technology in conversational workflows, and the company is investing heavily in ElevenAgents as a dedicated enterprise agent platform.
Still, the $500 million includes revenue from the broader ElevenLabs business: voice generation, dubbing, media tools, models and other audio products. We do not have a reliable public split showing how much belongs to autonomous conversations.
For now, Sierra remains the cleaner answer to “largest conversational AI startup.” ElevenLabs is the bigger AI company and increasingly one of Sierra's most serious voice competitors.
If you want more recent data on this point, please see our latest conversational AI market report.
Why are customer-service AI agents making more money than AI companions?
Customer-service agents currently monetize far better per deployment because companies can compare their price directly with the cost of human support, while consumer AI companions still depend on relatively small subscriptions and in-app purchases.
Sierra charges for useful outcomes. Fin charges around a dollar for successful resolutions at list price. Decagon can charge per conversation or resolution. If an AI agent handles a customer problem that previously required several dollars of human labor, the buyer has an immediate financial reason to increase usage.
Consumer companions have huge engagement but weaker economics. Appfigures estimated that romantic and NSFW AI companion apps generated about $163 million of app-store spending during the first half of 2026 across more than 200 apps. Zeta, the largest in that dataset, generated roughly $33 million.
Character.AI remains one of the best-known consumer products, but recent app intelligence shows a very different monetization model: subscriptions at $4.99 or $9.99 and in-app purchases ranging from less than a dollar to around $100. That can build a substantial consumer business, but it does not produce the same contract sizes as automating millions of enterprise support interactions.
The result is visible in the numbers. One enterprise specialist, Sierra, is already running around $200 million annually. An entire large slice of the companion-app market generated a similar order of magnitude across hundreds of products in six months.
In our conversational AI market deck, we identify pain points entrepreneurs should prioritize
How concentrated is conversational AI revenue today?
Conversational AI revenue is already concentrating around a surprisingly small group of companies, despite the huge number of startups selling AI agents.
Sierra contributes roughly $200 million ARR on its own. Cresta has passed $100 million. Kore.ai is around $110 million. Fin is roughly $100 million to $110 million, and the latest outside estimate puts Decagon around the same level.
Those five businesses alone therefore represent roughly $600 million of annualized conversational or closely related customer-experience revenue before we add Parloa, PolyAI, Respond.io or the AI portions of Gupshup and ElevenLabs.
That concentration tells us more than the raw startup count. Building a convincing customer-service bot has become much easier. Winning deployments at Cigna, United Airlines, Deutsche Telekom or other huge companies remains difficult because the product needs integrations, security, monitoring, evaluation, multilingual performance and enough reliability to touch real customer accounts.
The category still looks crowded when we count logos. Revenue tells a different story: a small group is already pulling away.
| Revenue tier | Companies we can reasonably place there now |
|---|---|
| ~$200M pure-play | Sierra |
| ~$100M tier | Cresta, Kore.ai, Fin, likely Decagon |
| $50M+ | Parloa |
| ~$35M–$40M | PolyAI, Respond.io |
| Larger mixed businesses | ElevenLabs, Gupshup, Intercom |
Which conversational AI startups are growing fastest right now?
Sierra is growing fastest at meaningful scale, while Decagon, Respond.io and Parloa are the challengers most capable of changing the ranking over the next few reporting cycles.
Sierra added roughly $100 million ARR in two quarters after already crossing $100 million. That combination of speed and scale is what separates it from smaller startups posting bigger percentage increases.
Respond.io grew 169% year over year to $35 million ARR. Parloa crossed $50 million while posting 150% net revenue retention. Decagon's last hard public reporting showed annualized revenue rising from roughly $10 million to significantly above $30 million, and the newer Sacra estimate around $100 million suggests the curve may have remained extremely steep.
Cresta offers a slower but more mature comparison. The company said revenue had nearly quadrupled over two years before it crossed $100 million ARR. That is still exceptional growth for a platform already selling into Fortune 500 contact centers.
We have lower confidence in Decagon's exact current position than Sierra's, but much more confidence that it belongs among the companies capable of reshuffling this ranking. With usage-based agent pricing, one or two very large deployments can move annualized revenue surprisingly fast.
This chart, included in our conversational AI market deck, illustrates revenue distribution by region across Europe, Asia, North America, Africa, and South America in the conversational AI market
Who are the top conversational AI startups by revenue today?
Sierra is the top pure-play conversational AI startup by revenue today, with roughly $200 million ARR, while Cresta, Kore.ai, Fin and probably Decagon make up the next group around $100 million.
The broader leaderboard depends on what we count. Intercom exceeds $400 million ARR, but most of that revenue predates Fin. Gupshup is above $350 million annually, with about 25% to 30% coming from AI products. ElevenLabs has now passed $500 million ARR and says voice agents are driving rapid enterprise growth, but its revenue still includes a large audio-generation business.
Among cleaner specialists below $100 million, Parloa has crossed $50 million, PolyAI's latest trajectory is around $40 million and Respond.io is at $35 million. Yellow.ai remains part of the established conversational-AI market, but its latest statutory revenue points to a much flatter business than the fastest-growing companies above.
The ranking has changed surprisingly quickly. A $30 million conversational-AI company would have looked large only a short time ago. These days, several independent specialists are already around or above $100 million, and Sierra has pushed the pure-play benchmark toward $200 million.
Sierra therefore leads the independent conversational-AI specialists by revenue. ElevenLabs could eventually take that title as voice agents become a larger share of its business, and Decagon looks like the fastest-moving threat inside the pure-play group. For now, no other specialist has published evidence strong enough to knock Sierra off the top.
If you want more recent data on this point, please see our latest conversational AI market report.
OUR METHODOLOGY
This analysis ranks conversational AI startups by the revenue that can reasonably be tied to conversational AI rather than simply comparing the largest company-wide numbers. We look at how central conversational AI is to each business, the latest usable revenue figure, what that figure actually measures, how recent it is, and how strong the underlying disclosure is.
We separate pure-play specialists from broader businesses when the distinction changes the answer. Sierra's roughly $200 million ARR is treated as predominantly conversational-agent revenue, while ElevenLabs' $500 million-plus ARR, Intercom's $400 million-plus ARR and Gupshup's $350 million-plus revenue are not automatically assigned in full to conversational AI because those companies also sell substantial non-agent products.
We also distinguish company disclosures from outside estimates. A directly announced milestone such as Cresta passing $100 million ARR carries more weight than an estimated annualized figure such as Decagon's roughly $100 million. When the evidence supports a revenue tier more strongly than an exact position inside that tier, we use the tier rather than forcing a precise ranking.
ARR, annualized revenue, run-rate revenue and recognized accounting revenue are not treated as interchangeable. This is especially important for usage- and outcome-based AI products, where a strong quarter can create an annualized figure that moves much faster than trailing twelve-month revenue.
Forecasts are used only as forward indicators. For example, PolyAI's expected movement toward roughly $40 million ARR and Kore.ai's ambition to move toward $200 million help show direction, but they are not treated as completed current revenue milestones.
We prioritized direct company announcements, executive disclosures and financial reporting, then used high-quality reporting or specialist estimates where a company had not published a current figure. Funding rounds and valuations are used as supporting context rather than substitutes for revenue.
Key sources include Sierra on reaching roughly $200 million ARR, Sierra's original $100 million ARR disclosure, Sierra's year-two update, Cresta on surpassing $100 million ARR, The Information on Kore.ai's revenue, Kore.ai's strategic growth update, The Information on Fin and Intercom, Intercom on making Fin central to the company, Salesforce on its agreement to acquire Fin, and Intercom's explanation of Fin's outcome-based pricing.
Additional sources include ElevenLabs on passing $500 million ARR, Decagon's Series D announcement and enterprise customer growth, Parloa on surpassing $50 million ARR, Parloa's $350 million Series D announcement, Forbes on PolyAI's revenue trajectory, PolyAI's Series D announcement, and Respond.io on reaching $35 million ARR, 169% year-over-year growth and its Series B.
This chart, included in our conversational AI market deck, shows annual VC investment in conversational AI startups
Related blog posts
