HomeArticle

The list of the world's 20 most profitable AI Agent companies is out, with the top one raking in up to $500 million. Two trends are worth paying attention to.

乌鸦智能说2025-07-29 19:57
Income is becoming a new threshold for measuring competitiveness.

Recently, data institution CB Insights released a highly - regarded list: "The top 20 AI Agent startups with the highest global revenues."

This list is different from traditional rankings that rely on financing amounts or valuations. Instead, it ranks companies directly by their real revenues. It can be said that it is the most direct window to observe the commercial implementation capabilities of the AI Agent track.

There are also many well - known names among the listed companies: from Cursor, a programming assistant for developers, to Harvey, which serves the legal teams of Fortune 500 companies; from Clay, which automates sales lead generation, to Fyxer, a personal AI assistant embedded in email. These companies not only represent the most cutting - edge exploration directions of AI Agent commercialization but also have clear monetization paths in their respective vertical scenarios.

Through this list, we can see two clear trends:

First, AI Agents are evolving from tools to "digital employees". They are no longer just assistants that answer questions but intelligent individuals capable of independently completing tasks and even being responsible for the results. They are starting to truly take over core business processes such as sales, legal affairs, customer service, and coding.

Second, revenue is becoming a new threshold for measuring the competitiveness of AI startups. After the era of "storytelling", the companies that can enter the revenue - generating stage faster will have more say.

Next, let's take a look at which 20 AI Agent companies have rushed to the forefront of global revenues with their real commercial implementation capabilities.

01 Cursor, AI Programming Agent (ARR: $500 million)

Cursor develops an AI - based "vibe coding" programming editor, which is an IDE integrated with an AI assistant. It can perform functions such as auto - completion, multi - line intelligent refactoring, code completion, and code library query. It can also detect logical and security vulnerabilities introduced by AI or humans in real - time through Bugbot.

Currently, there are more than 360,000 paying users, and hundreds of millions of lines of code are generated daily. Its customers include Stripe, OpenAI, Spotify, Midjourney, etc.

In May 2025, Anysphere just completed a $900 million financing round, and its valuation soared more than three times compared to the beginning of the year to approximately $9 billion. As of now, Anysphere's ARR has reached $500 million, and each employee generates $3.205 million in revenue.

02 Glean, Enterprise Search Agent (ARR: $100 million)

Headquartered in Palo Alto, California, Glean positions itself as a "Work AI platform". Its core business is enterprise - level work search + AI agent functions. Employees can interact with the system through natural language to retrieve information from multiple SaaS applications, automatically execute work processes, generate content, etc.

Currently, the Glean Agents platform has processed more than 100 million agent operations and is widely used in the optimization of multiple internal company processes, human collaboration, and knowledge management.

In June this year, Glean received $150 million in financing, and its valuation is $7.2 billion. As of now, Glean's ARR has reached $100 million, and each employee generates $102,000 in revenue.

03 Mercor, Recruitment Agent (ARR: $100 million)

Mercor is an AI - driven recruitment platform aiming to simplify the recruitment process through AI technology and help companies efficiently match suitable candidates. The platform uses advanced AI models to automatically screen resumes, conduct interview evaluations, and match candidates with company needs. (For details, see: "Disrupting Recruitment with AI! Three 00s dropouts built a $2 billion unicorn in 2 years")

Currently, the company's ARR is $100 million, and each employee generates $4.545 million in revenue.

04 Replit, AI Programming Agent for Developing Apps with Natural Language (ARR: $100 million)

Replit is an Agent that can develop apps and websites based on natural language. It can not only write code but also handle all matters from setting up the environment to deploying applications.

Replit was born out of a core pain point: the high entry threshold for programming. The traditional software development process is cumbersome and complex. Developers need to configure an integrated development environment (IDE), install the runtime environments of various programming languages, manage dependency packages, and understand how to deploy programs to the server. This series of preparatory work often discourages many people interested in programming.

The founders of Replit acutely captured this pain point and put forward a grand vision: to make software development interesting and accessible. To achieve this goal, Replit is committed to building a one - stop development platform that integrates all the tools required for programming in a browser - based environment. This means that users don't need to make any local installations or configurations. They can start coding immediately by opening a browser, selecting the required programming language or template.

This "zero - configuration" development experience has greatly lowered the programming threshold, making Replit one of the most user - friendly integrated development environments on the Internet.

According to previous news, Replit is in talks with investors for a new round of financing, and its valuation is expected to nearly triple to $3 billion. Currently, the company's ARR is $100 million, and each employee generates $637,000 in revenue. It is worth mentioning that Replit only took 6 months to increase its ARR from $10 million to $100 million.

05 Lovable, The Fastest - Growing Agent in History (ARR: $100 million)

Lovable is an AI startup headquartered in Stockholm, Sweden. It is committed to enabling users to build complete web applications without coding through natural language prompts. Users only need to describe their ideas, and Lovable can generate applications with a front - end interface, back - end logic, and database support in a few seconds. (For details, see: "The Fastest - Growing AI Company in Europe! Lovable Achieved $120 million in Revenue from Scratch in Just 3 Months")

As one of the fastest - growing companies in the AI field in terms of commercialization, Lovable took only 8 months to reach an ARR of $100 million. The per - capita revenue is as high as $2.222 million, and its latest valuation is $1.8 billion.

06 Crescendo, AI Customer Service Agent (ARR: $91 million)

In the field of AI customer service, Crescendo stands out with its human - machine collaboration solution. It packages the AI and human customer service solutions into an integrated solution, smoothly integrating the service processes of AI and human teams.

Specifically, through CX (customer experience) - type AI chatbots, Crescendo can handle 90% of routine consultations and answer independently by integrating internal company data (such as product catalogs and policy documents).

When encountering complex disputes, real - life experts can quickly intervene with the help of the conversation context sorted out by AI. They can even retrieve the user's return and exchange records and shopping history from three months ago to achieve precise services.

In terms of business models, Crescendo has also abandoned the traditional customer service billing model based on the number of people or hours and adopted a result - oriented pricing model. With its innovative model and payment method, Crescendo has achieved an ARR of $91 million, and its latest valuation is $500 million.

07 Harvey, AI Legal Agent (ARR: $75 million)

Harvey's legal AI assistant, Harvey Assistant, can automatically perform core tasks such as legal research, document drafting, contract analysis, and due diligence. It is built on a large - language model, capable of integrating court documents, providing legal insights with citations, and assisting in proofreading drafts, translation, and filling in contract fields. It is widely used in law firms and corporate legal processes.

In June this year, Harvey AI received $300 million in Series E financing, bringing its valuation to $5 billion. Currently, the company's ARR is $75 million, and each employee generates $133,000 in revenue.

08 StackBlitz, AI Programming Agent (ARR: $40 million)

StackBlitz is a browser - based integrated development environment (IDE) that allows developers to build, run, and deploy full - stack web applications directly from their web browsers.

StackBlitz keenly observed that few developers build web applications within the browser. To address this issue, they launched WebContainers technology, enabling web - based application building.

This means that StackBlitz has opened up another channel for application development. From now on, software development is no longer the exclusive domain of traditional developers. Anyone can create their own software without learning programming and code.

In January 2025, the AI programming company StackBlitz completed a new round of financing, with the financing amount exceeding $100 million. At the same time, the company announced that the number of registered users had exceeded 2 million. Currently, the company's ARR is $40 million, and each employee generates $1.143 million in revenue.

09 Clay, Sales Agent for Automatically Mining Sales Leads (ARR: $30 million)

Clay is a pre - sales customer lead generation and management platform. Through its strong AI capabilities and the integration of data from more than 100 external sources (including public databases and social media), it helps companies optimize their marketing workflows and provides accurate insights for sales and marketing teams.

Clay operates as a SaaS platform, providing API interfaces and integrating various data tools, and automatically generating spreadsheets. From finding customers, contacting customers to retaining customers, Clay drafts personalized promotional emails for each customer to arouse their interest and resonance.

In the latest round of financing, Clay's valuation reached $3 billion. Currently, the company's ARR is $30 million, and each employee generates $1 million in revenue.

10 Torq, AI Security Agent (ARR: $20 million)

Torq is an AI - driven security automation platform for security operation teams, focusing on automating the entire SOC process through no - code/low - code agents.

Its core products, the HyperSOC and Hyperautomation platforms, combine multi - step workflows, RAG models, and multi - agent architectures to achieve closed - loop automation from alert detection, investigation to response.

Torq integrates the company's existing security infrastructure (including SIEM, MDR, CASB, IAM, endpoint detection, etc.) and supports natural - language - triggered and executed processes. The platform has more than 150+ enterprise - level customers. Through automatic case management and preset response measures, it has significantly shortened the response time (the investigation efficiency has been increased to 3 - 5 times the original, and the response time has been saved by up to 90%).

Currently, the company's ARR is $20 million, and the per - capita revenue is $83,000.

11 Sierra, AI Customer Service Agent (ARR: $20 million)

Sierra is a customer service software that can handle tasks such as customer consultations and voice calls, providing a personalized and brand - consistent conversation experience. It focuses on building a conversational AI platform for companies to improve the efficiency and quality of customer service.

Sierra's technical architecture uses a multi - model combination, including the large - language models of OpenAI, Anthropic, and Meta, to ensure the accuracy and reliability of information. In addition, Sierra has introduced a "supervisory agent" mechanism to reduce AI "hallucination" through real - time monitoring and quality control and ensure the credibility of customer interactions.

As of October 2024, Sierra had completed $175 million in financing, and its valuation reached $4.5 billion. Its customers include well - known brands such as WeightWatchers, SiriusXM, and Sonos. Currently, the company's ARR is $20 million, and each employee generates $85,000 in revenue.

12 Sana, Enterprise - Level AI Assistant (ARR: $20 million)

Sana is an enterprise - level AI workflow and knowledge automation platform, focusing on helping organizations build their own AI assistants (Agents) to achieve task automation and intelligent information retrieval. Through a no - code approach, companies can easily create cross - tool, multi - step workflow agents, covering scenarios such as onboarding, customer support, contract review, and RFP writing.

Sana's core ability lies in the integration and modeling of the company's internal structured and unstructured data. Combining large - language models and RAG (Retrieval - Augmented Generation) technology, each AI assistant can not only "find" information but also "do" tasks. The platform itself supports knowledge - base access, connections to multiple systems such as Slack/Notion/CRM, and has built - in collaboration tools, suitable for deployment and expansion by teams of different sizes.

Currently, Sana serves multiple multinational customers such as Electrolux, Merck, and Hinge Health, helping them achieve more than 80% efficiency improvement in multiple business processes.

In the latest round of financing, Sana received a $55 million investment, and its valuation reached $500 million. Currently, the company's latest ARR is $20 million, and the per - capita revenue is $53,000.

13 Nabla, Medical AI Assistant (ARR: $16 million)

Nabla is an AI assistant platform focusing on the medical field, aiming to provide Agent solutions for clinicians from document generation to complex clinical workflow automation.

Its core capabilities include real - time voice transcription, automatic generation of SOAP - style medical records, ICD - 10 coding suggestions, extraction of patient consultation summaries, etc., and it is widely used in primary care, chronic disease management, and inpatient care scenarios.

Nabla's system can be seamlessly integrated into mainstream electronic medical record systems (such as Epic, Cerner, athenahealth). Doctors can get real - time AI assistance in the background without changing their original processes. The "Adaptive Agentic Platform" provided by the platform supports context awareness, multi - model deployment, and workflow automation, enabling AI not only to "understand" but also to actively suggest operations and promote decision - making.

Currently, Nabla has been deployed in more than 130 medical institutions globally, covering more than 85,000 doctors and nursing staff, and supporting more than 20 million patient consultations annually. Data shows that after using Nabla, the average time doctors spend on document work has been reduced by 50%, and patient satisfaction has been significantly improved.

As of now, Nabla's ARR is $16 million, and the per - capita revenue is $175,000.

14 Hebbia, Enterprise Knowledge Work Assistant (ARR: $13 million)

Hebbia is committed to building AI software for the finance and legal industries, providing high - level knowledge work organization services for companies through data and information organization.

In professional service fields such as financial services, law, and consulting, the ability to effectively find information is particularly important. Many investigation processes require manual analysis of a large number of dense documents.

However, what is particularly troublesome is the disorder and overflow of information caused by the excessive use of SaaS within the organization. It takes a lot of time to find effective information. According to a 2022 survey, 47% of digital workers said it was difficult to find the information needed to perform their jobs effectively.

This is where Hebbia comes in. Hebbia was established to provide neural search functions for investment banks and private equity analysts so that they can solve complex and ambiguous queries.

Their core product is Matrix. Users can upload any type of document, including PDFs, presentations, emails, and images, specify the questions they want to answer in each document, and immediately get detailed insights in a table format similar to Microsoft Excel.

At the beginning of 2025, Hebbia recently completed a $130 million Series B financing, led by Andreessen Horowitz, with participation from Index Ventures, Google Ventures, and Peter Thiel. This round of financing brought the company's valuation to $700 million. Currently, Hebbia's ARR is $13 million, and each employee generates $97,000 in revenue.

15 Decagon, AI Customer Service Agent (ARR: $10 million)

Decagon mainly builds AI Agents for large enterprises in the fields of customer support and customer experience. Most of the previous intelligent customer service systems were rigid robots, but relying on the continuously evolving AI Agent Engine, AI can truly understand customer needs and solve problems.

Decagon's core product, the AI Agent Engine, uses generative AI technology to handle complex customer consultations, provide personalized responses, and seamlessly integrate with existing customer support systems.

Moreover, the platform supports multi - channel communication, including chat, email, and voice calls, helping companies achieve 24/7 service. In addition, Decagon has launched DecagonVoice to provide a high - fidelity voice customer service experience.

Currently, Decagon's latest ARR is $10 million, and each employee generates $101,