The most innovative companies in applied AI for 2026

Three years (plus) after the arrival of ChatGPT, chatbots are morphing into AI agents. As generative AI models have improved and become able to reason in real time, the major AI labs, starting with Anthropic, have begun to shift their research focus from models that compose and comprehend text to ones that reason, use tools, and work autonomously. 

The first kind of agent that matured to the point of having real-world impact was an agent that can write, test, and document computer code. Coding agents, powered by language models, can understand plain language, which has democratized software development and made “vibe coding” possible. Products like Lovable and Bolt allow nontechnical or semi-technical product managers or marketing chiefs to quickly mock up working prototypes of apps or site features. Both products saw significant gains in both users and revenue over the past year. As did Cursor, which works alongside human software engineers in their normal user interface to build within large preexisting codebases. Anthropic’s Claude Code and OpenAI’s Codex also saw surges in usership in the last half of 2025. 

Customer service was another early application of AI agents, but many have been limited in their knowledge and reasoning ability, and therefore limited in scope and utility. That too is changing. Sierra, which was founded by ex-Salesforce exec (and OpenAI board member) Brett Taylor and ex-Google hardware chief Clay Bavor, is developing agents that act like all-purpose concierges or “brand representatives.” Sierra agents remember a consumer’s past interactions with the company and are versatile enough to handle a range of tasks including product returns, account updates, subscription issues, and appointment setting. Cognigy (now a part of Nice) formally launched its AI agent platform in 2025; its customer service AI agents work with human operators to plan and carry out tasks, such as handling complex transactions and coordinating across systems. 

ServiceNow, which provides a sort of operating system for customer support, is broadening its platform to enable the deployment and management of AI agents (both its own and those from third parties) across many departments across the enterprise including customer service, human resources, and IT. As AI agents proliferate, governance of the agents will become a big priority for enterprises. That’s why Credo AI’s Agent Registry is timely; the system gives enterprises a way to register all agents used by the company, as well as real-time oversight of the actions agents are taking, what data they’re accessing, and how they’re arriving at their decisions. 

In the first half of 2026, coding agents and customer service agents have already begun to reshape how organizations manage—and staff—those key business functions. By this time next year, we may look back at those applications as just the low-hanging fruit, the first of many to come. 

1. Sierra

For developing more thoughtful agents for business

Most AI agents in customer service still behave like glorified chatbots—fast, yet they fail to retain conversational context. Sierra, which is building AI agents that function less like ticket responders and more like long-term brand representatives, is focused on a core challenge in enterprise AI: memory. The company was founded in 2023 by Bret Taylor, who currently chairs OpenAI’s board and previously co-created Google Maps before serving as CTO of Facebook and co-CEO of Salesforce, and Clay Bavor, who spent 18 years at Google leading AR and VR initiatives and overseeing product and design for Workspace apps such as Gmail and Docs.

Sierra’s agents retain context across interactions, allowing them to recall previous purchases, unresolved issues, and customer preferences. The result is AI that functions less like a support chatbot and more like a persistent digital relationship layer for businesses. The approach has resonated with enterprises seeking deeper automation as Sierra reached $150 million in annual recurring revenue just eight quarters after launching in February 2024. It raised $350 million in funding in 2025 at a $10 billion valuation. In 2025, Sierra added voice capabilities to its platform, enabling customers to converse with memory-aware agents over phone calls, messaging, and web channels.

2. Cursor

For putting agents where developers need them

Software creation often requires engineers to toggle between writing code and managing the invisible labor around it, such as reviews, debugging, and deployment. Cursor aims to collapse the ongoing developer friction by embedding autonomous AI agents directly into coding environments.

In 2025, Cursor 1.0 introduced an AI-native development environment that integrates Agent, Bugbot, and Background Agent across IDEs, the command line, Slack, Linear, and the web. The Cursor Agent can complete complex end-to-end tasks, from building features to testing and migration, while the Background Agent runs long-horizon coding tasks in the cloud, allowing developers to parallelize work without interrupting flow. Cursor’s Bugbot has reviewed more than 3 million pull requests, identifying more than 1.5 million issues with resolution rates exceeding 50%.

The platform currently serves 31 million-plus users and is used by 64% of the Fortune 500, including Nvidia, Adobe, and OpenAI, generating more than 100 million lines of enterprise code daily. Cursor recently surpassed $2 billion in annualized revenue, doubling its recurring revenue in just three months.

3. Coactive

For making sense of vast video libraries

Video is one of the largest untapped data assets inside modern enterprises. Petabytes sit in cloud buckets and on local servers, yet almost none of it is truly searchable. At best, companies rely on manual tags and file names. Coactive turns raw visual content into structured, queryable intelligence.

The platform scaled its Multimodal Application Platform (MAP), a production-ready system that analyzes images and video directly—without relying on pre-labeled metadata. MAP applies multimodal AI to detect contextual signals across billions of frames, in turn automatically generating rich metadata insights to feed moderation systems, recommendation engines, ad targeting, and other discovery workflows. Coactive’s Intelligent Sampling isolates high-value video moments in seconds, while Intelligent Search allows organizations to build custom taxonomies tailored to their domains.

MAP processes billions of images and video frames, with usage growing fivefold across large-scale deployments at NBCUniversal, Thomson Reuters, and other media-heavy enterprises. At Fandom, MAP automated 88% of image labeling within weeks, cutting manual review time by 74% and reducing moderation costs by 50%.

4. Lovable

For democratizing software development by making it conversational

For decades, building software required fluency in programming languages and algorithmic concepts, as well as access to specialized teams for mid- to large-scale projects.

Launched in late 2024, Lovable is breaking that barrier through its coding assistant platform, enabling development through natural language. The platform launched Lovable Agent, an agentic development system that turns a written idea into a deployed application. The agentic AI can autonomously manage files and coordinate front-end and backend logic and data layers. Lovable Cloud comes with a backend for authentication, storage, and secure connections, while Lovable AI allows for functionalities such as content summary, image creation, and translation based on prompts.

Lovable’s annual recurring revenue (ARR) reached $200 million in November, and in December it raised a $330 million Series B at a $6.6 billion valuation. More than 35 million projects have been built on Lovable, which finished February with 500,000 paid users. Enterprise customers include Zendesk, Uber, Hubspot, Microsoft, and ElevenLabs.

5. ServiceNow

For taking its generative AI agents beyond customer service to support other departments

Enterprise generative AI deployments are often clustered around chatbots and customer support, leaving other departments untouched. ServiceNow is helping extend agentic automation by treating AI agents as a coordinated digital workforce spanning entire businesses.

Rather than confining AI innovations to support desks, ServiceNow expanded its platform into HR, finance, IT, security, and field operation domains through its single architecture and data model to coordinate work across departments. Two breakthrough capabilities anchored the shift. AI Control Tower provides executives with a central console to monitor and govern every agent in operation, whether built by ServiceNow or third parties. Workflow Data Fabric connects live data from hundreds of enterprise systems, allowing agents to act with real-time context instead of scripted prompts.

The platform’s customers, including AstraZeneca, reported saving 30,000 hours annually by automating research workflows, while the city of Raleigh, North Carolina, digitized 20 municipal processes and enhanced its citizen portal through the offerings. Likewise, Bell Canada optimized more than 2 million service jobs powered by AI-driven scheduling. Nearly 1,000 organizations have deployed agentic AI with ServiceNow, and the momentum helped drive subscription revenue to $12.88 billion in 2025, representing a 21% year-over-year growth.

6. Credo AI

For helping build confidence in AI by creating trust scores for the latest models

Choosing an AI model for a specific use case has started to feel like buying software without a spec sheet, and Credo AI is trying to fix that. In 2025, the company launched Model Trust Scores, a use-case-based leaderboard that evaluates 55 leading models across 95 representative use cases in 21 industries. 

Instead of ranking models on raw performance alone, the system measures risk, reliability, and policy alignment, which enables enterprises to weigh trade-offs before deployment. Credo also introduced an AI Agent Registry to monitor autonomous systems in real time, tracking data access, decision pathways, and policy compliance as agents act over data.

The offering has resonated with customers looking for governance at scale. Meanwhile, a new integration with Microsoft Azure AI Foundry integrates governance directly into the model development process. Credo AI doubled its revenue in 2025 and grew its enterprise customer base by 150%.

7. Gong

For arming sales teams with purpose-built AI agents

Sales teams sit on a gold mine of customer conversation data. However, most of it is forgotten the moment the call ends. Gong is helping turn those interactions into action. The company expanded its Revenue AI platform with 18 specialized AI agents designed specifically for go-to-market teams. Instead of generic copilots, these agents can analyze unstructured data such as calls, emails, and meetings to surface deal risks, flag competitive threats, and recommend next steps. The platform’s Orchestrate product allows revenue leaders to define sales plays once and automatically guide teams through execution, measuring adherence and impact in real time. Teams using Gong’s Smart Trackers—an agent trained to detect contextual deal signals—report a win rate increase by 35%. Upwork reached 95% forecast accuracy after deploying Gong Forecast. Specifically, the Ask Anything agent, which allows reps to ask questions about their entire deal history, achieved a 26% increase in win rates.

Gong surpassed $300 million in annual recurring revenue in 2025 and now serves more than 5,000 customers, including Google, ADP, and Uber for Business.

8. Graphite

For developing agents that fix AI-generated code

AI can now write code in seconds, thanks to innovations in LLMs. But reviewing the generated output still takes hours for developers. Graphite is helping enterprises automate the harder half.

The company launched Diamond, an agentic code reviewer designed to analyze pull requests with the judgment of a senior engineer. Instead of flooding developers with generic comments, Diamond prioritizes real risks such as bugs, security gaps, and logic errors, then adapts to each team’s style guides and repository history. Powered by fine-tuned LLMs from Anthropic and OpenAI, the system has reportedly reviewed more than 900,000 pull requests with an over 95% positive feedback rate and a 52% comment acceptance rate, outperforming most AI reviewers and even rivaling human benchmarks.

The platform also introduced Graphite Chat, a conversational layer that lets engineers query their codebase directly inside a pull request, ask for fixes, and merge changes without switching tools. The goal is to make the review process the control center of AI-era software development. Companies including Snowflake, Figma, and Notion now rely on Graphite to review code at scale. Ramp reported a cut median time between merged pull requests by 74%, while Shopify increased PRs shipped per developer by 33%. Graphite raised $52 million in Series B funding in 2025 from Accel and Anthropic’s fund.

9. Adobe

For activating PDFs with AI

The PDF was designed to preserve information. Today, Adobe is using it to enable conversation. The company’s Acrobat Studio blends Acrobat’s document intelligence with Adobe Express’s creative tools to turn static files into interactive workspaces. At its core are AI-powered “PDF Spaces,” where users can upload documents, assign AI assistants roles such as analyst or instructor, and query collections of files with source-linked citations. Instead of copying text into separate apps, teams can compare versions, extract insights, and instantly transform findings into additional content—all within the same environment.

The shift is practical as corporate knowledge (policies, reports, agreements) often lives inside PDFs, yet remains difficult to analyze at scale. Acrobat Studio surfaces information without moving data outside secure environments and includes encryption, sandboxing, and centralized controls. The move comes as Adobe navigates regulatory scrutiny over subscription practices and AI training claims, even as its digital media revenue grew to $17.65 billion in 2025, up 11% year over year.

10. Sardine

For using AI agents to stop financial crime

Scammers and their malpractices have evolved to become fast, coordinated, and AI-assisted. To help fight automation with automation, Sardine launched what it calls the first agentic AI platform for overseeing financial crime management, unifying fraud prevention and anti–money laundering into a single system. Instead of simply flagging suspicious transactions, the platform’s AI agents investigate them—pulling device intelligence, behavioral biometrics, and transaction history into a real-time narrative before money moves. Its models can analyze typing cadence, hesitation patterns, and session flow to detect when a customer may be coached by a scammer during a “safe transfer.” If behavior shifts, the system can pause or trigger intervention.

Beyond individual accounts, Sardine operates a consortium network spanning more than 3 billion devices, allowing institutions to share anonymized risk signals and disrupt mule networks. The platform currently supports more than 300 businesses in 70 countries. Sardine raised $70 million in Series C funding in early 2025.

11. Nice

For putting AI agents to work in call centers

The modern call center is no longer just rows of headsets. It’s increasingly staffed by software—and that software needs to be increasingly flexible. Last fall, AI platform giant Nice acquired AI voice agent pioneer Cognigy for $955 million. Cognigy is known for advancing agentic AI systems that don’t just answer scripted questions but can plan, execute, and resolve complex customer requests under human oversight. The platform’s AI agents can handle customer inquiries across industries, from rebooking flights during disruptions to processing financial inquiries, coordinating real-time backend orchestration. For live representatives, Cognigy’s copilot tools surface context and suggested actions mid-conversation, reducing friction without removing human judgment.

Customers have seen reductions in average handling times of up to 30%, and a major airline increased the number of self-service interactions from 10,000 to 30,000 per day with the help of Cognigy’s agents. The company realized 100% year-over-year revenue growth in 2025, expanding its services into the healthcare, finance, and aviation sectors.

12. Baseten

For hosting the infrastructure for AI startups

Behind every fast AI answer is a slower, more complicated layer, dependent on the model’s inference capabilities, that most users never see. Baseten helps startups and enterprises scale inference—the moment when a model responds to a user query—and allows startups to deploy and scale the best open-source models without having to manage a GPU cluster. The company launched production-grade model APIs for popular models such as Qwen 3, Kimi K2, and DeepSeek R1, with a focus on low latency and enabling reliable data processing. It also introduced Baseten Training, allowing companies to pretrain and fine-tune models on the same infrastructure they use for deployment. Customers, including Abridge, Writer, and Superhuman, rely on Baseten to keep models responsive in domains such as healthcare, productivity, and enterprise applications. The company’s revenue increased more than tenfold year over year, while customer count grew 5.5x. Baseten recently raised a $300 million Series E, valuing its platform at $5 billion. 

13. ConcertAI

For increasing the odds of successful clinical trials

Clinical trials are one of the most expensive and uncertain steps in bringing new medicines to market. Promising therapies often fail not because the science is wrong but because trials are poorly designed, recruit the wrong patients, or encounter operational hurdles. ConcertAI applies AI to large clinical and real-world healthcare datasets to help pharmaceutical companies predict trial outcomes and identify bottlenecks that might derail a study.

Its Accelerated Clinical Trials (ACT) system uses AI to help researchers design protocols, select trial sites, and match patients to studies. By forecasting enrollment and outcome probabilities, the platform allows teams to test trial strategies computationally before real-world launch. In 2025, the company expanded its capabilities with the Precision Suite, a set of generative and agentic AI tools designed to help life sciences organizations analyze clinical data and accelerate drug development. Biopharma companies, including Bayer and Bristol Myers Squibb, use the technology to optimize studies.

ConcertAI claims its systems can shorten clinical trial timelines by up to 10 to 20 months and reports serving 75% of the top life sciences companies. As drug development costs continue to rise, the company is positioning AI as a research layer that helps promising therapies reach patients faster.

14. Bolt.new

For letting non-coders prototype apps

What if building software felt more like texting than coding? Bolt.new, launched late 2024, turns a browser into a full development environment powered by StackBlitz’s WebContainers. Users only need to specify what they want to create, and the tool automatically generates, executes, and launches functional web or mobile applications without requiring any local development environment, servers, or setup. Bolt is designed for rapid development from scratch, making it ideal for product managers, designers, and entrepreneurs who want to prototype before engaging their engineering teams.

In 2025, the firm grew with Bolt B2, incorporating sophisticated coding agents such as Claude Code and OpenAI Codex right into its UI, along with cloud hosting and multi-agent support. It lets teams move from prototype to deployment within the browser. The company grew to $40 million in ARR in six months and has more than 8 million users.

15. Temporal

For making agents more reliable

AI agents are good at starting tasks, but not always good at finishing them. Temporal was built for the messy middle. By enabling AI agents to take on real operational responsibilities, the company is positioning reliability as infrastructure, rather than an afterthought.

As enterprises began to move their agents from demo environments to actual workflows, errors proliferated, including APIs timing out, rate limits being reached, and long-running tasks losing state. Temporal’s platform acts as a durable execution layer, ensuring that multistep workflows pick up where they left off instead of having to start over. This attention to reliability was further emphasized with the inclusion of Temporal’s platform in OpenAI’s Agents SDK, which was already being leveraged by the AI model developer to power production systems such as ChatGPT image creation and Codex.

Every step within a workflow is recorded, recoverable, and observable across clouds. Developers no longer need to write custom retry logic or rebuild brittle state machines. Temporal reported 380% year-over-year revenue growth in 2025, and raised a $300 million Series D at a $5 billion valuation in February 2026.

16. Cleo

For producing an AI financial assistant with attitude

Most budgeting apps quietly log your spending and send polite reminders. Cleo has opinions—and isn’t afraid to share them. Cleo has built an AI financial assistant that mixes blunt humor with behavioral science, aiming to keep users engaged long enough to change habits. Last year, the platform rolled out Debt Reset, which automatically pulls in users’ liabilities and constructs adaptive payoff strategies based on interest rates, due dates, and spending behavior. It also launched Money IQ, a gamified voice quiz that surfaces hidden subscriptions and recurring charges, turning financial blind spots into weekly prompts for action.

Under the hood, Cleo’s AI processes an average of 6.4 billion transactions per day, using memory and real-time reasoning to tailor advice. The company reported $333 million in annual recurring revenue in December 2025 and more than 1.5 million monthly active users, with four consecutive years of doubling revenue and subscribers. Cleo’s real differentiator is tone as its AI layers reasoning with personality. Its much-discussed “Roast Mode,” shared more than 500,000 times and drawing 2.4 million TikTok likes, turns overspending into a blunt, sometimes biting, intervention. The joke lands—but so does the message. The company is betting that financial coaching works best when it feels personal.

17. Aurora

For making autonomous trucking a reality

In April 2025, autonomous trucking moved from pilot programs to paying customers. Aurora put driverless trucks on public highways hauling freight every day—with no human in the cab—marking the first sustained commercial deployment of autonomous long-haul trucking.

At the core is the Aurora Driver, which pairs deep learning with a rules-based “verifiable AI” architecture built to make its decisions traceable and auditable—an answer to growing unease around opaque, black-box autonomy systems. Backed by its proprietary FirstLight lidar and millions of simulated scenarios, the system now runs routes along the Dallas–Houston corridor and has logged more than 50,000 commercial driverless miles with a perfect safety record.

Last year, Aurora generated $3 million in commercial revenue, hauling freight for partners including FedEx, Werner, Schneider, Hirschbach, and Uber Freight. Moreover, the company is also working with truck manufacturers Paccar and Volvo to integrate the Aurora Driver at the factory level. The trucking sector is experiencing a chronic shortage of drivers, which continues to threaten the supply chain economy. With its solution moving autonomy onto active routes, Aurora is demonstrating that self-driving trucks aren’t a distant concept—they are operating in traffic today.

18. Mastercard

For laying the pipes for agent-driven commerce

AI agents are set to start shopping on our behalf—and Mastercard is positioning itself as the infrastructure layer that makes sure they can actually complete the payment. In 2025, the payments giant introduced Agent Pay, a framework that allows verified AI agents to complete transactions on a user’s behalf using tokenized credentials. Built on Mastercard’s existing tokenization network, the system issues “agentic tokens” that authenticate both the consumer and the software agent, helping distinguish legitimate AI assistants from malicious bots.

Supporting tools include an Agent Toolkit, Insight Tokens, and an Agent portal—giving banks and platforms a way to embed payments directly into conversational and agent-based systems.

Partners such as Microsoft, IBM, OpenAI, Braintree, and Checkout.com are integrating Mastercard’s capabilities into AI platforms, while all U.S.-issued cards will be enabled for agentic commerce, with global expansion to follow. Mastercard’s payment network handled more than 175 billion switched transactions in 2025, up from 159 billion in 2024.

19. Applied Intuition

For making vehicles “intelligent companions”

Instead of building its own cars, Applied Intuition builds the software brains that other manufacturers rely on. The company introduced its Self-Driving System (SDS) for Automotive, a white-box autonomy stack designed for passenger vehicles. Unlike opaque, black-box systems, SDS provides automakers with visibility into perception, planning, and control layers while remaining hardware-agnostic. 

The aim is to help deliver advanced driver-assistance features with higher autonomy levels, without locking OEMs into a single chip or sensor supplier. Applied Intuition also expanded into in-vehicle intelligence through a collaboration with OpenAI, which integrates LLMs into vehicles, turning dashboards into conversational interfaces that can remember preferences, assist with tasks, and connect with users’ mobile devices.

The company’s domain reach currently spans automotive, trucking, mining, and defense, with partnerships including Traton Group and Komatsu. In 2025, Applied Intuition closed a $600 million Series F round at a $15 billion valuation, signaling investor confidence in its platform approach.

20. Delphi

For building AI proxies for real people

What if an autonomous expert built atop your skills could keep working after you log off? Delphi developed what it calls “digital minds”—interactive AI versions of real people trained on their books, podcasts, videos, and writing. Users link their content, verify their identity, and generate a conversational agent that answers questions in their voice. To maintain fidelity, the company says each digital mind includes guardrails designed to reduce hallucinations and keep responses aligned with its human source. The platform refined this concept with Interview Mode, which lets the system ask creators clarifying questions to fill knowledge gaps, and a Mind Quality Score that measures how accurately the AI reflects its human source. 

More than 3,000 customers, including Arnold Schwarzenegger, Jay Shetty, Simon Sinek, and other celebrities, are currently using the service.

Delphi reported $4.8 million in ARR as of September 2025 and more than 8 million monthly interactions with digital minds. While the AI model has raised questions about identity and authenticity, Delphi claims that all accounts are verified and consent-driven. In a world where AI impersonation and synthetic content are on the rise, Delphi is vying to establish itself as an infrastructure for digital identity.

Explore the full 2026 list of Fast Company’s Most Innovative Companies, 720 honorees that are reshaping industries and culture. We’ve selected the companies making the biggest impact across 59 categories, including advertising, applied AI, biotech, retail, sustainability, and more.

View the full article