The AI Agent Company Map 2026: Who's Building Agents, Who's Funded, and Who's Hiring
By Fast AI Startup Jobs
TL;DR
- 301 companies in our database are tagged as building AI agents. That's about 18% of the 1,692 companies we track, which tells you how much of the AI startup world has reoriented around agents rather than chatbots or single-shot models.
- 51 of them have raised $100M or more. Only six have crossed $500M total: Cognition (Devin), Sierra, Legora, Genspark, Temporal, and a stealth lab called Ineffable Intelligence.
- 167 of the 301 are actively hiring, with 2,640 distinct open roles between them (deduped by job ID, because our raw feed double-counts roles posted to multiple locations by about 36%).
- The single biggest agent hirer right now is Legora, a legal-AI company, with 185 open roles. Sierra (119), Decagon (107), EliseAI (104), and LangChain (103) round out the top five.
- The money is splitting along a clear line: horizontal "build-your-own-agent" platforms (LangChain, Dust, Lindy) versus vertical agents that own one job end to end (legal, support, finance, security). The verticals are raising larger rounds.
In this article: How we built this · Coding agents · Customer support · Voice agents · Sales & GTM · Security agents · Browser & web · Vertical agents · Infra & frameworks · What it means for job seekers · FAQ
How we built this map
Two things make a list like this either useful or worthless: where the company list comes from, and whether the numbers are real.
For the company list, we used the Agents tag in our own database. Every company on fastaijobs.com gets hand-tagged when we ingest it, and 301 carry that tag as of June 2026. We trust the tag more than a keyword scan: searching descriptions for "agent" pulls in plenty of companies that use the word loosely (a "customer agent dashboard" is not an agent company). The tag is the cleaner signal. Where a company sits in an ambiguous spot, we read its actual product description before placing it in a category below, and a few judgment calls are noted as we go.
For the numbers, every funding figure is the company's recorded total raised, and every job count is deduped by job ID. This matters more than it sounds. Our raw job feed contains 37,693 rows, but only 24,006 are distinct postings. The same role gets listed once per location, so a company hiring one "Software Engineer (Remote)" across five office cities shows up as five rows. We count distinct IDs, which knocks roughly 36% off the naive total. If a number here looks lower than what you'd see scraping a careers page, that's why: we're counting jobs, not listings.
One honest limitation: our company records don't store headquarters location, so this map is organized by what a company does and how it's funded, not by geography. Funding stages and amounts are as of each company's last recorded round.
Coding agents
The most mature agent category, and the one where the money is heaviest. These companies build agents that write, review, debug, and ship software, ranging from autonomous "do the whole ticket" systems to agents that pair with a human engineer.
| Company | Stage | Total Raised | Open Roles | What it does |
|---|---|---|---|---|
| Cognition AI | Series D | $1.7B | 60 | Makes Devin, the autonomous software engineer. The category's flagship. |
| Magic | Series C | $465M | 6 | Long-horizon coding agents with proprietary long-context models. |
| Factory | Series C | $220M | 25 | "Agent-native" development; its Droids run across the full SDLC. Used by MongoDB, EY. |
| Resolve AI | Series A | $200M | 13 | An autonomous "Production Engineer" that investigates and fixes incidents. Founded by OpenTelemetry's creators. |
| Imbue | Series B | $232M | 3 | Research lab training agents that reason and code; ships Sculptor and Bouncer. |
Cognition is the obvious leader, both in capital ($1.7B, latest round May 2026) and in name recognition. Devin defined what people mean by "coding agent." But the interesting story is the spread below it. Factory and Resolve are betting that the winning agent isn't a general coder but one wired into a specific workflow: Factory into enterprise dev pipelines, Resolve into production incidents. That's the same vertical-vs-horizontal split you'll see in every category on this list.
Worth flagging: this is the category most exposed to acquisition. Cursor (Anysphere), the most valuable coding-tools company of all, was just acquired by SpaceX for $60B. When the platform owners decide they want the coding layer, they buy it.
Customer support and CX agents
The category with the clearest business case. Support is expensive, measurable, and repetitive, which makes it the first place enterprises actually deploy agents in production rather than in a demo.
| Company | Stage | Total Raised | Open Roles | What it does |
|---|---|---|---|---|
| Decagon | Series D | $481M | 107 | Enterprise CX agents across chat, voice, email, SMS, with controlled deployments. |
| Sierra | Series E | $1.58B | 119 | Bret Taylor's conversational platform; $10B valuation, voice now the primary channel. |
| Kore.ai | Growth | $296M | 0 | Conversational and agentic AI for regulated-industry service and support. |
| Ada | Series C | $200M | 8 | Resolves support at scale in 50+ languages for brands like Meta. |
| Serval | Series B | $127M | 34 | Internal IT and employee-support agents for enterprises. |
Sierra is the heavyweight here at $1.58B raised and 119 open roles, the most of any pure-play support company. It's also a useful illustration of how fast this category is moving: the company describes voice as now its primary channel, which is why it shows up on both this list and the voice list below. The lines between "support agent" and "voice agent" are dissolving.
Decagon is the one to watch on hiring momentum, with 107 open roles after a $250M Series D in January 2026. If you want to join a support-agent company that's clearly in scale-up mode, Decagon and Sierra are where the headcount is going.
Voice agents
Voice was a research problem two years ago and is a product category now. Latency dropped, the models got good at interruptions and turn-taking, and suddenly an agent that handles phone calls is shippable. This is the fastest-growing slice of the map by company count.
| Company | Stage | Total Raised | Open Roles | What it does |
|---|---|---|---|---|
| Sesame AI | Series B | $308M | 23 | Lifelike voice companions; flagship model Maya, plus audio eyewear hardware. |
| PolyAI | Series D | $204M | 13 | Enterprise voice assistants for hospitality, healthcare, finance. Cambridge spinout. |
| LiveKit | Series C | $174M | 19 | Open-source realtime voice/video infra; powers ChatGPT Voice Mode. |
| Assort Health | Series B | $102M | 32 | Voice agents for healthcare call centers and patient scheduling. |
| HappyRobot | Series B | $62M | 77 | Voice workers for supply-chain ops: calls, emails, routine comms at scale. |
| Vapi | Series B | $72M | 30 | Developer platform for building voice agents; abstracts the realtime infra. |
| GigaML | Series A | $65M | 54 | Voice agents for customer care, claiming 90%+ resolution accuracy. |
| Bland | Series B | $65M | 12 | Low-latency phone agents for support, sales, and call-center workflows. |
Two patterns stand out. First, voice agents are going vertical fast: Assort owns healthcare scheduling, HappyRobot owns freight and supply chain. The generic "voice agent for anything" pitch is losing to companies that pick one industry's phone calls and nail them. Second, HappyRobot (77 roles) and GigaML (54 roles) are hiring far above their funding weight, which usually signals a company that's found product-market fit and is racing to capture a vertical before competitors notice.
LiveKit is in a different position: it's the picks-and-shovels play. It doesn't build voice agents, it builds the realtime infrastructure other people's agents run on, including OpenAI's. If you'd rather build the layer everyone depends on than compete in a vertical, that's the profile.
Sales and GTM agents
The "AI SDR" wave. These companies build agents that prospect, write outreach, and qualify leads, pitched explicitly as replacing or augmenting sales-development reps.
| Company | Stage | Total Raised | Open Roles | What it does |
|---|---|---|---|---|
| 11x | Series B | $76M | 0 | Digital workers for GTM teams: prospecting, outreach, lead qualification. |
| Unify | Series B | $59M | 14 | Agentic GTM combining intent data with automated outreach. |
| Artisan AI | Series A | $46M | 8 | "AI employees" for GTM; Ava handles AI SDR workflows. |
This is the category I'd approach with the most caution as a job seeker. The funding is real but the round sizes are smaller than the other categories (nothing here is past Series B), and the hiring is thin: 11x, which raised $76M, shows zero open roles in our data. That doesn't mean the company is in trouble, but a well-funded company that isn't hiring is a signal worth asking about in an interview. The AI-SDR space also faces an unusually direct question about whether outbound email actually works better when it's automated, and that question isn't settled.
If the category is thin in your search, that's the honest reality of the data, not an omission on our part.
Security agents
Security is one of the few places where "autonomous agent that takes action" is both genuinely useful and genuinely fundable, because a security operations center is drowning in alerts that a human can't triage fast enough. These companies build agents that investigate and respond to threats.
| Company | Stage | Total Raised | Open Roles | What it does |
|---|---|---|---|---|
| TENEX.AI | Series B | $277M | 54 | AI-native managed detection and response; AI models plus human analysts. |
| 7AI | Series A | $166M | 28 | "Swarming" AI agents that automate SOC investigation and response. |
| Reco | Series B | $85M | 9 | Agentic SaaS security posture management. |
| Cogent Security | Series A | $53M | 13 | Autonomous vulnerability management agents. |
| Prophet Security | Series A | $41M | 7 | Agentic SOC analyst that triages and investigates alerts. |
| Dropzone AI | Series B | $57M | 0 | AI SOC analyst for autonomous alert investigation. |
TENEX.AI leads on both funding ($277M) and hiring (54 roles), and its model is telling: it pairs proprietary security AI with human analysts rather than going fully autonomous. That hybrid framing shows up across the category, because a security agent that's wrong is a different kind of problem than a chatbot that's wrong. For job seekers with a security background, this is one of the better-funded, actively-hiring corners of the agent world, and the founder pedigrees here (ex-Splunk, ex-Mandiant) tend to be strong.
Browser and web agents
A smaller but architecturally important category: agents need to use the web, and most of the open internet wasn't built for them. These companies provide the browsers, proxies, and APIs that let agents actually click, scroll, and read.
| Company | Stage | Total Raised | Open Roles | What it does |
|---|---|---|---|---|
| Parallel | Series B | $230M | 15 | Web infrastructure and APIs so agents can search and browse reliably. |
| Browserbase | Series B | $68M | 5 | Cloud-hosted headless browsers built for agents; proxies and stealth. |
| Kernel | Series A | $22M | 7 | Browser infrastructure for AI agents. |
| Browser Use | Seed | $17M | 0 | Open-source framework for agents to navigate sites; 80,000+ GitHub stars. |
Parallel's $230M raise is the surprise here, the largest in a category most people don't know exists, and it tells you investors believe "web access for agents" is real infrastructure rather than a feature. This is a small list with only a handful of well-funded players, so we're being upfront: the category is early, and most of the companies are seed or Series A. But it's the kind of unglamorous infrastructure bet that occasionally turns into the layer everything depends on (see what happened to LiveKit in voice).
Vertical agents (legal, finance, health)
The biggest bucket by count, and where some of the largest rounds outside of foundation models are landing. These companies take one profession's most painful workflow and build an agent that owns it.
| Company | Stage | Total Raised | Open Roles | What it does |
|---|---|---|---|---|
| Legora | Series D | $866M | 185 | Collaborative AI workspace for lawyers to research, draft, review. |
| EliseAI | Series E | $392M | 104 | Automates the housing and healthcare consumer journey; $2.2B valuation. |
| Writer | Series C | $369M | 50 | Full-stack enterprise platform for building and governing agents on company data. |
| Rogo | Series D | $310M | 42 | Financial-analysis agents for investment banking; built by ex-bankers. |
| AppZen | Series D | $290M | 12 | Agentic AI for finance: auditing expenses, invoices, contracts. |
| Hebbia | Series B | $161M | 0 | Document-analysis agents for asset managers and investment banks. |
| Norm AI | Series B | $136M | 4 | Turns regulations into autonomous compliance agents. |
Legora is the headline. At $866M total raised and 185 open roles, it's the single biggest agent hirer in our entire database, ahead of every coding or support company. A legal-research tool out-hiring Devin's maker is the clearest evidence that vertical agents, not horizontal platforms, are where the operational scale is right now. EliseAI (104 roles, $2.2B valuation) tells the same story in real estate and healthcare.
The logic is straightforward: a vertical agent has proprietary data, a defined buyer, and a workflow that's hard to replicate as a feature inside ChatGPT. That defensibility is exactly what's keeping these companies independent while feature-level tools get absorbed. If you're optimizing for a company that's likely to still exist (and still be hiring) in two years, the vertical agents are the safer bet on this map.
Agent infrastructure and frameworks
The layer everyone building an agent has to choose: the frameworks, orchestration, and observability tools. Lower-margin and more contested than the verticals, but it's where a lot of the genuinely interesting engineering work lives.
| Company | Stage | Total Raised | Open Roles | What it does |
|---|---|---|---|---|
| LangChain | Series B | $160M | 103 | Open-source frameworks (LangGraph) and managed platform (LangSmith) for building and observing agents. |
| Dust | Series B | $62M | 25 | Platform for building custom org agents connected to company knowledge. |
| Lindy | Series B | $50M | 0 | Build-and-deploy "AI employees" for repetitive knowledge work. |
| Mastra | Series A | $35M | 2 | Open-source TypeScript framework for agents; built by the Gatsby team. |
| CrewAI | Series A | $18M | 2 | Open-source multi-agent orchestration framework. |
| Gumloop | Series B | $70M | 6 | No-code AI automation for building agentic business workflows. |
LangChain's 103 open roles are the standout, the second-most of any infrastructure-layer company on this map, which makes sense given how much of the agent ecosystem runs through LangGraph and LangSmith. But this is the category most squeezed by the foundation-model companies moving down the stack. As we documented in the middle-layer analysis, when OpenAI or Anthropic needs a piece of agent infrastructure, they increasingly buy the team rather than integrate the tool. The open-source frameworks here (Mastra, CrewAI) face that risk most directly: huge adoption, thin revenue, and a buyer who only wants the people and the code.
That's a real consideration if you're weighing a job offer from a framework company, and we'd rather say it plainly than pretend the category is risk-free.
What this map means if you're job hunting
A few things fall out of the data that should shape where you point your search.
The hiring is concentrated in verticals, not platforms. The five biggest agent hirers (Legora, Sierra, Decagon, EliseAI, LangChain) include four companies that own a specific workflow and only one pure framework. If you want the most open roles and the most operational scale, follow the vertical agents into legal, support, real estate, and finance.
Funding and hiring don't always move together, and the gap is information. 11x raised $76M and shows zero open roles; Hume-adjacent infra plays sit on capital without expanding headcount. A well-funded company that isn't hiring isn't necessarily failing, but it's a question to ask directly: is this a deliberate efficiency stance, or a sign the round is being used to extend runway rather than grow?
Match your category to your risk tolerance. Vertical agents (Legora, Decagon, Rogo) have defensible data moats and tend to stay independent. Infrastructure and framework companies (LangChain, Mastra, CrewAI) do the most technically interesting work but face the highest absorption risk from foundation-model companies buying down the stack. Coding agents sit in between: huge upside, but also the category big platforms most want to own outright, as the SpaceX-Cursor deal just demonstrated.
And read the round date, not just the amount. Several companies here last raised in 2023 or 2024. In a market moving this fast, a two-year-old round and a stalled headcount is worth more scrutiny than a smaller, fresher raise with active hiring.
You can browse all 301 agent companies, with verified funding data and live deduped job counts, at fastaijobs.com/companies.
The bigger picture: "AI agent" stopped being a single category some time in the last year. It split into coding, voice, support, security, vertical SaaS, and the infrastructure underneath all of them, each with different economics and different odds. The companies that will still be hiring in 2028 are the ones that picked a specific job and made an agent genuinely better at it than a person, not the ones selling "agents for everything."
FAQ
How many AI agent companies are there in 2026?
In our database of 1,692 tracked AI startups, 301 are tagged as building AI agents as of June 2026, about 18% of the total. Of those, 51 have raised $100M or more and 167 are actively hiring. The real number across the whole market is higher, since we only track companies we've ingested and verified, but the proportions are a good read on how much of the AI startup world has reoriented around agents.
Which AI agent company has raised the most money?
Among pure agent companies in our data, Cognition (Devin, coding) leads at $1.7B total raised, with Sierra (conversational/CX) close behind at $1.58B. Legora (legal) has crossed $866M. The largest deal connected to the agent world recently wasn't a fundraise at all: SpaceX's $60B acquisition of Anysphere, the maker of Cursor.
What's the difference between an AI agent company and a regular AI startup?
An AI agent company builds software that takes multi-step actions autonomously: it plans, calls tools, browses, writes code, makes phone calls, and completes a task end to end, rather than answering a single prompt. A regular AI startup might wrap a model in a chat interface or generate content on demand. The line is whether the product does work on its own versus assisting a human one response at a time.
Which AI agent companies are hiring the most right now?
By deduped open-role count in our data, the top agent hirers are Legora (185), Sierra (119), Decagon (107), EliseAI (104), and LangChain (103). The pattern is that vertical agents (legal, real estate, support) are hiring at larger scale than horizontal platforms, with the exception of LangChain on the infrastructure side.
Are AI agent jobs concentrated in coding agents?
No. Coding agents get the most press, but the most open roles are in vertical agents and customer-support agents. Legora (legal) alone has more open positions than any coding-agent company in our database. Coding is the most mature category by funding, but support, voice, and vertical SaaS are where headcount is growing fastest.
Is it risky to join an AI agent infrastructure startup?
It carries a specific risk: foundation-model companies (OpenAI, Anthropic) increasingly acquire agent-infrastructure and framework teams rather than integrate their tools, especially open-source projects with big adoption but little revenue. That's not a reason to avoid the category, but it's worth asking about a company's revenue model and acquisition exposure before you sign. Vertical agents with proprietary data tend to be more durable as independent companies.
Data sourced from fastaijobs.com as of June 2026. We analyzed 1,692 tracked AI startups, of which 301 carry our Agents tag, cross-referenced against 24,006 distinct open job postings (deduped by job ID from a raw feed of 37,693 rows). Funding figures reflect each company's last recorded round. Our records do not store company headquarters, so this map is organized by category and funding rather than geography.