Enterprise AI Agents: Labelbox Raises $100M+ Series E

Labelbox, the AI platform powering reinforcement learning for frontier models and managed agents, announced it has raised more than $100 million in Series E financing to scale its Enterprise AI Agents platform, according to a company announcement on October 14, 2026. The round combines equity with a credit facility and was led by Infinitum Partners, with participation from Wipro Ventures, Andreessen Horowitz (a16z), B Capital, Tau Ventures, Snowpoint Ventures, and others.

Chief executive and co-founder Manu Sharma put the pre-money valuation at $1.25 billion, reported by Unite.AI. The company said the financing will fund continued research and engineering across its two core products, Horizon and Recursion Managed Agents, along with expansion of the teams bringing those products to customers. Sharma indicated that access to compute will be central to the company's next chapter, noting that Infinitum's John Yetimoglu was an early investor in Groq and is involved in building large-scale compute infrastructure at GroqCloud.

The raise lands as investor money pours into the infrastructure layer of the agent economy. Where the last cycle funded chatbots, this one funds Enterprise AI Agents that execute real work inside real business systems, with measurable outcomes and compounding improvement. Labelbox is betting that the winners will own the full stack: the sandbox, the agent harness, the reinforcement-learning loop, and the data flywheel connecting them.

Recursion Managed Agents: agents graded like employees

Labelbox opened Recursion Managed Agents to enterprise customers earlier in October 2026, and the product reads like a direct answer to the governance questions holding back Enterprise AI Agent deployment. A Managed Agent receives a goal, a trigger such as a schedule, an alert, a Slack message, a webhook, or an API call, and a rubric defining what good work looks like. A separate grader agent then scores every run against the rubric line by line, pointing to evidence for each judgment; when the work misses, the agent revises it, according to Unite.AI.

Agents can operate in three modes: multi-agent, in which a coordinator chains specialists that each carry their own model, instructions, tools, and permissions; team, in which copies of an agent split work on a shared board and check each other's results; and auto, the default, in which the agent decides for itself when a job needs specialists or a team. The agents work across more than 100 business applications and any MCP server, with every tool scoped as read only, changes data, or destructive, and credentials held in a vault the model never sees, according to the company's announcement as syndicated by World Lifestyler.

Sessions run in isolated sandboxes, and Labelbox says the platform runs with zero data retention at every LLM provider. Its listed compliance program covers SOC 2 Type II, GDPR, CCPA, NIST 800-171, and PCI DSS. Pricing is set per agent session hour, combining compute and model spend in a single estimate, with existing Labelbox enterprise customers receiving a one-time $100 credit.

The most persuasive evidence may be Labelbox's own internal use. The company reports running 51 Enterprise AI Agents across 11 teams, spanning software engineering, reliability, security, machine-learning research, evals, product, finance, sales, strategy, and procurement. In one internal example, a pull-request review agent ran 39 graded reviews on a single agent version and model; across the 33 reviews that passed the grader, median cost fell from $13.15 for the first 11 reviews to $8.90 for the last 11, a 32% reduction, while review time held at about 27 minutes and the agent's memory was updated 15 times. Under the product's memory mechanics, once five new session reports exist, and at most once an hour, a curator agent reads them and updates the agent's memory, and customers can browse what an agent has learned and delete lessons that no longer hold.

Horizon: the reinforcement-learning engine under frontier labs

The other half of the funding thesis is Horizon, which provides reinforcement-learning environments and evaluation infrastructure for frontier AI labs. According to the announcement, Horizon's WorldSim system builds simulated companies, complete with code, dashboards, spreadsheets, decks, and documents filled with realistic business data, in which a model attempts a difficult multi-step task end to end while every step is recorded as a trajectory. Deterministic verifiers score each trajectory identically on every run, and when two trajectories are both correct, an expert's preference teaches the model what Sharma describes as taste.

The company describes operating at a scale of thousands of simulated worlds and millions of graded trajectories, tuned to a lab's pass@k targets across coding, computer use, research, multimodal, and voice work. The Horizon product line covers autonomous AI research, agent coding and software engineering, multimodal knowledge work, voice with agentic tool use, computer use, and cybersecurity, reported by AI Brief.

Labelbox says it serves more than 400 customers, including over 90% of leading U.S. AI labs, with a customer base spanning frontier labs, hyperscalers, startups, and Fortune 500 companies. The company reached profitability in 2026 while growing more than fivefold year over year, a rare combination in the agent-infrastructure cohort. Founded in 2018, Labelbox describes its path toward Enterprise AI Agents in three chapters: training-data software for people building AI; reinforcement-learning environments built after the ChatGPT moment, when the most capable models began improving by practicing real work; and Enterprise AI Agents, which the company built for itself first before opening the system to customers.

Why this round matters for the agent ecosystem

The Labelbox Series E sits inside a broader re-rating of agent infrastructure. In the past few weeks alone, Pluto Security emerged from stealth with roughly $42 million to secure shadow AI inside companies, and Reco added $55 million to govern how Enterprise AI Agents connect to business systems. Each raise targets a different layer of the same problem: Enterprise AI Agents are getting real permissions inside real companies, and every layer, from training to security to governance, needs purpose-built tooling for Enterprise AI Agents.

Labelbox's bet is that the highest-leverage layer is the improvement loop itself. The company's framing is that frontier labs and enterprises alike will converge on AI systems that get better with every run, whether the run is a graded trajectory in a simulated world or a pull-request review inside a company's own codebase. With $1.25 billion of valuation and more than $100 million of fresh capital behind that thesis, the Enterprise AI Agents built on Labelbox's stack will be worth watching, as the company's announcement put it, to see whether agents that learn on the job can close the gap between demo and dependable production software.

The open questions are familiar ones: how quickly enterprise customers move from pilot to production, how grader-based quality control holds up for Enterprise AI Agents outside the vendor's own dogfooding, and whether per-session-hour pricing stays attractive as model costs keep falling. For now, Labelbox arrives at its Series E with the two assets most agent startups lack: profitable growth and 51 of its own agents already doing the work. It is the same enterprise-adoption thesis behind recent agent rollouts like Hiver Omni's agentic support platform.