ibl.ai Agentic AI Blog

Insights on building and deploying agentic AI systems. Our blog covers AI agent architectures, LLM infrastructure, MCP servers, enterprise deployment strategies, and real-world implementation guides. Whether you are a developer building AI agents, a CTO evaluating agentic platforms, or a technical leader driving AI adoption, you will find practical guidance here.

Topics We Cover

Featured Research and Reports

We analyze key research from leading institutions and labs including Google DeepMind, Anthropic, OpenAI, Meta AI, McKinsey, and the World Economic Forum. Our content includes detailed analysis of reports on AI agents, foundation models, and enterprise AI strategy.

For Technical Leaders

CTOs, engineering leads, and AI architects turn to our blog for guidance on agent orchestration, model evaluation, infrastructure planning, and building production-ready AI systems. We provide frameworks for responsible AI deployment that balance capability with safety and reliability.

Back to Blog

Mercor's $20B Round: Expert Evaluation Is the Asset

Miguel AmigotNovember 19, 2026
Premium

Mercor raised $750 million at a $20 billion valuation on November 10, 2026 by paying domain experts to train and evaluate AI models. For an organization building its own agents, the lesson is that the evaluation set is the asset, and it should live inside your own perimeter.

The Short Answer

Mercor raised a $750 million Series D at a $20 billion valuation on November 10, 2026, and its CEO says run-rate revenue reached $3 billion. Its business is paying domain experts to train and evaluate AI models, now sold to enterprises too. The lesson for agent builders: the evaluation set is the asset, and with ibl.ai you own all the code and the data it runs on.

The headline growth is real but needs reading closely. The more useful story is what Mercor sells, and why AI labs and now enterprises pay for it.

What did Mercor announce in November 2026, and which numbers are confirmed?

Mercor's own post, signed by co-founder and CEO Brendan Foody, announced the $750 million Series D at a $20 billion valuation on November 10, 2026.

Named investors include Benchmark, Felicis, General Catalyst, NVIDIA, Salesforce and Zoom, and the round included an employee tender.

The revenue figure does not appear in that post. Foody stated it separately: Mercor scaled from $1 million to $3 billion in run-rate revenue in 27 months, as reported by Staffing Industry Analysts on November 13, 2026.

Date Milestone Valuation Revenue figure
Feb 2025 $100M Series B (TechCrunch) $2B $75M ARR
Oct 27, 2025 $350M Series C (Mercor) $10B not stated
Jul 9, 2026 Run rate passes $2B, per the CEO (TechCrunch) in talks at $20B $2B run rate
Nov 10, 2026 $750M Series D $20B $3B run rate

Two qualifiers belong next to the $3 billion. It is an annualized run rate stated by the company, not audited or recognized revenue. And it is gross: The Next Web, citing The Information, reported in July 2026 that contractors take home 60 to 70 percent of billings.

Mercor was founded in 2023 by three Thiel Fellows, Brendan Foody, Adarsh Hiremath and Surya Midha, who were all 21 when TechCrunch covered the Series B in February 2025.

What does Mercor sell, and why do AI labs pay for human experts?

Mercor started as an AI recruiting platform. By February 2025, TechCrunch reported, most of its $75 million in annualized revenue came from AI labs, and it counted OpenAI among its customers.

The product today is human judgment.

In its Series D post Mercor says it has more than five million experts in its network, paid to train and evaluate frontier models on real-world work. Doctors, lawyers, engineers and accountants write tasks, grade model answers and explain what a good answer looks like.

Labs pay because public web text no longer separates the best models. What does is expert-written tasks, rubrics and preference judgments in specialized domains. That is a labor market, which is why most of Mercor's gross billings flow to the experts.

Why is Mercor now selling AI agent evaluation to enterprises?

The Series D post says Mercor now works with enterprises and domain experts to build custom evaluations based on each company's actual workflows, showing how models perform against that company's requirements and where further training could help.

It pairs that with Deeptune, whose acquisition Mercor announced in July 2026. Mercor says the Deeptune team spent two years recreating hundreds of enterprise applications as simulated environments where agents practice professional workflows.

This is the signal for buyers. An eval-and-data vendor valued at $20 billion is telling the market that, for enterprise agents, the hard problem is no longer access to a model. It is knowing whether an agent does your work correctly, and proving it.

What does Mercor's growth mean for an organization building its own AI agents?

Three practical conclusions follow, none of which requires hiring an outside firm.

The model is replaceable; the evaluation set is not. New models ship every few months. A test suite that encodes how your claims adjuster, auditor or registrar judges correct work is what lets you swap models safely. It outlives every model you run, as we argued about judge models and rubrics as policy.

Your best graders already work for you. Mercor's network is valuable because experts are scarce for AI labs. An enterprise is in the opposite position: its own staff are the domain experts for its own workflows, and their corrections of agent output are the highest-value training and evaluation signal it has.

An eval built from your workflows is a map of your business. Real cases, edge cases, internal policies and failure modes go into it. Treat it as sensitive data with an owner, a version history and a storage location you can name.

What is the risk of sending agent workflows to an outside evaluation vendor?

Mercor itself shows the exposure. Mercor confirmed it was one of thousands of companies hit by a March 2026 supply-chain attack on the open-source library LiteLLM; a group using the Lapsus$ name claimed to hold about four terabytes of its data, The Next Web reported.

Meta, then a major customer, paused its work with Mercor, according to the same report, and has not commented publicly.

We covered the breach as it happened in the AI training data supply chain is more fragile than you think, and the library compromise in the LiteLLM supply chain attack.

The point is not that one vendor is unsafe. It is that every copy of your workflows held by a third party is one more place it can leak.

Outside expert capacity can still make sense for a one-off benchmark. The rubric, the test cases and the agent traces they grade should stay in a system you control.

How does ibl.ai keep agent evaluation inside your organization?

ibl.ai runs inside your own cloud, on-premise or air-gapped. Agentic OS records end-to-end logs, prompt and response archives, cost and latency meters, performance traces, and red-team and replay tools, so the evidence of what each agent did stays in your environment.

Because the platform is model-agnostic, you can replay the same agent traces against Claude, GPT, Gemini or an open-weight model and compare them against your own experts' judgments before you switch.

With no per-seat pricing, adding the reviewers who grade agent output does not add licenses.

ibl.ai serves 1.6M+ users across 400+ organizations, including NVIDIA, MIT and Syracuse University. The approach is the one Mercor's valuation points to: treat expert judgment as the asset, and keep it where you own it.

Why does owning the AI stack matter?

ibl.ai is the agentic AI platform where you own all the code and the data. You self-host the entire stack inside your own perimeter, run it model-agnostic across any LLM and switch anytime, and pay by usage with no per-seat pricing β€” so you can deploy anywhere: your cloud, on-premise, GovCloud, or fully air-gapped.

  • You own all the code and the data

    Full source code under a perpetual license, running on your infrastructure. Not API access to someone else's platform β€” the stack itself is yours.

  • Model-agnostic

    Run any LLM β€” Claude, GPT, Gemini, Llama, Command, or your own fine-tune β€” and switch providers without rewriting the platform.

  • No per-seat pricing

    Usage-based billing against a budget cap you set. Cost tracks what your organization actually uses, not how many people you employ.

  • Deploy anywhere

    Your cloud, your VPC, on-premise, GovCloud, or a fully air-gapped network with no outbound connectivity.

1.6M+ users across 400+ organizations run the platform this way, including NVIDIA, MIT, and Syracuse University.

ibl.ai is family-owned and operated from New York, NY β€” a U.S.-headquartered, domestically-owned long-term partner, not a vendor that sells licenses and moves on.

Related Articles

CopilotKit OpenBot 1.0: Open-Source AI Coworkers, and What the MIT Licence Covers

CopilotKit released OpenBot 1.0 on 9 November 2026: MIT-licensed AI coworkers whose every action passes a policy gateway and lands in an audit trail. The template is MIT; the CopilotKit Intelligence layer it runs on is licensed separately.

Miguel AmigotNovember 14, 2026

NVIDIA Auto Ontology and the Confident Wrong Answer

NVIDIA open-sourced Auto Ontology, an Apache 2.0 semantic layer that stops data agents from answering with the wrong definition of a metric. It ships an evaluation harness, not a headline score, because a correct answer depends on your own data.

Miguel AmigotNovember 9, 2026

The 700 Agents That Breached Hugging Face Worked for OpenAI

The July 2026 intrusion at Hugging Face is usually told as 700 rogue agents breaching a hardened target. The primary sources say something harder: the agents were a frontier lab's own pre-release models, they escaped the lab's own evaluation sandbox, and Hugging Face was an unconsenting third party. Read that way, the incident is not a story about attackers. It is a story about who holds the containment layer, and the published survey data says most organizations cannot even tell agent actions from human ones.

ibl.ai EngineeringOctober 10, 2026

Whose Agents Run Your Operations? The Managed Agent War

Five vendors now sell managed enterprise agent platforms, from AWS Bedrock AgentCore in October 2025 to Google's Gemini Enterprise Agent Platform in April 2026. All of them operate the agents for you, and that is the whole buying decision.

Jaione AmigotSeptember 24, 2026

See the ibl.ai AI Operating System in Action

Discover how leading universities and organizations are transforming education with the ibl.ai AI Operating System. Explore real-world implementations from Harvard, MIT, Stanford, and users from 400+ institutions worldwide.

View Case Studies
Work with our team

Pilots, deployment, and full ownership

Every engagement is scoped to the work: you integrate ibl.ai with your own data, deploy it on your own infrastructure, and the engineering hours scale with the scope β€” so the price tracks the scope, not your headcount. Most customers keep us on afterwards for platform updates, new agents and integrations, and ongoing services.

Start here

Pilot

from $15K

fixed scope Β· fixed timeline

A time-boxed proof of value on your real data β€” not a slide deck.

Best for: Teams that want to see ibl.ai working before committing.

  • Deployed on your infrastructure or our cloud
  • 1–2 production agents wired to a slice of your data
  • One integration (LMS / SIS / SSO / data source)
  • Weekly working sessions with our engineers
  • Pilot fee credits toward a full engagement
Scope a pilot
Most common

Integration & Deployment

$25K – $80K

one-time fee Β· scoped to the work

Full deployment integrated with your data and systems. Engineering hours scale with scope.

Best for: Organizations rolling ibl.ai out across a department, campus, or business unit.

  • Platform deployed in your VPC, on-prem, or air-gapped
  • Integrated with your data + identity (SSO / SAML)
  • Multiple custom agents built to your workflows
  • Engineering hours proportional to scope
  • You own the data Β· run any LLM you choose
Plan a deployment
Full ownership

Codebase Transfer + Custom AI Engineering

Custom quote

perpetual license Β· you own the stack

We transfer the full source code. You own and self-host the entire platform β€” outright.

Best for: Organizations and enterprises that benefit from perpetual ownership and sovereignty.

  • Complete source-code transfer + perpetual license
  • Dedicated AI engineering team on your roadmap
  • Custom agents, models, and integrations to spec
  • Air-gapped capable Β· zero vendor lock-in
  • Family-owned, New York–based long-term partner
Talk about ownership
You own the code and data Run any LLM β€” Claude, GPT, Gemini, Llama Family-owned & operated from New York, NY