Open Source AI Agent Maker Nous Research Targets Business Users
Nous Research, maker of the free Hermes agent, closed a $90 million Series B that values the company at $1.5 billion and plans to launch AI agents for business users, TechCrunch reported on October 7. Robot Ventures led the round, and Samsung, Nvidia and the venture firms Menlo and Union Square also joined.
Here is the short version for founders: a free, open-source AI agent is turning into a paid business product. If you are deciding whether to build on open tools or pay for a closed platform, this is a live case study. It also fits the wave of AI agents for business that investors keep funding.
The Numbers Behind Nous Research
The three-year-old company has brought in $158 million overall. The company estimates that developers have copied Hermes over 24 million times, and that the agent drives roughly 2.5 percent of worldwide AI token consumption. The Wall Street Journal reported annualized revenue of roughly $36 million by mid-September, with a goal of passing $100 million by the end of 2026.
| Metric | Reported figure |
|---|---|
| Series B | $90 million |
| Valuation | $1.5 billion |
| Total raised | $158 million |
| Annualized revenue, mid-September | About $36 million |
| 2026 revenue target | More than $100 million |
From Free Download to Paid Product
The new money will fund Hermes for Businesses. The idea is to let companies run customized agents for complex, multi-stage tasks without exposing their data. In plain terms, Nous is building a commercial layer on top of a popular free project.
That path is familiar in software. A community adopts the free tool, and the company sells the support, hosting and controls that larger customers need. The risk is that the free version competes with the paid one, so the paid layer must be clearly better.
What This Means for Teams Weighing Open Tools
Open tools can lower cost and reduce lock-in, but they also shift responsibility to you. Somebody on your team must handle updates, monitoring and security. Closed platforms charge more because they take on that work.
Before you commit, map your needs. Ask how sensitive your data is, how fast you need support and how much engineering time you can spare. Reliability matters most once an agent runs real tasks, which is why AI agent infrastructure keeps attracting funding.
A Simple Checklist for Piloting a Business Agent
Pick one workflow that is repetitive and low risk, such as sorting inbound leads or drafting weekly reports. Define what success looks like in numbers, such as hours saved or errors avoided. Then run the pilot for two to four weeks before you widen it.
Next, set permissions on purpose. Give the agent access only to the data it needs, log what it does and keep a human approval step for anything that touches money or customers. The NIST AI Risk Management Framework offers a free structure for thinking through these controls.
Guardrails That Protect Your Company
Privacy claims deserve testing. Ask vendors where data is stored, who can read it and whether it trains any models. Get the answers in writing, and review them again each time terms change.
Security belongs in the plan from day one. Funding for the control side of the market is growing, as the story on AI agent security shows. A small team can borrow the same ideas at a smaller scale.
Budgeting for AI Without Surprises
Agent costs can creep up quietly. Usage-based pricing means a busy week can produce a bigger bill than a quiet one, so set alerts and monthly caps from the start. Track cost per completed task, not just total spend, to see whether the tool is really paying off.
Compare that figure with the cost of doing the same job by hand. If an agent saves two hours of a teammate’s week, translate that into dollars and see whether it clears the subscription and any setup time. In addition, include the cost of review, since human checks are part of the real price.
Finally, plan an exit. Keep your prompts, workflows and data in formats you control, so you can switch vendors if prices rise or quality slips. Flexibility is cheap to build early and expensive to add later.
Keep a written log of every vendor decision and the reason behind it. When your team grows, that record saves debate, and it helps you revisit choices calmly as the tools change.
What to Watch as the Agent Race Heats Up
Watch whether Nous reaches its revenue target, since that will show if open-source popularity converts into paying customers. Watch pricing, too. If open projects keep lowering the cost of capable agents, your vendors will have to justify their fees.
For now, the practical move is to experiment cheaply and keep your options open. Test one open option and one closed option on the same task, then compare cost, speed and results before you sign anything long-term.
