UpShaqo
Intelligence desk
Agent Trust & Security Source-backed analysis

Nous Research Wants Your Enterprise Data on an Open-Source Agent

A $90 million round and a $1.5 billion valuation give Nous Research credibility to sell AI agents to businesses. The open-source pedigree that built its reputation is also the thing buyers need to interrogate hardest.

UpShaqo Editorial IntelligenceOctober 7, 20266 min read
Intelligence standard

Independent UpShaqo analysis built from fresh, attributed sources. We explain the impact instead of repeating the announcement.

Read for leverage: focus on the workflow change, the customer problem, and the next action—not only the product announcement.

A procurement team evaluating Hermes for Businesses this quarter faces a narrower question than "is this a good product." It's this: do you hand multi-step workflows and private data to a three-year-old startup whose entire reputation was built on a free, endlessly cloned developer tool, just as that company pivots to selling the same technology as a paid enterprise product?

That's the decision Nous Research is now asking founders, ops leaders, and IT buyers to make. The company has confirmed a $90 million Series B at a $1.5 billion valuation, and it's using the capital to launch Hermes for Businesses, a version of its open-source Hermes Agent aimed at companies that want customized agents handling multi-step tasks without sending sensitive data through someone else's black box.

What's actually confirmed

Strip away the narrative and the hard facts are these, all drawn from TechCrunch's reporting: the round was led by Robot Ventures, with Nvidia, Union Square Ventures, Menlo Ventures, Samsung, and 1789 Capital participating, bringing Nous Research's total funding to $158 million. The open-source Hermes Agent has been cloned more than 24 million times and, by the startup's own estimate, drives roughly 2.5% of global AI token usage. Revenue was approximately $36 million annualized by mid-September 2026, with the company projecting it will cross $100 million before year-end, a figure originally reported by The Wall Street Journal and cited in TechCrunch's coverage.

That's a real trajectory. It's also, notably, a trajectory built almost entirely on individual developers and hobbyists adopting a free tool — not on enterprise contracts, security audits, or compliance track records. Hermes for Businesses is the first real test of whether that grassroots momentum converts into something a general counsel or CISO would sign off on.

The gap between "private and secure" and proof of it

Nous Research's pitch for the enterprise product centers on letting companies deploy customized agents while "keeping their data private and secure," per the reporting. That phrase is doing a lot of work. The research packet contains no detail on what architecture, certifications, data residency guarantees, or audit trail enables that claim. There's no mention of SOC 2, ISO 27001, HIPAA readiness, or any third-party security assessment.

This is a point worth separating clearly: the valuation, the funding syndicate, and the revenue numbers are confirmed facts. The specific technical and contractual mechanisms that would make an enterprise comfortable running agents on customer or financial data are not described anywhere in the available reporting. Any buyer-facing claim about security posture should be treated, for now, as marketing language awaiting independent verification — not as settled fact.

Why the open-source lineage cuts both ways

Hermes Agent's open-source status is the reason it got cloned 24 million times and reportedly accounts for 2.5% of global AI token usage — numbers that, again, come from the company itself rather than an independent measurement firm. That adoption curve is genuinely impressive and gives Nous Research something most enterprise AI vendors lack: a massive, organic developer base that already trusts the core model.

But open weights and open code also mean the underlying agent architecture has been inspected, forked, and modified by anyone who wanted to. For a hobbyist building a personal automation script, that transparency is a feature. For a business considering Hermes for Businesses to run workflows touching customer records or financial data, the relevant question shifts: is the enterprise layer — the access controls, the data isolation, the audit logging — equally open to scrutiny, or is it a proprietary wrapper around an open core that buyers can't actually inspect? The source material doesn't answer this, and it's the single most important unresolved question for any serious evaluation.

Reading the investor roster

The backer list is worth a moment of analysis, clearly separated from fact. Nvidia's participation signals infrastructure-level confidence — a chipmaker doesn't typically invest in agent startups it thinks will burn compute inefficiently. Union Square Ventures and Menlo Ventures are established enterprise and consumer software investors, which suggests institutional diligence happened. Samsung's involvement hints at possible hardware or device integration ambitions, though nothing in the reporting confirms a specific product partnership.

1789 Capital's participation, notable because Donald Trump Jr. is a partner there, is a fact worth naming plainly — it's disclosed in the reporting — but it is not evidence of political favoritism, regulatory advantage, or any specific business outcome. Readers should treat it as a disclosed data point about the cap table, not as a predictor of how Hermes for Businesses will perform or how it will be regulated. Drawing a causal story from that single line item would be speculation this brief won't indulge in.

A concrete scenario: what adoption actually looks like

Imagine a 200-person logistics company considering Hermes for Businesses to automate vendor-onboarding paperwork — a genuinely plausible multi-step workflow use case. The upside is real: an open-source-rooted agent with a large, battle-tested developer community behind it, backed by investors with deep enterprise software experience, at a fraction of the cost of building a comparable system in-house.

The exposure is also real. The company has no publicly reported enterprise security certifications in this packet, no named enterprise customers, and a business model that only recently pivoted from free developer tooling to paid deployment. A logistics company handing vendor banking details to this system today would be an early adopter in the truest sense — betting on a trajectory, not a track record.

Controls before you sign anything

For leaders actually considering Hermes for Businesses, a few concrete steps separate due diligence from hope:

  • Request documentation of any third-party security audit or certification before piloting with real customer or financial data — and treat the absence of one as a material finding, not an oversight.
  • Ask specifically whether the enterprise layer (access controls, logging, data isolation) is open-source and inspectable, or proprietary and opaque, since the base Hermes Agent's openness doesn't automatically extend to the business product.
  • Pilot with synthetic or low-sensitivity workflows first, given the product's newness, and scale only after seeing how the agent handles edge cases and failure modes in your own environment.
  • Clarify data residency and retention terms in writing, since "private and secure" is a marketing claim until it's a contractual one.

What remains unanswered

The reporting confirms the money, the valuation, and the revenue trajectory. It does not confirm named enterprise customers, specific compliance credentials, incident history, or how Nous Research's open-source roots will be reconciled with enterprise-grade confidentiality requirements. Those gaps aren't disqualifying — plenty of fast-growing infrastructure companies closed them after their first big round. But they are the specific, unresolved questions any buyer should be asking before Hermes for Businesses touches anything that matters.

Sources

#Nous Research#Hermes Agent#enterprise AI agents#open-source AI#vendor due diligence#AI funding

Two doors. Pick one.

Hire the team.
Or become it.