Benchmark and Index put $60 million behind Hone's long-running business agents
Hone raised a $60 million seed round led by Benchmark and Index Ventures for AI agents that pursue business goals over weeks or months. Buyers have only Hone's own account of its safeguards to judge them by, because Hone has not disclosed revenue or customer results.
The Engineer · Build desk
What happened
- Bloomberg reported a $285 million valuation, attributed to Hone, and did not say whether the figure is pre-money or post-money.
- Cognition and AI inference startup Modal are early customers, and Hone is building go-to-market agents with Cognition.
- Hone calls its agents Engines, which are assigned an outcome, connected to company systems and given limits on what they can do.
- Benchmark partner Peter Fenton joins the board and told Bloomberg that Hone is working on safeguards against cybersecurity incidents involving agents.
Why it matters
- What it costs cost A pilot judged on lead conversion has to run on the buyer's live sales pipeline for weeks or months before the outcome metric can be read, so the buyer carries the trial's business risk.
- Who's exposed exposure The first customer results will come from Cognition, whose CEO invested in Hone personally and employed Stephan, so buyers need evidence from an unrelated customer before treating results as independent.
- What doesn't add up contradiction With the one outside benchmark pointing against models making long-horizon business calls alone, Hone's case rests more on its approval gates and staged rollout than on the model's judgment.
- Who has to decide decision Each buyer has to settle the boundaries and the list of actions needing human sign-off up front, and those settings decide how much of the outcome the agent is actually allowed to own.
Next story · Product · 2d ago
Modal counts $300 million a year from containers where AI agents run codeConfirmed6 publishers
Hone's site describes how an Engine takes on work. A customer assigns it an outcome, connects it to company systems and sets boundaries on what it can do [7]. Hone says it simulates an agent's decisions before the agent gets more responsibility. Sensitive actions need human approval, and actions are logged for review [16]. These are Hone's descriptions of its own system [16].
I think staged responsibility is the right design for this problem. It is also the part a buyer can test first: whether the approval gate fires on the actions that matter, and whether the log lets a reviewer reconstruct why the agent acted.
The difficulty is duration. A coding agent such as Cognition's Devin takes on defined software work [8]. In Bloomberg's sales example, Hone envisions an agent that goes past scoring incoming leads and works on improving lead conversion over time [9]. That objective stays open for months while conditions change, so the agent has to hold context, adjust its approach and know when a person should step in [10]. Conversion also lags. I'd expect a change to lead handling in week one to show up in closed deals weeks later, mixed in with everything else the sales team did that quarter. Runtimewire framed the buyer's problem as judging whether the agent is improving the right metric without creating new problems elsewhere [11].
Stephan has worked on a long schedule before. As a teenager he was a co-founder of Team Tumbleweed, a project in the European Space Agency's business incubator that built a prototype Mars rover powered by wind [20]. Its planned 2020 test in Israel's Negev desert was postponed because of COVID-19, and a prototype was tested there in 2021 [21]. A year lost to a pandemic is decent rehearsal for an objective whose conditions change while it is still open. He later joined Cognition as chief of staff to CEO Scott Wu [5].
Stephan said the aim is to build AI that "owns outcomes" [6]. The closest outside evidence on that ambition is a simulation. CEO-Bench, a June 2026 paper, runs a fictional startup for 500 days of pricing, marketing and budgeting decisions [17]. Its authors report that most evaluated models struggled and that every one finished below the benchmark's rule-based baseline [18]. Because it is a simulation, it bears on Hone only indirectly [19]. For its result to transfer, an Engine would have to make those calls end to end with no approval step. Hone describes gated sensitive actions and staged responsibility instead [16].
Bloomberg's figures leave the investors' stake open. If $285 million is post-money, the $60 million bought about 21% of Hone [22]. If it is pre-money, the post-money value is $345 million and the round's investors hold about 17% [23][24]. The report did not include the round's terms [14].
What to watch
- A Cognition or Modal result stated as a change in a named business metric over a fixed window, measured against a baseline.
- An independent evaluation of Hone's Engines, or a long-horizon benchmark that keeps human approval steps in the loop.
- Confirmation of whether the $285 million valuation is post-money, and the terms of the round.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence45
- Adoption15
- Hype gap+40
- Incentives65
- Confidence55
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Hone raised a $60 million seed round to build AI agents that pursue business goals over weeks or months.
- [2]
Benchmark and Index Ventures led Hone's seed round.
- [3]
The round values Hone at $285 million, according to the company as reported by Bloomberg; the valuation is attributed to Hone and Bloomberg did not label it post-money.
- [4]
Hone's early customers include Cognition and AI inference startup Modal, Bloomberg reported; Hone is working with Cognition on agents for go-to-market work.
- [5]
Hone CEO Moritz Stephan was previously chief of staff to Cognition CEO Scott Wu.
- [6]
Stephan says the aim is to build AI that "owns outcomes."
- [7]
Hone calls its persistent agents "Engines" and says they can be assigned an outcome, connected to company systems, and given boundaries for what they can do.
- [8]
Coding agents such as Cognition's Devin can take on defined software work; Stephan wants to apply agent systems to responsibilities that cut across a business.
- [9]
Bloomberg's example is sales: Hone envisions an agent that does more than score incoming leads, instead handling the work of improving lead conversion over time.
- [10]
A business objective can stay open for months while conditions change, so an agent must keep track of context, adjust its approach, and know when a person needs to step in, unlike a discrete coding or customer-service task.
- [11]
Runtimewire posed the question of how a customer can judge whether the agent is improving the right metric without creating new problems elsewhere.
- [12]
Cognition CEO Scott Wu is a personal investor in Hone.
- [13]
Benchmark partner Peter Fenton is joining Hone's board and told Bloomberg that the company is working on safeguards against cybersecurity incidents involving agents.
- [14]
The report did not include the round's terms.
- [15]
Hone has not disclosed revenue, customer outcomes, or independent performance data.
- [16]
Hone's site describes simulating an agent's decisions before it takes on more responsibility, with sensitive actions requiring human approval and actions logged for review; these are Hone's descriptions, not independently evaluated results.
- [17]
CEO-Bench, a June 2026 research paper, simulates operating a fictional startup for 500 days, including pricing, marketing, and budgeting decisions.
- [18]
The CEO-Bench authors report that most evaluated models struggled, and every model finished below the benchmark's rule-based baseline.
- [19]
Because CEO-Bench is a simulation, it does not evaluate Hone directly.
- [20]
As a teenager, Stephan co-founded Team Tumbleweed, a European Space Agency business-incubator project that developed a wind-powered Mars rover prototype.
- [21]
A planned 2020 Team Tumbleweed test in Israel's Negev desert was postponed because of COVID-19, business@school reported; a prototype was later tested there during the AMADEE-20 mission in 2021.
- [22]
If the $285 million valuation is post-money, the $60 million round bought about 21% of Hone.
- [23]
If the $285 million valuation is pre-money, the post-money valuation is $345 million.
- [24]
At a $345 million post-money valuation, the $60 million round is about 17% of Hone.
Sources
2 independent publishers whose own reporting we read for this story.
- cryptobriefing.comFormer OpenAI, Cognition staffers launch Hone to build AI agents that run business operations
1 article · October 8, 2026
- runtimewire.comHone raises $60M to build AI agents for business work that lasts months
1 article · October 8, 2026