Investor Vinod Khosla believes Wajo, a startup founded by an ex-DeepMind engineer, will win the AI agent market on trust, TechCrunch reported on October 6, 2026. The company's Fo agent can hire humans to complete a task, which makes accountability a central question for the product.
What is Wajo's Fo agent and what can it do?
The Wajo Fo agent is an AI agent that, according to TechCrunch's reporting, can hire humans to complete a task. That is the core capability the outlet highlighted, and it sets the product apart from agents that only act through software.
An AI agent is a program that takes a goal, breaks it into steps and carries those steps out with limited human direction. Most agents today work through digital tools such as browsers, email and code. They search, draft, schedule and click.
Fo reportedly goes a step further. When a task needs a person, the agent can bring one in to do the work. TechCrunch's summary is brief, so details are not public in the material this article draws on. The summary does not say how Fo finds workers, how it pays them, what kinds of tasks it handles, or how it checks the results. Readers should treat any claim about those mechanics as unconfirmed until Wajo or TechCrunch provides it.
What the reporting does establish is the direction. The agent is not just a tool a person operates. It can recruit people as part of getting a job done.
Why does Vinod Khosla think Wajo will win the agent market on trust?
Khosla's stated belief, as TechCrunch reported it, is that Wajo will win the agent market on trust rather than on raw capability. The reporting frames trust as the competitive edge he expects to matter most.
Read next Laika's Wildwood: Stop-Motion Fantasy at TIFF 2026Khosla is a well-known venture investor, and his views on AI draw attention across the technology industry. TechCrunch's headline attributes the belief to him directly. The summary available here does not include his reasoning, any direct quotes, or details of his financial stake in Wajo, so this article does not supply them.
What can be offered is context, labeled as analysis. Agents are only useful if people hand them real responsibility. A person will let software draft an email without much worry. Letting it spend money, contact strangers or commit to deadlines is a bigger step. The more consequential the task, the more a user needs to believe the agent will behave predictably and own its mistakes.
That logic suggests why an investor would see trust as a durable advantage. Features are easy to copy. A track record of reliable behavior is harder to build and slower to lose. Whether Wajo earns that record is something only time and customer experience can show.
Why does it matter that an AI agent can hire people?
It matters because an agent that hires people moves AI from advising on work to organizing it. That raises practical questions about responsibility, oversight and the experience of the people being hired.
The following points are analysis, not reported fact.
Accountability. If an agent engages a human and something goes wrong, who is responsible? The user who gave the instruction, the company that built the agent, or the worker? Clear answers matter for any business that might rely on such a product.
Oversight. A person who delegates a task to an agent may never see the hiring step. Good design would let users review who was engaged, for what, and at what cost before commitments are made. Whether Fo offers this is not stated in the source.
Worker experience. The people hired by an agent deal with a counterpart that is software. They need to know who they are working for, how they will be paid and where to turn with a dispute.
Quality control. Human work varies. An agent that hires must judge whether the result is acceptable, and that judgment is itself a risk.
None of this is unique to Wajo. Online labor marketplaces have long connected buyers with freelancers, and they have grappled with identity checks, payment protection and dispute handling. The new element is that the buyer is an AI system acting on a user's behalf. This is why Khosla's emphasis on trust, as TechCrunch reports it, fits the product's nature.
How does Wajo compare with other AI agent products?
On the single point TechCrunch highlights, Wajo's agent differs from most agent products because it can bring humans into the loop as hired workers. The source does not provide a feature-by-feature comparison, so the picture below is a general framing, labeled as analysis.
| Approach | How work gets done | Where humans appear |
|---|---|---|
| Software-only agents | Through apps, browsers and code | Mainly as the person giving instructions |
| Freelance marketplaces | A person searches for and hires workers | The buyer hires directly |
| Wajo's Fo agent (as reported by TechCrunch) | The agent can hire humans to complete a task | As workers the agent engages |
This table describes categories, not named competitors, because the source does not name any. Pricing, speed, accuracy and customer numbers for Fo are not in the reporting, so no performance comparison is possible. Readers evaluating agent products should ask each vendor how it handles errors, what approvals it requires and how it records its actions.
What are the open questions about Wajo's approach?
The biggest open questions are how Fo actually works and what safeguards surround it. TechCrunch's summary confirms the hiring capability and Khosla's belief, but little else.
Questions worth watching, framed as analysis:
- Process. How does Fo select and pay the people it hires, and what tasks is it built for?
- Controls. Can users set limits, approve hires in advance or review work afterward?
- Liability. What happens when hired work is wrong, late or harmful?
- Evidence of trust. What would count as proof that Wajo is more trustworthy than rivals, such as audits, published policies or customer results?
- Background. TechCrunch describes the founder as an ex-DeepMind engineer. That signals technical experience, though the summary does not detail the founder's work there or how it shapes Fo.
A short timeline of what is publicly reported as of October 2026:
- October 6, 2026: TechCrunch reports that Khosla believes Wajo will win the agent market on trust, and that Fo can hire humans to complete a task.
That single dated entry reflects how early the public record is.
What this means
This section is analysis, not reported fact. For readers weighing agent tools, the Wajo Fo agent is a signal that vendors are starting to compete on assurance as much as on features. Before delegating real tasks to any agent that can spend money or engage people, check who is accountable, what you can approve in advance, and what record the system keeps. Treat Khosla's trust thesis, as TechCrunch reports it, as a claim to be tested, not a conclusion. The next useful evidence will be concrete: published safeguards, named customers and how the product behaves when a hired task goes wrong.
Related coverage
Source: TechCrunch
