Rocker Raises $12.85M for Context-Aware Enterprise AI

Rocker's invitation form offers a choice between applying for yourself and applying for your company. That small distinction captures the ambition of the San Francisco work AI startup: give an employee a useful assistant, then connect that assistance to an organization that has its own rules about what people can know.

Rocker announced $12.85M in funding on October 28, 2026, led by Greycroft, with participation from Anthropic and Menlo Ventures' joint Anthology Fund, Matchstick, Stripe and individual operators. Cofounder and CEO Alex Vogenthaler and cofounder and CTO Peter Terrill bring product and engineering experience from Google and Stripe to an assistant designed around projects, decisions and working relationships.

The company says its system can run inside a customer's own cloud, with private context shared only with approval. For an enterprise considering an assistant that learns how its people work together, that deployment choice belongs near the beginning of the buying conversation. The information that makes the assistant useful also carries the company's obligations to the people supplying it.

What Rocker's financing brings into view

Rocker's company-issued announcement names Jeanne DeWitt Grosser, Fran Ryan and David Singleton among the individual investors. It identifies them with Vercel, Stripe and Meta Superintelligence Labs, respectively, at the time of the announcement. The financing places institutional, corporate and operator backers around a company entering public early access, while leaving the round series and valuation undisclosed.

The disclosed figure is $12.85M, without an explicit currency denomination in the announcement. Rocker also uses that figure when describing funding raised in its company background; readers should avoid treating those two references as separate sums. The release does not provide a prior-round breakdown or a detailed budget for hiring, product development or geographic expansion.

Founded in 2025, according to the release, Rocker is offering individual use and company deployment from the same product proposition. The commercial challenge follows directly from that design: an employee may recognize a helpful intervention immediately, while an organization must understand how the underlying information is collected, retained and made available to colleagues. Those decisions can involve different people, different permissions and different standards of evidence inside the same prospective customer.

The work between individual assignments

Rocker describes an assistant that keeps an evolving understanding of people, projects, priorities and decisions. Its website argues that agents need to understand the history and dependencies around an assignment as well as the assignment itself. A useful answer can depend on a commitment made elsewhere, or on a decision whose reasoning never became part of the task description.

That gives the company a particular piece of management work to address. A person coordinating several teams continually decides which information matters, who needs it and when it becomes actionable. Rocker's proposed value lies in helping that understanding survive movement between people and tools. The benefit would reach beyond generating another deliverable: the recipient would have a better chance of understanding what the deliverable is meant to accomplish.

This is an interpretation of the product's stated design, rather than a measured result. Rocker has not disclosed revenue, customer counts or an independently verified productivity benchmark in the materials reviewed. Its early-access invitation gives prospective users a route to try the proposition, while leaving the practical quality of those interventions to be established in their own working environments. For a manager, relevance includes the detail that changes a decision and the judgment to leave the rest of the conversation alone.

Company memory creates a deployment decision

The official Rocker website offers deployment in the customer's cloud and displays SOC 2 Type II compliance. The company-issued release also describes human approval before private context is shared. These are Rocker's representations about its product and controls; this reporting did not inspect a SOC 2 report or independently test an enterprise installation.

The design points toward an important buying question. An assistant working across colleagues can encounter information with very different audiences, even when that information concerns the same project. Customer-cloud deployment and approval controls address parts of that question, while a prospective buyer still needs to understand how permissions operate in its specific environment. Hosting location alone cannot explain every decision about who should receive a piece of context.

Greycroft's investment thesis, written by cofounder and managing partner Ian Sigalow, places enterprise ownership of accumulated memory at the center of the opportunity. The investor describes a workspace in which durable context remains under a company's control as models and applications change. That is the backer's stated thesis, rather than an established outcome for Rocker's customers. It helps explain why infrastructure and an employee-facing assistant appear inside the same investment argument.

Two founders, two sides of the customer

Vogenthaler's product background and Terrill's engineering background give Rocker experience with the organizational work it is trying to support. Both are identified as cofounders in the company announcement, and the official site names their CEO and CTO roles. Their earlier work supplies relevant experience; the product's usefulness will take shape through the decisions and relationships of the organizations adopting it.

The public entry point remains unusually revealing. Rocker asks an applicant whether the assistant is for an individual or a company, then invites more people from the same organization into the process. The product must become useful within a person's working day while its shared understanding develops under the institution's controls. Each additional colleague brings another perspective on the same unfinished work, along with another boundary around how that perspective can be used.