OpenAI’s former sales leader joins VC firm Acrew: OpenAI taught her where startups can build a ‘moat’

OpenAI’s first-ever sales leader, Aliisa Rosenthal, is stepping into a new role in venture capital. Rosenthal has joined Acrew Capital as a general partner, where she will work closely with founding partner Lauren Kolodny and the firm’s broader leadership team, Rosenthal and Kolodny confirmed to TechCrunch.

Rosenthal exited OpenAI roughly eight months ago, closing out an intense three-year chapter at the AI lab. During her tenure, the company launched a string of significant products, including DALL·E, ChatGPT, ChatGPT Enterprise, Sora, and several others.

“I wasn’t initially looking to join a VC fund,” Rosenthal told TechCrunch. After leaving OpenAI, she spent time meeting founders and evaluating a wide range of AI startups. However, after scaling OpenAI’s enterprise sales organization from just two people into a team numbering in the hundreds, she began to see venture capital as a way to extend her impact. Kolodny’s pitch ultimately resonated: instead of advising a single company on its go-to-market strategy, Rosenthal could help shape the growth of many startups at once.

Reflecting on her time at OpenAI, Rosenthal said the experience gave her deep insight into buyer behaviour and enterprise decision-making. “I learned a lot about behaviour, both on the side of the buyers, how people are thinking about these purchases, and the gap between what most organizations believe is possible and what they can actually deploy today,” she explained.

That perspective also gave her a clear view into one of the most pressing questions for AI founders: how to build a defensible moat in a market where large model providers can quickly introduce competing products.

Will OpenAI build everything and crowd out startups? Rosenthal doesn’t think so. “They’re already doing a lot — they’re in consumer, they’re in enterprise, they’re building a device,” she said. “I don’t think they are going to go after every possible enterprise application.” One of the strongest defences for startups, in her view, is specialization: focusing intensely on specific enterprise needs rather than trying to be broadly applicable.

Context as a competitive advantage

Beyond specialization, Rosenthal believes that “context” will emerge as a robust and durable moat for AI companies. Context, in this sense, refers to the information an AI system retains and uses within its context window as it processes tasks and requests.

“Context is dynamic. It’s adaptable. It’s scalable,” she said. She sees the industry moving past basic Retrieval-Augmented Generation (RAG) — which has been the standard approach since around 2025 for reducing hallucinations by grounding large language models in trusted data sources — toward something more advanced. That next step, she suggests, is a persistent “context graph” that continuously evolves as systems interact with data and users.

There is still significant technical work ahead to make this vision a reality, particularly around long-term memory and reasoning that goes beyond pattern matching. “I expect real innovation here,” Rosenthal said. “I think this year we will see new approaches around the idea of context and memory.”

She also believes that enterprise applications that embed and manage this context directly will gain a meaningful edge. “Ultimately, when we talk about moat, I think whoever owns and manages the context layer will have a major advantage in AI products,” she said.

Another opportunity Rosenthal sees lies outside the race for the most powerful, most expensive models. Rather than relying exclusively on state-of-the-art offerings from major AI labs, she believes there is room in the market for more affordable, lighter-weight models. These systems may not top benchmark leaderboards, but they can still deliver substantial real-world value while innovating on inference costs.

“I think there’s space for cheaper models that focus on efficiency,” she said, noting that many enterprise use cases do not require the absolute cutting edge in model performance to be effective.

As an investor, Rosenthal says her primary interest is in the application layer. “Where I’m really excited to invest is on the application layer,” she said. “I want to see which durable applications are built on top of all these different models — not just the foundational ones.” She is particularly interested in startups with compelling use cases that help enterprise employees work faster, smarter, and more efficiently.

In terms of sourcing deals, Rosenthal plans to tap into OpenAI’s expanding alum network. Now that the organization is a decade old, many former employees have gone on to found or join startups that have raised significant funding at high valuations. That list includes OpenAI’s prominent rival Anthropic as well as younger, high-profile companies such as Safe Superintelligence.

There is also a growing trend of former OpenAI leaders moving into early-stage investing. About a year ago, Peter Deng, OpenAI’s former head of consumer products, joined Felicis as a general partner. Deng has since been involved in major early deals, including investments in LMArena and Periodic Labs.

“I actually spoke with Peter a few months ago, and that conversation helped me make the decision,” Rosenthal said of her move into venture capital.

Rosenthal may also bring a unique advantage to founders seeking funding. Beyond her experience inside OpenAI, she has extensive relationships with enterprise buyers — the very customers and beta users that early-stage AI startups need to validate and refine their products.

Many enterprises, she believes, still underestimate just how transformative AI can be. “There’s a huge gap,” Rosenthal said, “and I’m very optimistic that it can be filled. That gap represents a huge green field for new applications and new companies.”