Superpower Daily
OpenAI Loses Data Center Chief While Pursuing a Reported $600 B...Alibaba Raises HK$80B for Full-Stack AI as Investors Question t...LLMScholarBench Tests 22 Models and Finds an Accuracy-Represent...Meta Adds Voice-Controlled AI and Navigator Home Screen to Ques...Harvard Uses 7 AI Faculty Avatars in a $699 Founder Bootcamp, W...DeepSeek V4 Pro Lands on Fireworks With $2.50 CyberGym Solves,...OpenAI Loses Data Center Chief While Pursuing a Reported $600 B...Alibaba Raises HK$80B for Full-Stack AI as Investors Question t...LLMScholarBench Tests 22 Models and Finds an Accuracy-Represent...Meta Adds Voice-Controlled AI and Navigator Home Screen to Ques...Harvard Uses 7 AI Faculty Avatars in a $699 Founder Bootcamp, W...DeepSeek V4 Pro Lands on Fireworks With $2.50 CyberGym Solves,...
Businesspublished

Perplexity Taps NYU’s Andrew Gordon Wilson for Continual-Learning Agent Research

Wilson’s remit joins continual learning with agent collaboration as Perplexity supports work that unfolds across extended sessions. The first test will be whether that mandate becomes a defined research team and publication program.

By 3 min read
Perplexity Taps NYU’s Andrew Gordon Wilson for Continual-Learning Agent Research
Sign in to save and react

Listen to this story

The audio brief

About 1:35
0:001:35
Read transcript
Perplexity has hired New York University professor Andrew Gordon Wilson to lead research into two problems that become more important as AI agents work for longer: learning from experience, and coordinating with other agents. Wilson announced the appointment on August 24. He will keep running his NYU research group, while building a Perplexity team that plans to work openly and is already recruiting researchers in these areas. The mandate is broader than simply giving an agent a longer memory. Continual learning asks what a system should retain, how it can transfer knowledge to unfamiliar tasks, and how it can avoid losing capabilities as it changes. Agent collaboration adds another layer: deciding when agents should cooperate, delegate a task, or use another agent’s output. The timing connects that research to Perplexity’s push toward longer-running work. In July, the company introduced SPACE, an internal sandbox that lets agents run code, preserve files, and continue multi-step tasks across extended sessions. Perplexity says SPACE powers every session in its Computer product and saw millions of sandbox creations during its first full-deployment week. Those figures come from the company. Wilson has also co-authored a 2025 paper on when multi-agent coordination improves decisions, including agent selection and outsourcing. But the organization is still undefined. Perplexity has not said how many researchers Wilson will oversee or what the first projects will be. The key question is whether this becomes a small open-research effort or a larger agent-research group with a visible publication program.

Story brief

3 key points

Perplexity’s appointment of NYU professor Andrew Gordon Wilson gives its longer-running agent strategy a dedicated research lead focused on memory across tasks and coordination among agents. Wilson will keep running his NYU group, conduct the Perplexity work openly, and recruit researchers, making the effort partly academic in structure. The timing follows Perplexity’s SPACE rollout for persistent, multi-step...

  1. 01

    Wilson announced the appointment on August 24 and will retain his NYU research group.

  2. 02

    The Perplexity group is recruiting and plans to conduct its research openly.

  3. 03

    SPACE powers every Computer session and saw millions of sandbox creations in its first full-deployment week, according to Perplexity.

Perplexity has hired New York University professor Andrew Gordon Wilson to lead research on two hard problems for more persistent AI systems: learning from experience over time and coordinating work among agents. Wilson will continue running his NYU research group alongside the new role.

Wilson announced his appointment on August 24. He said his Perplexity group will focus on continual learning and collaboration between AI agents, with research conducted openly. He also invited researchers working in those areas to contact him, signaling that the group is recruiting.

One role, with an academic group retained

  • Wilson will lead research at Perplexity while continuing to run his research group at New York University.
  • The stated Perplexity agenda is continual learning and collaboration between AI agents.
  • Wilson has said the work will be open and invited prospective researchers to get in touch.

The product context is longer-running work

The appointment follows Perplexity’s move toward agents that can work across longer sessions. In July, the company introduced SPACE, an internal sandbox that lets agents run code, preserve files, and carry out multi-step work over extended sessions. Perplexity said SPACE powers every session in its Computer product and handled millions of sandbox creations in its first week of full deployment; those usage figures are company-reported.

SPACE provides an environment for work that persists beyond a single answer: code can run, files can remain available, and a task can extend through several steps. Wilson’s mandate addresses a separate research layer that becomes more consequential in such settings: how a changing system preserves useful knowledge, transfers it to new tasks, and avoids degrading abilities it already has; and how multiple agents decide whether another agent should handle part of the job.

Wilson has studied the coordination question before

The fit is clearest in Wilson’s recent work on multi-agent orchestration. He co-authored a 2025 paper examining when coordinating multiple agents improves decisions, including decisions about selecting other agents and outsourcing tasks. That is narrower than a product plan, but it directly overlaps with the questions Perplexity has assigned to his group: when agents should work together and when work should move elsewhere.

Perplexity Research already identifies search, reasoning, agents, and systems as its core areas. It is also advertising AI researcher roles in New York City, Palo Alto, and San Francisco. Wilson’s public invitation gives that existing hiring footprint a more specific research focus, while his continued NYU role preserves an active academic base alongside the company position.

The next move is organizational

The hire establishes a research lead and a stated agenda, but not yet the operating scale of the effort. Perplexity has not specified how many researchers Wilson will oversee or identified the first projects the group plans to publish. Those two details will show whether the role becomes a small open-research effort, a larger agent-research organization, or something in between.

Related coverage

Continue reading