Perplexity SPACE Runtime Launches: 2026 AI Agent News
Perplexity has launched SPACE, a runtime for AI agent tasks, according to TestingCatalog. If you use AI for research or daily work, this explainer breaks down what SPACE is, why it matters, and how to try Perplexity's agent features today.
📰 What Happened: Perplexity Ships a Runtime Called SPACE
TestingCatalog, an outlet that tracks unreleased and newly shipped AI features, reported that Perplexity has launched a runtime called SPACE for AI agent tasks. In plain terms, a runtime is the environment where an AI agent actually does its work: the place where it can hold files, run multi-step jobs, and keep going after you close the tab.
The name is a clear evolution of Perplexity Spaces, the feature that lets you group research into project folders with custom instructions and uploaded files. Until now, Spaces were mostly a way to organize conversations. A runtime turns that organizational layer into a working layer, where the AI does not just answer questions but executes tasks inside your project.
This fits a pattern Perplexity has been building all year. The company already shipped Comet, its agentic browser that can click through websites on your behalf, and Tasks, which run searches and reports on a schedule. SPACE looks like the connective tissue: a home base where agent work runs, persists, and produces output you can come back to.
One honest caveat: this is early reporting from TestingCatalog, and Perplexity often rolls features out gradually, sometimes to Pro and Max subscribers first. Exact capabilities and availability may shift by the time it reaches your account, so treat the details as a snapshot, not a spec sheet.
What Does "Agent Runtime" Actually Mean?
Think of a chatbot as a phone call: it only works while you are on the line. A runtime is more like hiring a contractor with a workshop. You hand over a job, the agent works in its own environment with the tools and files it needs, and it comes back with a finished result. Anthropic's Claude Code, OpenAI's agent tools, and Google's Gemini 3 agent features all rely on the same core idea. Perplexity giving its agents a dedicated runtime signals it wants to compete on completed work, not just fast answers.
💼 Why It Matters for Solopreneurs and Knowledge Workers
If you run a one-person business, the practical difference between a chatbot and an agent runtime is time. A chatbot answers one question per prompt, and you assemble the pieces yourself. An agent with a runtime can take a whole job, such as "research these five competitors and build me a comparison document," and deliver the finished artifact while you do something else.
Perplexity's angle is search. Its agents sit on top of a live research engine, which makes them naturally suited to work that starts with finding accurate, current information: market research, lead lists, price monitoring, content research, and weekly industry digests. For a solopreneur, those are exactly the tasks that eat afternoons but do not directly earn money.
There is also a bigger market signal here. In 2026, every major AI company is racing from "answers" to "outcomes." OpenAI, Anthropic with Claude, and Google with Gemini 3 have all pushed hard into agents that complete tasks end to end. When Perplexity, best known as an AI search engine, builds dedicated agent infrastructure, it confirms where the entire industry believes the value is: delegation, not conversation. Learning to delegate work to AI agents is quickly becoming a baseline skill, the way learning search was in the 2000s.
The Shift From Asking to Assigning
The habit change is subtle but important. Instead of asking "What are the top email tools for small teams?", you assign: "Compare the top five email tools for a two-person agency, check current pricing, and give me a recommendation with a table." Runtimes exist to make that second kind of request reliable. The people who benefit most will be the ones who practice writing task briefs, not just questions.
🧩 How SPACE Fits Into Perplexity's Agent Stack
Perplexity's agent story now has several layers, and it helps to see where a runtime slots in. Each existing piece solves one part of the delegation problem, and SPACE appears to tie them together.
Spaces gave you a persistent project: files, custom instructions, and a focused set of conversations. Comet gave the AI hands, letting it browse and act on real websites. Tasks gave it a clock, running work on a schedule without you present. A runtime gives it a workbench, a place where multi-step jobs execute and results accumulate.
For context, Perplexity also lets subscribers choose among frontier models such as Claude, GPT-5 class models, Gemini 3, and its own Sonar models for search. That matters because agent tasks lean heavily on model quality: a better reasoning model inside the same runtime produces noticeably better finished work.
| Feature | What It Does | Best For |
|---|---|---|
| Spaces | Project folders with files and custom instructions | Organizing ongoing research by client or topic |
| Comet browser | Agentic browsing that navigates sites for you | Web tasks like form filling and comparison shopping |
| Tasks | Scheduled searches and recurring reports | Weekly digests, monitoring, routine research |
| SPACE runtime | Execution environment for multi-step agent jobs | Delegating whole tasks that produce finished output |
🚀 How to Try It Today: Practical Steps
You may not see SPACE in your account immediately, since Perplexity typically rolls new capabilities out in stages, often starting with paid tiers. Perplexity Pro runs 20 dollars per month, with a higher Max tier for heavy users. The good news is that you can prepare today with features that already exist, so that when the runtime reaches you, your workflow is ready.
Start by creating a Space for one real project you are working on right now. Upload the relevant files, write custom instructions describing your business and preferred output style, and run your research inside it for a week. This is exactly the context an agent runtime will draw on.
Then practice one delegation-style request per day. Take a task you would normally do in five browser tabs and write it as a single complete brief with a clear deliverable. Compare what the AI produces against what you would have made. This calibrates your sense of what is safe to delegate, which is the real skill the SPACE era rewards.
- ✔Check perplexity.ai and the TestingCatalog article for SPACE availability on your plan
- ✔Create one Space with real files and custom instructions for a current project
- ✔Write one full task brief per day: goal, sources to check, and the exact deliverable format
- ✔Try a scheduled Task, such as a weekly competitor or industry digest
- ✔Verify agent output against sources before using it in client work
- ✔Follow TestingCatalog or Perplexity's changelog to catch the wider rollout
🔭 What to Watch Next: Open Questions
Early reports leave several questions open, and the answers will determine how useful SPACE becomes for non-developers. Watch for three things as the rollout widens.
First, availability and limits. Which plans get the runtime, how many concurrent agent tasks can you run, and how long can a job execute? Runtime economics usually show up as usage caps, and those caps decide whether this is a toy or a workhorse.
Second, integrations. An agent runtime becomes dramatically more valuable when it can touch your actual tools: email, calendars, spreadsheets, and cloud drives. Perplexity's competitors are all racing to add these connections, so expect pressure here.
Third, reliability and safety controls. Agents that act on your behalf need clear logs of what they did, confirmation steps before anything irreversible, and easy ways to review sources. How Perplexity handles this will matter more than any single feature, especially for people using it in client-facing work.
❓ Frequently Asked Questions
What is the Perplexity SPACE runtime?
Based on TestingCatalog's reporting, SPACE is a runtime, meaning an execution environment where Perplexity's AI agents can carry out multi-step tasks. Instead of only answering questions in a chat, agents get a persistent workspace to run jobs and produce finished output. It builds on Perplexity's existing Spaces feature for organizing research projects.
Is SPACE free, or do I need Perplexity Pro?
Perplexity has not published full availability details, and early features often reach paid tiers first. Perplexity Pro costs 20 dollars per month, and the Max tier serves heavier usage. Check your account settings and Perplexity's official announcements for current access on your plan, since rollouts frequently expand over weeks.
How is this different from ChatGPT or Claude agents?
The core idea is the same: give the AI an environment to complete whole tasks rather than single answers. Perplexity's differentiator is that its agents sit on top of a live search engine with citations, which suits research-heavy work. OpenAI, Anthropic's Claude, and Google's Gemini 3 offer stronger ecosystems in other areas, such as coding or office document integration.
Do I need to be a developer to use an agent runtime?
No. Runtimes like SPACE are infrastructure that runs behind a normal chat interface. Your job is simply to write clearer requests: state the goal, the sources or constraints, and the deliverable you want back. If you can brief a freelancer, you can brief an agent.
🏁 Final Thoughts
Perplexity's SPACE runtime is another clear signal that 2026's AI race is about delegation, not conversation. The company that made AI search mainstream is now giving its agents a persistent place to work, joining OpenAI, Anthropic, and Google in the push toward AI that finishes tasks. You do not need to wait for full rollout to benefit: set up a Space, practice writing complete task briefs, and build the habit of reviewing agent output. Those skills transfer to every platform. If you found this explainer useful, subscribe to Agents at Work for plain-English breakdowns of AI agent news, and drop a comment with the first task you would hand to an agent.
Last updated: July 18, 2026 · Keyword: Perplexity SPACE runtime · Agents at Work
Comments
Post a Comment