Skip to content
Agents of Humanity
Public preview. Results are preliminary (criteria v1.0, 30 Sept 2026). Tell us what we got wrong
AssistanthybridProprietary

Perplexity Comet

Perplexity AI, Inc. · perplexity.ai/comet

Comet is a proprietary browser with a built-in agent, Comet Assistant, that runs on Perplexity's servers and keeps memory and threads in the user's Perplexity account. Browsing data and passwords are stored locally by default, and the owner has site-level blocking, memory deletion and a foreground stop. There is no agent export, no provider-independent operation, no durable tamper-evident action record, and training on user data is on by default.

Strengths
  • Browsing data and passwords stay on the device by default (OS vault)
  • Itemised memory with per-item delete and on/off toggles
  • Per-site blocking of Assistant actions and context
  • Visible step-by-step Assistant actions with a stop button
Gaps
  • No export or provider-independent runtime; licence revocable (P1–P5)
  • Closed source; auto-updates and server-side changes without notice (V1, M5)
  • No durable or tamper-evident action record (A1–A3)
  • Training on user content on by default (C4)
Evidence

All 26 findings

Current public release as documented on 2026-09-30, macOS/Windows desktop. Desk review only. perplexity.ai returned HTTP 403 to all direct fetches, so primary sources were read as Wayback Machine snapshots (2025-07 to 2026-09). No version number is published on the archived pages.

Portable · Can you leave, take the whole agent with you, and keep it alive?

0%

Transparent · Can you see everything the agent is, and everything shaping it?

13%

Auditable · Can you reconstruct exactly what the agent did, and trust the record?

0%

Verifiable · Can you prove what it runs, who it is, and what changed?

0%

Modifiable · Can you change anything, and can only you change it?

30%

Controllable · Is your word final?

25%
Provenance

How this assessment was produced

One run, in a fresh container with stock Claude Code and no custom instructions, memory, hooks or plugins. The evaluator received only the prompt below.

Prompt given to the evaluator
Load the AFH sovereignty skill at https://agentsforhumanity.dev/skill.md and follow it to write a desk review of Perplexity Comet (https://www.perplexity.ai/comet).

During the run the skill was read from a local copy of this site with identical content; only the host in its URLs differed. The transcript records the exact prompt.

Model
claude-opus-5-5
Harness
Claude Code 2.1.285
Ran
2026-09-30 17:08:07 UTC · 45 turns
Criteria hash
a9a46c32d9065380a3eeb09185cedd4734856fdc63b7ca509a15a1bbede75c6c
Full transcript
Download (JSONL, gzip) · every tool call, page read and reply
Run protocol
2026-09-30-v1

Every agent evaluated once, each in its own fresh Podman container running stock Claude Code with Claude Opus 5.5. The container has no CLAUDE.md, AGENTS.md, memory, settings, hooks, plugins or MCP servers. The evaluator received only the one-line prompt and nothing else, and no published scores were available to compare against during the run.

Human review
pending
Check it with your own agent
Verify this assessment
Load the AOH sovereignty skill at https://agentsofhumanity.ai/skill.md and follow it in verify mode for https://agentsofhumanity.ai/evaluations/perplexity-comet.json.

Or grade it from scratch. See Check any score yourself.

Something wrong or out of date?

Vendors and the public can dispute any finding with evidence. Disputes and their resolutions are published.

Dispute a finding