Is Perseus (YC F25 codebase search startup) profitable?

No revenue, customer, or profitability data has been disclosed for Perseus, the YC F25 'Search in Latent Spaces' lab. As of mid-2026 there are no named customer logos and no confirmed funding beyond YC's accelerator investment, per the company's own YC and product pages.

Could Anthropic or OpenAI replace Perseus's codebase search product?

Yes, structurally. Perseus retrieves code context for coding agents, but Claude Code and Cursor already ship native real-time codebase indexing as of mid-2026, and MCP standardization (97 million plus installs as of May 2026) is commoditizing the exact integration layer Perseus relies on.

Who are Perseus's biggest competitors in codebase search for AI agents?

Greptile, which has named customers including PostHog, Raycast, Vouch, and Podium; Sourcebot, marketed as a cheaper Sourcegraph alternative; and native features from Anthropic's Claude Code, OpenAI, and GitHub Copilot, all as of mid-2026 per company sites.

Has Perseus raised funding?

No independently confirmed funding round has been found for Perseus beyond its YC Fall 2025 batch acceleration investment. Unlike its same-named YC batchmate Perseus Defense, which closed a $6 million seed in September 2025, no equivalent disclosure exists for this Perseus as of mid-2026.

Is Perseus the same company as Perseus Defense?

No. Perseus (YC F25, 'Search in Latent Spaces') is a codebase-search lab for AI coding agents, while Perseus Defense (YC S25) is an unrelated Austin, Texas anti-drone startup that closed a $6 million seed round on September 18, 2025.

Does Perseus have a moat against AI labs building the same feature?

Little evidence of one. Perseus's core asset is an unverified 'iterative branching search' retrieval algorithm; a related open-source GitHub project (Perseus Computing LLC, as of July 2026) is MIT-licensed, meaning any lab, including Anthropic or OpenAI, could fork the approach freely.

YC Fall 2025 · India

Perseus

Early-stage startup building fast codebase search for AI coding agents.

Losing its protection

Assessed 27 July 2026 · Evidence as of 27 July 2026

The structural protection is eroding.

On the recordSince 27 July 2026 · Checkable from mid-2026 · Every call on the record Checkpoint: Window to pivot toward regulated, air-gapped enterprise customers or stay generic middleware

The call

Perseus has a working search tool but no confirmed customers or funding, and the labs it depends on can build the same feature into their own agents at any time.

Perseus makes software that helps AI coding assistants find the right parts of your codebase to look at. The problem is that OpenAI, Anthropic, and GitHub have already built this exact feature into their own products, and they have no reason to stop improving it since better code-finding makes their assistants work better. The one opening for Perseus is companies that cannot use cloud AI tools because of rules or security locks; if Perseus focuses only on those locked-down customers and builds case studies in heavily regulated industries, it could survive. If it keeps trying to sell itself as a faster add-on for Claude or ChatGPT, it will lose to the companies that own those products.

Why

AI pressure: Anthropic Claude Code and OpenAI Codex-style coding agents. Frontier labs already ship native repo-wide search inside their own agents, so better retrieval just makes their products stronger, not Perseus's. Could AI do the core job? Yes, likely.

Bigger players: GitHub Copilot Workspace (Microsoft). Copilot Workspace already offers built-in codebase context, cutting the reason a developer would add a third-party search layer. Could a giant copy it? Yes, easily.

What protects it. None confirmed yet: the only real protection would be locking in air-gapped, compliance-barred enterprises the big labs cannot serve, and Perseus has not shown it can do that.

What would change this call

Perseus announces seed funding or names paying customers within 12 months of Demo Day. Shows whether Perseus can convert YC status into real capital or traction.
Claude Code or Copilot Workspace ships native repo-wide search matching Perseus's speed claims. Confirms frontier labs are commoditizing Perseus's core retrieval feature.
Perseus publishes case studies with air-gapped or regulated-sector (defense, fintech, healthcare) clients. Would show a pivot to the only niche not already served by Anthropic or GitHub.

Key numbers

Claims 94% fewer prompt tokens used in search (2026-07-15)
Claims 37ms median search latency at 10,000 docs (2026-07-15)
YC F25 Demo Day held (2025-12-03)

Named risks

No confirmed funding, customers, or revenue beyond YC accelerator investment as of mid-2026. Severity: high
Core retrieval feature can be replicated natively by Anthropic, OpenAI, or GitHub with zero business conflict. Severity: high
MCP standardization at 97 million plus installs commoditizes the integration layer Perseus depends on. Severity: high
Related open-source repo is MIT licensed, allowing free forking by any competitor or lab. Severity: medium
Persistent brand confusion with unrelated, better-funded Perseus Defense creates SEO and reputational risk. Severity: low

Rivals, as read in July 2026

Rival	Direction	Evidence
Greptile	losing to	Has named paying customers (PostHog, Raycast, Vouch, Podium) while Perseus has none disclosed as of mid-2026
Anthropic (Claude Code)	losing to	Ships native real-time codebase indexing as of mid-2026, overlapping Perseus's core product
OpenAI (Codex-style agents)	losing to	Iterating native codebase search weekly per April 2026 product update, no P&L conflict in improving retrieval
GitHub/Microsoft (Copilot Workspace)	losing to	Copilot for Xcode already supports @workspace codebase search natively as of April 2025
Sourcebot	contesting	Positions directly against Sourcegraph with self-hosted cross-repo context, competing niche as of 2026

Questions people ask

Is Perseus (YC F25 codebase search startup) profitable?: No revenue, customer, or profitability data has been disclosed for Perseus, the YC F25 'Search in Latent Spaces' lab. As of mid-2026 there are no named customer logos and no confirmed funding beyond YC's accelerator investment, per the company's own YC and product pages.
Could Anthropic or OpenAI replace Perseus's codebase search product?: Yes, structurally. Perseus retrieves code context for coding agents, but Claude Code and Cursor already ship native real-time codebase indexing as of mid-2026, and MCP standardization (97 million plus installs as of May 2026) is commoditizing the exact integration layer Perseus relies on.
Who are Perseus's biggest competitors in codebase search for AI agents?: Greptile, which has named customers including PostHog, Raycast, Vouch, and Podium; Sourcebot, marketed as a cheaper Sourcegraph alternative; and native features from Anthropic's Claude Code, OpenAI, and GitHub Copilot, all as of mid-2026 per company sites.
Has Perseus raised funding?: No independently confirmed funding round has been found for Perseus beyond its YC Fall 2025 batch acceleration investment. Unlike its same-named YC batchmate Perseus Defense, which closed a $6 million seed in September 2025, no equivalent disclosure exists for this Perseus as of mid-2026.
Is Perseus the same company as Perseus Defense?: No. Perseus (YC F25, 'Search in Latent Spaces') is a codebase-search lab for AI coding agents, while Perseus Defense (YC S25) is an unrelated Austin, Texas anti-drone startup that closed a $6 million seed round on September 18, 2025.
Does Perseus have a moat against AI labs building the same feature?: Little evidence of one. Perseus's core asset is an unverified 'iterative branching search' retrieval algorithm; a related open-source GitHub project (Perseus Computing LLC, as of July 2026) is MIT-licensed, meaning any lab, including Anthropic or OpenAI, could fork the approach freely.

Timeline

2026-07-26YC lists Perseus among 16 funded search startups in its industry directory
2026-07-15Perseus Computing LLC GitHub repo publishes vendor-claimed benchmarks (94% fewer tokens, 37ms P50 search)
2026-05-22Industry newsletter reports MCP protocol surpasses 97 million installs, commoditizing context layer
2025-12-03YC F25 Demo Day occurs; Perseus enters seed fundraising window
2025-11-20YC F25 batch breakdown published, confirming Perseus's batch membership
2025-09-18Same-named but unrelated Perseus Defense (YC S25) closes $6M seed round

Revision history

27 July 2026

First assessed. Losing its protection. The text above is fixed as of this date and is not edited afterwards.

In the same band

Nexus Terrain Metorial