Is Moss profitable in 2026?

No profitability or revenue figures have been disclosed. Moss (InferEdge Inc.) self-reports 6 enterprise design partners and 3 paying customers with roughly 100% week-over-week usage growth as of its YC page, but no independent financial data or audited revenue exists as of July 2026.

Who is Moss's biggest competitor?

LanceDB is Moss's closest architectural peer, an embedded Rust-based vector database also targeting edge/on-device use, per a third-party comparison dated 2026-05-17. Pinecone remains the dominant managed vector database incumbent as of March 2026, while Qdrant, Chroma, and pgvector also compete for the same niche.

Could AI or platform vendors replace Moss?

Yes, this is Moss's core structural risk: LiveKit and Pipecat, the exact voice-AI platforms Moss integrates with, could bundle native low-latency retrieval and disintermediate Moss's distribution. No such feature has shipped yet as of July 2026, but foundation labs are also absorbing memory/retrieval into base models.

Has Moss raised a Series A?

No. As of 2026-07-10, Moss has no priced Series A or later round and no disclosed valuation in independent sources. Moss went through Y Combinator's Fall 2025 batch and was still actively hiring and fundraising-adjacent as of mid-2026 job postings.

How many employees does Moss have?

Moss's team size is listed as 5 people as of a 2026-07-10 job posting on Y Combinator's site, though the company is actively hiring for backend, SDK, and founding design engineer roles to grow from that base.

What does Moss actually build?

Moss builds an embeddable, Rust/WebAssembly retrieval runtime for real-time semantic search inside conversational AI agents, positioned as an alternative to hosted vector databases like Pinecone. It offers SDKs for Python, TypeScript, Swift, and Android, with integrations for Pipecat and LiveKit as of June 2026.

YC Fall 2025 · United States

Moss

Moss builds an embedded Rust/WASM retrieval engine for real-time voice and agent search.

Protection under test

Assessed 27 July 2026 · Evidence as of 27 July 2026

Something worth protecting exists and is being tested right now.

On the recordSince 27 July 2026 · Checkable from 12-18 months from mid-2026 · Every call on the record Checkpoint: LiveKit or Pipecat shipping native retrieval, or Moss closing a priced Series A with audited revenue

The call

Moss has working technology and early integrations, but sits in a category where the platforms it plugs into could build the same thing themselves.

Moss builds a search tool that helps AI voice agents find information quickly. The problem: similar tools like LanceDB already exist, Pinecone is the market leader, and the platforms Moss works with (LiveKit and Pipecat) could easily add this feature themselves without needing Moss. The biggest risk is that LiveKit or Pipecat decide to bundle search directly into their own products, which would cut off Moss from customers. Also worth watching: the company's reported numbers (5 million voice minutes and 380,000 installs) jumped dramatically in two weeks, raising questions about whether those figures are real.

Why

AI pressure: OpenAI, Anthropic and agent SDK makers adding native memory/retrieval. As foundation models and agent frameworks grow context windows and add built-in memory, the retrieval layer Moss sells could get absorbed for free. Could AI do the core job? Yes, likely.

Bigger players: LiveKit and Pipecat, the voice agent platforms Moss integrates with. Both platforms could bundle native low-latency retrieval directly, cutting Moss out of the exact pipelines it depends on for distribution. Could a giant copy it? Yes, easily.

What protects it. Fast embedded Rust/WASM runtime and existing SDK integrations, but no lock-in, data moat, or certification yet backs this up.

What would change this call

LiveKit or Pipecat ships native bundled retrieval/memory in their platform SDK. Would cut Moss out of its main distribution channel entirely.
Independent audit confirms or debunks Moss's 5M voice minutes and 380K installs figures. Tests whether Moss's traction claims are real or inflated.
Moss announces a priced Series A with disclosed, audited revenue figures. Would show real investor confidence beyond self-reported YC metrics.

Key numbers

5-person team (2026-07-10)
3 paying customers, 6 design partners claimed (n.d.)
~100% week-over-week growth claimed, unverified (n.d.)

Named risks

Vendor-claimed metrics escalated implausibly within two weeks (5M voice minutes, 380K installs). Severity: high
LiveKit or Pipecat could bundle native retrieval, cutting off Moss's primary distribution channel. Severity: high
No proprietary data, lock-in, or regulatory moat identified in a saturated vector-search category. Severity: high
No confirmed priced funding round beyond an unverified aggregator-reported seed. Severity: medium
Five-person team must scale engineering, sales, and compliance simultaneously. Severity: medium

Rivals, as read in July 2026

Rival	Direction	Evidence
LanceDB	contesting	Identical embedded Rust/edge architecture already shipped, per comparison dated 2026-05-17
Pinecone	losing to	Remains dominant managed vector database default per third-party comparisons as of 2026-03-08
LiveKit / Pipecat	contesting	Moss depends on these for distribution but they could bundle native retrieval, risk flagged 2026-04-28
pgvector	losing to	Cited as pragmatic default for teams under 10M vectors on Postgres, per 2026-04-28 comparison
Qdrant	contesting	Rust-based open-source competitor cited as fastest OSS option in 2026-04-28 comparison

Questions people ask

Is Moss profitable in 2026?: No profitability or revenue figures have been disclosed. Moss (InferEdge Inc.) self-reports 6 enterprise design partners and 3 paying customers with roughly 100% week-over-week usage growth as of its YC page, but no independent financial data or audited revenue exists as of July 2026.
Who is Moss's biggest competitor?: LanceDB is Moss's closest architectural peer, an embedded Rust-based vector database also targeting edge/on-device use, per a third-party comparison dated 2026-05-17. Pinecone remains the dominant managed vector database incumbent as of March 2026, while Qdrant, Chroma, and pgvector also compete for the same niche.
Could AI or platform vendors replace Moss?: Yes, this is Moss's core structural risk: LiveKit and Pipecat, the exact voice-AI platforms Moss integrates with, could bundle native low-latency retrieval and disintermediate Moss's distribution. No such feature has shipped yet as of July 2026, but foundation labs are also absorbing memory/retrieval into base models.
Has Moss raised a Series A?: No. As of 2026-07-10, Moss has no priced Series A or later round and no disclosed valuation in independent sources. Moss went through Y Combinator's Fall 2025 batch and was still actively hiring and fundraising-adjacent as of mid-2026 job postings.
How many employees does Moss have?: Moss's team size is listed as 5 people as of a 2026-07-10 job posting on Y Combinator's site, though the company is actively hiring for backend, SDK, and founding design engineer roles to grow from that base.
What does Moss actually build?: Moss builds an embeddable, Rust/WebAssembly retrieval runtime for real-time semantic search inside conversational AI agents, positioned as an alternative to hosted vector databases like Pinecone. It offers SDKs for Python, TypeScript, Swift, and Android, with integrations for Pipecat and LiveKit as of June 2026.

Timeline

2026-07-10Job posting claims 5M+ voice minutes and 380K+ installs, up from prior figures
2026-07-08VoiceAISpace profile confirms legal entity InferEdge Inc, San Francisco HQ, founded 2024
2026-06-28GitHub usemoss/moss-samples shows Pipecat, LiveKit, Agora, Vercel AI SDK integrations
2026-06-24Job posting claims 1M+ voice minutes and 350K+ installs, two weeks before higher figures
2025-11Unconfirmed aggregator report of a seed round from Pioneer Fund (unverified)
2025-10-30Launch YC page published; YC Fall 2025 batch, partner Pete Koomen
2024Moss founded as InferEdge Inc

Revision history

27 July 2026

First assessed. Protection under test. The text above is fixed as of this date and is not edited afterwards.

In the same band

Deeptrace Lightberry Unsiloed AI