Open source · EU-jurisdictional · Post-quantum ready

Search died. AI memory is next. This time, you own it.

Every major AI model is building memory about you. Most extract value and tell you nothing. Oursly is the layer where YOUR memory belongs to YOU — with the observability, revocation, and equity rights every model owes you.

Reserve early access See the observability gap
Open source on GitHub Matrix protocol Zero-knowledge encryption Post-quantum ready (ML-KEM-768, ML-DSA) EU-jurisdictional
The framing

The AI memory you don't control.

Search moved from a page you see to an API you don't. AI memory is going the same way — silently, behind every chatbot, agent, and summary. The pipeline that decides what's remembered about you operates without your consent, your visibility, or your equity. Oursly inverts the pattern.

Extraction model · typical AI memory

You feed the model. The model feeds the platform.

  1. Model trained on your data without per-event disclosure.
  2. Model summarizes you to other users, agents, and search surfaces.
  3. You never see what was remembered, cited, or said about you.
  4. Value flows to the platform. The bill flows to you.

Default for every major hosted assistant in 2026.

Oursly model · observability + ownership

You hold the memory. You see the citations. You earn the upside.

  1. YOUR memory store, on YOUR device, with YOUR keys.
  2. YOU control what enters the graph and what gets revoked.
  3. YOU see what every model says about you — Citation Watch, Q3 2026.
  4. YOU receive equity in the platform you helped build.

Default for Oursly, in writing, in the ledger.

Why it matters now

Regulation just acknowledged the asymmetry.

The EU AI Act Article 50 (applicability 8/2/26) is the first regulation to require agentic AI to disclose itself to the humans it interacts with. That is the platform-side mirror. Oursly is the consumer-side mirror — disclosure of what's been remembered about you, by whom, and what got cited downstream.

If the model has to disclose, you should be able to verify.

As the trade press put it this spring: infrastructure that provides structured access to AI isn't building AGI — it's building observability. TechSpot, 6 May 2026.

The wedge

Your AI should remember you. Not someone else's servers.

Personal memory is the substrate every useful AI assistant runs on. Today that memory either lives on the AI vendor's servers (OpenAI, Anthropic, Google) or on third-party memory layers (Mem0, Graphiti, LangMem) that hold the graph and rent it back to you. Oursly Memory flips it: the graph runs on your device, the keys stay on your device, and you carry your memory between assistants.

Federated personal knowledge graph

A typed graph of entities, relations, and time-anchored events that any AI assistant can query through a standard interface. Local-first; syncs across your own devices over Matrix; never crosses an inference boundary unless you authorize it.

Post-quantum-ready encryption Live · verify →

Standardized on the NIST post-quantum cryptography winners: ML-KEM-768 (FIPS 203) for key encapsulation and ML-DSA (FIPS 204) for digital signatures. Defends against harvest-now-decrypt-later threats years before the cryptographically relevant quantum computer arrives. Our live oursly ML-KEM-768 + ML-DSA-65 public keys and signed genesis attestation are at /crypto.

You hold the keys

Zero-knowledge architecture. The server cannot read your memory graph, your stored files, or your sync state. If we are subpoenaed, the data we have to hand over is ciphertext. The only people who can decrypt it are you and anyone you have explicitly authorized.

  • Carry your memory between Claude, ChatGPT, local Llama, or whatever you switch to next
  • Export-by-default - your graph is yours in a portable, signed format
  • Audit-by-default - every read and write is logged in an append-only journal you control
  • Disability and privacy law friendly - GDPR Article 17 (erasure), HIPAA-amenable, EU data residency
CapabilityOursly MemoryMem0 / Graphiti / hosted memory
Graph stored on your device Yes No
You hold the encryption keys Yes No
Post-quantum encryption ML-KEM-768, ML-DSA RSA / ECC only
Portable across AI assistants Standard API Per-vendor SDK
EU-jurisdictional Frankfurt + your device US default
User earns equity in the platform Equity ledger No
The model

Equity for users, not extraction.

Every action you take on Oursly - storage you contribute, memory you grow, friends you onboard - earns equity tokens in a transparent ledger. When the platform grows, you own a piece of it. Not points. Not "credits." A line in a verifiable ledger that maps to actual platform equity at the appropriate liquidity event.

Oursly Equity Ledger · preview mockup
Live as of Q3 2026 target
User Usage units Equity tokens earned Total pool %
@elena.r 12,840 128.4 0.0042%
@kenji 8,610 86.1 0.0028%
@aisha.k 6,295 62.9 0.0020%
you 0 0 0.0000%
@mateo 4,180 41.8 0.0013%

Mockup illustrative. Token-to-equity mapping, vesting curve, and ledger contract finalize Q3 2026. Cap-table allocation reserved for users: 30% of fully diluted equity, allocated proportional to usage and contribution. No payment required to earn.

Every action you take earns equity in the platform.

Register interest to be first
Coming next

One wedge first. The rest follows.

Memory and storage ship first because they are the foundation - every other surface depends on them. Email, search, voice, and video are next. They are not the lead. They are the backlog you get for free because the foundation is right.

Coming Q3 2026

Citation Watch

What does ChatGPT, Claude, Perplexity, and Gemini actually say about you?

  • Periodic prompts to the major models — "Who is [your name]?" / "What do you know about [your brand]?" / "Cite [your domain]"
  • Verbatim model responses delivered to your inbox weekly
  • Diff against last week — see what each model added, dropped, or changed
Roadmap · Q3 2026 · waitlist counts toward equity ledger
Coming next

Encrypted email

Your own domain, your own keys. No scanning, no ads, no tracking. Plugs into your memory graph so your assistant can summarize your inbox without an outside server reading it.

Coming next

Private metasearch

Federated search that does not track you. Results pulled into your memory graph so your assistant remembers what you researched without anyone else watching.

Coming next

Voice and video

End-to-end encrypted calls on Matrix-compatible rails. Optional transcript ingestion into your memory graph - on your device, on your terms.

Coming next

Chat

Matrix-protocol chat that shares your encrypted memory graph with your assistant. Your conversations help your assistant learn you without the assistant vendor reading the chat.

Simple, fair pricing.

Memory and storage on day one. Equity tokens at every tier. EU-jurisdictional by default.

Free

$0/month
  • 5GB encrypted storage
  • 10K memory-graph nodes
  • Post-quantum-ready encryption
  • Equity tokens accrue
  • Community support
Reserve free tier
Most Popular

Pro

$9/month
  • 100GB encrypted storage
  • 500K memory-graph nodes
  • Cross-device sync (Matrix)
  • Audit journal export
  • Priority support
  • Higher equity-token multiplier
Reserve Pro tier

Team

$29/month
  • 500GB encrypted storage
  • Unlimited memory-graph nodes
  • Shared graphs with team policies
  • Admin controls + audit logs
  • Dedicated support
  • Team equity-token allocation
Contact us

Reserve early access.

Be among the first owners. We email you when Oursly Memory opens to early users. Equity tokens accrue from day one.

No spam. No tracking pixels. Your address is hashed and rate-limited.