Si-Mind is a sovereign decision substrate. Capture your life — voice, photo, text — through infrastructure you own. Ask hard questions of it. Get decision-grade answers synthesized adversarially across multiple models, with provenance back to the moment you captured it.
Nobody else can do this. Because nobody else has your data. And we don't keep it either.
I.The loop
The moat is the loop, not a feature. Markdown is exportable. Structural capital is not.
Capture lands in your private graph. The synthesis surface reads from the graph and reasons over it across multiple models simultaneously, each citing back to the captures that informed it. The decision goes back into the graph. The next question is smarter — about you, with you.
ChatGPT can't do this; it has no graph of your life. Anytype can't do this; it has no adversarial synthesis. We have both. The switching cost rises with every captured thought.
Si-Mind is one room in a house. The house is Sicelium.
One substrate, multiple altitudes. Self-host or use a managed surface; the interface is identical — only the chrome and what's editable differ.
i.Si-Brainthe substrateMarkdown plus property graph plus vector retrieval. Your hardware, your keys, your data. Always present, never empty.
ii.Si-Prismthe synthesisThree frontier models read the graph together and disagree well. The synthesis is what survives the disagreement. Single-model confidence is the trap.
iii.Si-Tunnelthe transportEnd-to-end encrypted device-to-substrate. Encryption happens in the open. You watch your secret become ciphertext; the friction is the trust.
III.What we won't compromise
Trust is a property of the substrate, not a claim we make.
Sovereign by default
Your data lives on infrastructure you control. The company doesn't see your captures. Self-host or use a hosted instance — your call, never ours.
Provenance is load-bearing
Every claim Si-Mind makes traces back to the capture that produced it. No floating context. No "the AI said so." Receipts always.
Adversarial synthesis, not single-model trust
Three models argue over your data. The synthesis is what survives the disagreement. Single-model confidence is the trap.
Eval-instrumented, radically transparent
Retrieval quality is measured weekly and published. If it doesn't improve, the work that week reverts. No vibes. No marketing language for capabilities we can't measure.
Open core, dual-licensed
The substrate is AGPL — copyable, inspectable, forkable. Commercial license for proprietary use. No hostage software, no surprise paywalls.
IV.Where we are
Substrate quality is measured against twenty of the founder's actual queries. Floor today, target 4.2 at ship.
Personal eval · faithfulness
Tracks weekly. Targets ≥4.0 by month 3, ≥4.2 at launch. If the score doesn't move, the substrate work for that period reverts.
2026-05-26 · baseline established
3.30/5
v0 is in build. Seventeen-week plan, shipping end of September 2026. iOS Companion + Si-Tunnel transport + Si-Brain substrate + Si-Prism synthesis, fused into one product. AGPL release at launch. Private TestFlight beta from week 13.
V.The pledge
Here is what we commit to. And what we will not do.
We will
Sign every model call cryptographically. Every claim ours, every claim yours.
Hold your data on infrastructure you control. Always.
Pass model pricing through transparently — what we pay, plus a margin we name.
Publish substrate quality metrics weekly.
Open-source the harness under AGPL.
We will not
Train on your data. Not for "improvements." Not for "research." Not ever.
Lock you in. Markdown out, graph out, identity keys yours.
Hide model costs behind opaque credit systems.
Hold your keys hostage. Lose password + recovery code = unrecoverable by design.
Ship a feature we cannot measure.
VI.Want in
Drop your email. One note at launch.
No marketing — just a note when the AGPL release lands and TestFlight invitations open. Maybe one more if there's something genuinely worth telling you.
No tracking. Email used once. Unsubscribe is delete.