Provenance Engine

Document how
people think.
Not what they submit.

ProveNotes is a provenance platform that makes thinking visible and verifiable. Cryptographic hashing, timestamped capture, and AI-scaffolded questioning create a tamper-evident trail of the process behind any piece of work.

See ReFraim Explore the Engine
09:14:22 Text
I think the main argument in Frankenstein isn't about technology being dangerous. It's about what happens when creators refuse responsibility for what they make.
SHA a7c3...f912 · SES-0412 · 2025-03-10T09:14:22Z
Explore mode
Who exactly is the "creator" here? And what does responsibility look like when the creation has its own voice?
09:17:05 Text
Shelley wrote the creature as articulate and self-aware. Maybe the argument is that creation always exceeds the creator's intent.
SHA e4b1...83d2 · SES-0412 · 2025-03-10T09:17:05Z
09:22:41 Voice
[00:47 voice memo transcribed]
SHA 7fa2...1bc9 · SES-0412 · 2025-03-10T09:22:41Z
What ProveNotes Is

A provenance engine
for authentic work.

ProveNotes captures the process of thinking, not just the output. Every keystroke, voice memo, and photograph gets a cryptographic hash and timestamp at creation. The trail is tamper-evident by design.

AI thinking modes scaffold the process with questions that widen, sharpen, or ground the work. Intentional friction (no-paste enforcement, inactivity questioning) creates genuine engagement rather than optimizing for speed.

The result: a verifiable record of how someone arrived at their ideas. Integrity becomes a side effect of working, not an enforcement problem.

Stack Next.js 14, Supabase, TypeScript
Hashing SHA-256, per-entry at creation
Input modes Text, voice, photo
AI layer Claude Haiku (scaffolding only)
Thinking modes Explore, Focus, Build
Friction model No-paste, inactivity questioning
Pricing Pay what you want. No feature gates.
Entity 404 AM Studios LLC
Core Capabilities

What the engine does.

SHA-256 Provenance
Every entry hashed at the moment of creation. Tamper-evident. The hash proves the content existed in that form at that time.
Intentional Friction
No-paste enforcement and inactivity questioning. The constraints create engagement. Friction is the feature, not a limitation.
AI Thinking Modes
Explore, Focus, Build. Three scaffolding operations grounded in self-regulated learning research. Questions, not answers.
Multi-Modal Capture
Text, voice recordings, photographs. All timestamped and hashed at the moment of creation. One trail, multiple input types.
Session Timelines
Continuous session tracking. When they started, paused, were questioned, pushed through, and stopped. The temporal shape of thinking.
Notebook Architecture
Composition notebook metaphor with marble-generated covers. The metaphor shapes the relationship to the work. Familiar, contained, personal.
Scaffolding Layer

Three thinking modes.

Grounded in self-regulated learning, metacognition, and Project Zero thinking routines. Each mode scaffolds a distinct cognitive operation. Students choose the one that fits what their work needs.

Explore
Widen the thinking. Generate questions, follow threads, shift perspective. Elaboration and inquiry.
"What haven't you said yet?"
Focus
Sharpen the thinking. Tighten claims, check sources, demand precision. Organization and evidence.
"What's the actual claim, in one sentence?"
Build
Ground the thinking. Notice patterns, find throughlines, decide next steps. Reflection and transfer.
"What's the next concrete move?"
Built on ProveNotes

ReFraim: reframing AI
in education.

The first product built on the ProveNotes engine. ReFraim gives schools a way to document student thinking instead of policing AI use. Students work in a focused workspace. Teachers see the process trail. Integrity becomes visible.

Visit ReFraim →
AI isn't the problem. Invisibility is. Make student thinking visible, verifiable, and valuable.
Focused workspace with intentional friction
Three thinking modes: Explore, Focus, Build
Cryptographic provenance on every entry
Teacher-visible process trails
Free pilot for schools. No feature gates.
Architecture Principles

Privacy is a consequence,
not a feature.

ProveNotes uses API-level access to external models. No training on prompts, no profile building, no retention on the provider side. The query goes out. The response comes back. Everything else belongs to the user.

Student entries and promptsEncrypted at rest
Model provider data retentionNone (API-level)
Model training on promptsDisabled
Provenance trailUser-owned, exportable
Teacher visibilityProcess only, not content
Third-party data sharingNone

Education is the first vertical.

The provenance problem exists everywhere people need to prove their process: research, creative work, legal documentation, professional development. ProveNotes is the engine. The applications are just beginning.

The trail is
the proof.

See how ProveNotes powers visible thinking in schools.

Visit ReFraim Get in Touch