◆Portable knowledge for humans + AI
Your knowledge should
outlive every tool.
Every conversation, decision, spec, and artifact adds to what your project knows. Today, that knowledge is trapped in the tool that created it.
Intransient turns it into a portable, evidence-backed knowledge platform. Knowledge Lenses give each human or AI exactly the context it needs—without losing the source.
◆The problem
Your project knowledge is trapped in the tools that created it.
Modern work happens across many systems. Each tool holds part of the story. None gives the project a durable, portable knowledge system.
ChatGPT
GitHub
Documents
Browser
People
Changing tools shouldn’t mean rebuilding what the project knows.
◆The idea
Turn interactions into knowledge that outlives the tool.
Intransient preserves evidence, identifies potential knowledge, and lets people decide what becomes portable, durable project knowledge.
Preserve authorized interactions and artifacts as source evidence.
◆Evidence before knowledge
Knowledge without evidence is just another hallucination.
Confirmed knowledge remains connected to the exact source it came from. The summary never replaces the record.
“Given our transactional requirements and existing operational experience, PostgreSQL is the clearest default.”
- Provider
- AI conversation
- Captured
- 2026-08-10 14:32
- Identity
- sha256:a91c…84e2
Use PostgreSQL as the primary transactional database.
- Derived fromDiscussion #17
- SupersedesDecision #09
- Related specDatabase Architecture
- Implemented byArtifact / PR #142
◆Human governed
AI proposes.
Humans decide.
Captured information does not automatically become project truth. Distillation creates proposals. People confirm what deserves to become durable knowledge.
Move authentication from custom JWTs to OAuth.
◆Knowledge network
Projects aren’t folders.
They’re networks.
Navigate from a topic to its discussions, decisions, specifications, artifacts, and exact source evidence.
contains · led to · defined by · supports
Architecture
2 specs 4 artifacts
◆Core concept / Knowledge Lens
A Lens gives each AI exactly the knowledge it needs.
Knowledge Lenses are how Intransient shares knowledge without dumping an entire project history into every interaction.
A coding agent might need implementation decisions. A product agent might need requirements and historical tradeoffs. Each Lens reveals a governed, purpose-specific part of the same portable knowledge network.
◆Provider neutral
Your knowledge shouldn’t disappear when you change AI.
Models will change. Tools will change. Your project’s knowledge should move with you.
KNOWLEDGE
◆Knowledge over time
Remember not only what you decided—but how you got there.
Today’s answer is useful. Knowing why yesterday’s answer changed is often more valuable.
Use REST API
Realtime needs emerge
Event-driven updates
Realtime Architecture v2
◆Portable by design
Your knowledge should travel with your project.
Canonical knowledge should remain understandable, portable, and rebuildable—not trapped inside one tool, provider, or opaque database.
Current architecture direction: human-readable Markdown/YAML as canonical knowledge, with derived indexes that can be rebuilt.
◆Clear boundaries
What Intransient is—and isn’t.
Not shared memory. Shared knowledge—shaped through Lenses.
NOTAnother memory-sharing tool
→BUTA portable knowledge-sharing platform
NOTOne giant context dump
→BUTGoverned Knowledge Lenses
NOTA search index
→BUTNavigable, connected knowledge
NOTAI secretly deciding truth
→BUTHuman-confirmed, evidence-backed knowledge
NOTKnowledge locked to one model
→BUTProvider-neutral knowledge
NOTInvisible employee monitoring
→BUTExplicitly authorized capture
◆The future experience
Imagine every AI joining the project already understanding the project.
No architecture recap. No lost tradeoffs. No hunt for the important file. A new AI navigates the project’s portable knowledge and gets ready to work.
◆Current status / Early stage
We’re building
the foundation.
Intransient is being explored and built as a portable, local-first knowledge system: capturing authorized interactions, preserving provenance, distilling structured knowledge, and connecting it into a network that humans and different AI systems can navigate.
Some foundations are being prototyped now. The larger vision is still evolving.