TEAL
Temporal · Evolving · Agentic · Ledger. A framework for reasoning about evolving organizational knowledge over time — not a database, not just another graph.
What TEAL is
TEAL is the conceptual and technical framework that powers Rutabaga. It treats knowledge as connected, time-aware, and agent-ready — rather than as rows in a table or static documents in a search index.
Each letter is a first-class dimension of the system: time is structural (not just a timestamp), facts evolve (not just updated-in-place), agents traverse meaning (not just vector neighborhoods), and every inference is auditable back to its source.
T — Temporal
Time is a first-class dimension of the graph.
Facts in TEAL carry valid-from and valid-to windows. That means the system can answer what was true at a point in time — not just what is true now. This is distinct from simply timestamping records: the graph itself is temporally structured, so you can traverse it at any moment in history without losing what the world looked like then.
For enterprise knowledge, this is the deepest technical differentiator. Organizational reality changes. Reasoning about it without temporal structure flattens everything into a misleading present.
E — Evolving
Knowledge isn't static, and neither is the model of it.
In TEAL, new evidence can supersede old facts without deleting history. Confidence and evidence quality are part of the model, not metadata bolted on the side. As the organization learns, the graph learns — and you can see what changed, when, and why.
Facts don't decay. They get updated, traceably.
A — Agentic
Structured for how AI agents actually reason.
TEAL is designed so agents can traverse causality, context, and decision pathways — not just retrieve chunks of text. This enables GraphRAG-style reasoning grounded in the graph's structure, where every hop is a semantically meaningful relationship.
The goal is knowledge agents can reason with, not just data they can cite.
L — Ledger
Every fact has a traceable source.
The ledger dimension maintains a chain of evidence so that every insight can be audited back to its origin. For regulated industries — and for any organization that wants to trust AI-assisted conclusions — this matters more than any visualization flourish.
Ledger is the trust and auditability layer, and it's woven into the model from the start.
Grounded in a proprietary semantic relationships
TEAL's ontology isn't ad-hoc. It's rooted in anthropological frameworks for how concepts relate.
These relationships become edge types in the graph, creating semantically meaningful traversal paths rather than arbitrary labeled connections. It's a richer, more principled basis for knowledge organization than a typical property graph, and it's part of why agents can reason over a TEAL graph instead of just walking it.
v1.0
Initial formalization of the four TEAL dimensions.
v1.1
Added temporal modeling with explicit valid-time windows.
v1.2
Added confidence metrics and evidence quality as first-class properties on nodes and edges.
What TEAL is not
Positioning matters, and TEAL is easy to mistake for adjacent things it isn't.
-
Not a database
TEAL is a framework and ontological approach, not a storage system. It sits above graph infrastructure and defines how knowledge is modeled, not where it's persisted.
-
Not a BI or analytics tool
Rutabaga doesn't replace your BI and analytics. It enables reasoning about knowledge in context, why things connect, how they have changed, and where the evidence comes from.
-
Not just RAG
Traditional RAG retrieves chunks of text by similarity. TEAL enables structured graph traversal grounded in semantic relationships and temporal context, so answers stand on structure rather than resemblance.
-
Not a memory system with decay applied to facts
Many AI systems borrow Ebbinghaus-style decay from memory research and apply it to facts. That's a category error, decay belongs to episodic memory, not factual knowledge. TEAL separates the two: facts don't decay, they get superseded with traceable history.
How the pieces fit together
Rutabaga is positioned as a business UX and intelligence layer above graph infrastructure. The framework makes the layering explicit.
| Layer | Component | Role |
|---|---|---|
| Business UX | Rutabaga | Interface, visualization, user-facing AI queries |
| Temporal layer & ontology | TEAL | Time as first-class dimension, Spradley semantics, confidence and evidence |
| Storage & agent plumbing | Graph infrastructure | Graph storage, agent execution |
Want to go deeper?
Rutabaga is in private beta. If TEAL resonates with a problem you're wrestling with, we'd like to hear about it.
Get in touch