An intent-native database paradigm
Intent doesn't live inside your app.
Most systems remember the session. Few preserve the purpose behind it. in10tDB is intent-native infrastructure for storing, resolving, and explaining purpose as a durable, queryable object — across platforms, models, and time.
- Anchor
- preserve user purpose
- Top strand
- complete blocked task
- Tension
- 0.78 — rising
- Confidence
- 0.71
Apps store what happened. Models predict what comes next. Platforms track who you are today. None of them hold why you showed up — and none carry that forward when you move on.
01 — The problem
Intent is human. Not a session variable.
Your intent when you open a health app doesn't reset when you close it. Your purpose in a conversation doesn't restart when you switch models. The motive behind a weeks-long project doesn't expire when your browser session does.
But every system you use treats it that way — inferring your why fresh each time, from scratch, with no memory of what came before and no continuity across what comes next.
Intent belongs to the person. It should live somewhere that isn't the app, isn't the model, and isn't the platform of engagement. Somewhere that persists, compounds, and stays honest about what it knows and what it's inferring.
02 — What in10t.org offers
Three properties that make intent computable.
in10tDB is not a label store or a scoring system. It formalizes intent through three structural properties that do not exist in any existing database paradigm.
Intent is not a point-in-time label. It is a thread that runs through signals, sessions, systems, and time.
in10tDB holds that thread. When a person moves from one app to the next, one model to another, their purpose doesn't reset — it accumulates, updates, and stays legible to any system that has been granted access to read it.
Coherence is what makes intent useful across a workflow. Continuity is what makes it trustworthy over time.
Intent often exists before it is written down. A person acts with purpose before they file a ticket, write a prompt, or state a requirement.
in10tDB captures and structures the signal that precedes specification — inferring motive from behavior, resolving it into a structured form, and making it available to downstream systems before a human has had to put it into words.
This is not prediction. It is structured synthesis from evidence — with provenance attached to every inference.
Intent weakens. A purpose declared six months ago has less authority than one active right now. A motive supported by recent signals should outweigh one that hasn't been reinforced in weeks.
in10tDB models this mathematically. Every strand of intent carries a decay function — so the system always knows not just what a person intends, but how confident that reading is given the recency and weight of the evidence behind it.
Stale intent doesn't disappear. It becomes appropriately uncertain — which is honest, and useful.
03 — How it works
From raw behavior to computable motive.
in10tDB formalizes intent as a structured object — not a label, not a score, not a segment. Something you can query, traverse, and hold accountable across time.
Signals enter cleanly
Clicks, prompts, telemetry, transactions, agent calls — normalized into a stable envelope. The runtime keeps raw events separate from derived motive, so noisy inputs do not silently become truth.
Intent resolves from evidence
The runtime computes a live distribution of active motives: competing strands, confidence levels, decay, friction, tension. A structured output with provenance — not a guess.
Actions keep their why
Every execution links through an Intent → Action → Outcome chain. When the session ends, the motive doesn't disappear with it.
Anchors hold the line
Declared policies and human mandates are pinned as Anchors — stable roots the system checks against before any consequential action proceeds.
04 — The runtime
Math first. Language models second.
The hot path is deterministic. Intent resolves through structured computation — fast, auditable, and tractable. Language models enter only when the math surfaces something it cannot explain: high drift, contradiction, or an effort-versus-outcome paradox.
When they do, their output is tagged as hypothesis — not ground truth. It enters the graph like any other strand, subject to the same decay and the same scrutiny.
resolve()current intent distributionexplain()why-chain traversalproject()future drift simulationalign()anchor-to-outcome audit05 — Who it's for
Built for people who can't afford opacity.
Intent-native infrastructure matters wherever a system acts on behalf of a human and needs to show — not just claim — that it understood the purpose behind it.
06 — Where it matters
For systems where decisions cannot be opaque.
07 — Early access
Interested in intent-native infrastructure?
We are building in10tDB and having early conversations with people working on agents, adaptive systems, and AI governance. Share a note. We read every one.