◎ OS PUB Apache 2.0 ← All specifications

P205 — AIEP — Environment State Interface

Applicant: Neil Grassby Classification: Patent Application — Confidential Priority: Claims priority from GB2519711.2 filed 20 November 2025 Architecture Layer: AIEP AGI Cognition Layer — Phase 2

Framework Context

[0001] This specification operates within an AIEP environment as defined in GB2519711.2 and GB2519798.9. The present specification defines the interface through which external environment observations and sensor data are normalised, admitted, and integrated into the Phase-2 AIEP world state model.

Field of the Invention

[0002] The present invention relates to environment sensing interfaces and data normalisation architectures for evidence-bound artificial intelligence systems.

[0003] More particularly, the invention relates to a canonical interface layer that accepts observations from physical sensors, digital monitoring systems, and external data feeds and normalises them into admissible AIEP evidence artefacts suitable for ingestion by the world state graph subsystem.

Background

[0004] Evidence-bound reasoning architectures require that all world state updates originate from verified, provenance-bound evidence records. Physical sensors and external monitoring systems produce raw observations in diverse, heterogeneous formats that are not directly compatible with the AIEP evidence artefact schema.

[0005] Without a canonical normalisation interface, sensor data must be manually formatted before admission, introducing inconsistency, loss of timestamp fidelity, and potential provenance breaks. A formal environment state interface ensures that sensor-derived observations achieve the same provenance quality as document-derived evidence.

Summary of the Invention

[0006] The invention provides an Environment State Interface (ESI) that: accepts observations from registered sensor endpoints and data feed adapters; normalises observations into canonical AIEP evidence artefact format; verifies observation freshness against configurable staleness thresholds; and submits normalised artefacts to the AIEP evidence admissibility gate.

[0007] The ESI maintains a sensor registry mapping sensor identifiers to normalisation adapters. Each adapter defines the transformation from sensor-native format to the canonical AIEP observation artefact schema.

[0008] The ESI assigns each admitted observation a three-level timestamp (observation time, publication time, admission time) consistent with the TEG (P202) temporal schema.

ASCII Architecture

Physical Sensors / Data Feeds
   |         |         |
   v         v         v
+---------------------------------------+
|      Sensor Adapter Registry          |
|  Adapter_A   Adapter_B   Adapter_C   |
+-------------------+-------------------+
                    |
                    v
+---------------------------------------+
| Environment State Interface (ESI)     |
|                                       |
| - Freshness validation                |
| - Schema normalisation                |
| - Three-level timestamp assignment    |
| - Provenance hash derivation          |
+-------------------+-------------------+
                    |
                    v
        AIEP Evidence Admissibility Gate
                    |
                    v
     Causal World State Graph (P200)

Definitions

[0009] SensorAdapter: A registered transformation module that converts sensor-native or feed-native data formats into canonical AIEP evidence artefact format. Each adapter is associated with one or more sensor or data feed identifiers.

[0010] SensorRegistry: A versioned registry mapping sensor and data feed identifiers to their associated SensorAdapters and freshness configuration parameters.

[0011] ObservationArtefact: A canonical AIEP evidence artefact produced by the ESI from a raw sensor observation or data feed record. Each ObservationArtefact includes: a provenance hash derived from the raw observation content; the adapter identifier used; and a three-level timestamp record.

[0012] FreshnessThreshold: A configurable maximum interval between an observation’s world time and the current admission time. Observations exceeding the freshness threshold are rejected with a staleness record rather than admitted to the ledger.

[0013] StalenessRecord: A negative-proof record produced by the ESI when an observation is rejected for exceeding the freshness threshold, retaining evidence that the observation was received but not admitted.

Detailed Description

[0014] Sensor Registration. Before an observation source may submit observations, it must be registered in the SensorRegistry. Registration records: the sensor or feed identifier; the canonical adapter class responsible for normalisation; the expected observation frequency; and the freshness threshold applicable to this source. Registration is a governed operation requiring P215 policy evaluation.

[0015] Observation Normalisation. When a raw observation arrives from a registered source, the ESI locates the corresponding SensorAdapter in the registry. The adapter applies its defined transformation, producing an ObservationArtefact whose content conforms to the canonical AIEP observation schema. The transformation is deterministic: identical raw observations from the same adapter always produce identical ObservationArtefacts.

[0016] Freshness Validation. After normalisation, the ESI validates the observation’s world_time against the configured freshness threshold for the source. If current_admission_time − world_time > freshness_threshold, the observation is rejected and a StalenessRecord is appended to the ledger. The StalenessRecord preserves the provenance of the rejected observation, allowing governance review to identify sources producing persistently stale data.

[0017] Three-Level Timestamp Assignment. For observations that pass freshness validation, the ESI assigns the three-level timestamp: world_time from the sensor-reported or feed-embedded observation timestamp; pub_time from the transmission timestamp in the source protocol; and admission_time as the wall-clock time of ESI processing. These three timestamps are encoded in the ObservationArtefact and preserved through the TEG (P202) integration.

[0018] Provenance Hash Derivation. The ObservationArtefact provenance hash is computed as SHA-256 over: the raw observation byte sequence; the adapter identifier; and the world_time value. This hash provides a deterministic, tamper-evident content identifier independent of the admission timestamp.

[0019] AIEP Evidence Admissibility Gate. After normalisation, freshness validation, and timestamp assignment, ObservationArtefacts are submitted to the standard AIEP evidence admissibility gate. The admissibility gate applies the same schema validation and provenance verification as document-derived evidence artefacts. Observation artefacts that pass this gate are admitted to the evidence ledger and ingested by the CWSG (P200) update cycle.

Technical Effect

[0020] The invention provides a canonical normalisation pipeline ensuring that sensor-derived and data-feed-derived observations achieve the same provenance quality as document-derived evidence artefacts in the AIEP architecture. By enforcing freshness validation, three-level timestamp assignment, deterministic adapter transformation, and SHA-256 provenance hashing, the ESI prevents stale, improperly timestamped, or untraceable sensor data from contaminating the evidence-bound world state model.

Claims

  1. A computer-implemented method for normalising environment observations into an evidence-bound reasoning architecture, the method comprising:

    (a) maintaining a sensor registry mapping sensor and data feed identifiers to registered sensor adapters and per-source freshness thresholds;

    (b) receiving a raw observation from a registered source and applying the registered sensor adapter to transform the raw observation into a canonical evidence artefact format;

    (c) validating the observation’s world time against the configured freshness threshold for the source, and rejecting observations exceeding the threshold by producing a staleness record in the evidence ledger;

    (d) assigning a three-level timestamp to admitted observations comprising a world time derived from the sensor-reported observation timestamp, a publication time from the transmission protocol, and an admission time from the processing wall-clock; and

    (e) computing a provenance hash over the raw observation byte sequence, adapter identifier, and world time, and submitting the normalised observation artefact with this hash to the evidence admissibility gate.

  2. The method of claim 1, wherein the adapter transformation is deterministic such that identical raw observations from the same adapter always produce identical canonical artefacts.

  3. The method of claim 1, wherein sensor registration is a governed operation requiring evaluation by the safety constraint and governance enforcement engine before a sensor source may submit observations.

  4. The method of claim 1, wherein a staleness record retains the provenance hash of the rejected observation, enabling governance review to identify sources producing persistently stale data.

  5. The method of claim 1, wherein admitted observation artefacts are submitted to the same evidence admissibility gate used for document-derived evidence, enforcing identical schema validation and provenance verification across all evidence types.

  6. An environment state interface system comprising one or more processors and a sensor registry, configured to perform the method of any of claims 1 to 5.

  7. A non-transitory computer-readable medium storing instructions which, when executed by one or more processors, perform the method of any of claims 1 to 5.

Abstract

A computer-implemented environment state interface is disclosed for evidence-bound AI reasoning architectures. Sensor-derived and data-feed-derived observations are normalised into canonical evidence artefact format using registered sensor adapters. Freshness validation rejects stale observations and produces staleness records, preserving provenance of rejected data. Admitted observations receive a three-level timestamp distinguishing world time, publication time, and admission time, consistent with the temporal event graph schema. A SHA-256 provenance hash computed over the raw observation, adapter identifier, and world time provides a tamper-evident content identifier. Normalised observation artefacts pass through the standard AIEP evidence admissibility gate, achieving parity of provenance quality with document-derived evidence. +-------------------+-------------------+ | v AIEP Evidence Admissibility Gate | v Causal World State Graph (P200)


---

## Detailed Description

[0009] **Sensor Registry.** The sensor registry maps `sensor_id` to a `SensorAdapter` object. Each adapter implements: `parse(raw_observation) -> ObservationRecord`; `get_observation_schema()` returning the expected output schema; and `validate_freshness(observation_time)` checking against the sensor's configured staleness threshold.

[0010] **Observation Normalisation.** The normalised observation record follows the AIEP evidence artefact schema with additional fields: `sensor_id` (registered sensor identifier); `observation_time` (ISO 8601 timestamp of actual observation); `published_at` (the time the sensor published the reading to the interface); `admitted_at` (AIEP ledger admission timestamp); `raw_hash` (SHA-256 of the pre-normalisation raw observation bytes); and `normalised_content` (the structured canonical representation).

[0011] **Freshness Gate.** Each sensor adapter specifies a `max_staleness_seconds` threshold. Before normalisation, the ESI checks whether `now - observation_time > max_staleness_seconds`. Stale observations are rejected with a NegativeProofRecord (per P129 governance model) rather than admitted to the evidence ledger.

[0012] **Provenance Continuity.** The ESI includes the `raw_hash` in the evidence artefact provenance chain, ensuring that the raw sensor reading can be independently verified against the normalised artefact. The raw observation bytes are retained in a separate raw store for configurable duration.

---

## Claims

1. An environment state interface for an evidence-bound reasoning architecture that normalises sensor observations into canonical evidence artefact format with provenance preservation.
2. The system of claim 1 wherein sensor observations are assigned three-level timestamps distinguishing observation time, publication time, and admission time.
3. The system of claim 1 wherein stale observations are rejected via governance records without admission.
4. The system of claim 1 wherein normalisation adapters are registered per sensor and implement a canonical transformation interface.
5. The system of claim 1 wherein the raw observation hash is preserved alongside the normalised artefact hash for independent verification.

Dependencies

This specification is published as open-source prior art under the Apache License 2.0. Copyright 2026 Protocol Owner. Inventor: Protocol Architect.

Patent information · Licensing · All open specifications