i6eal.de
DEEN
Start a project
Public AI signal chronicle

What changed across the public AI landscape?

The signal chronicle preserves changes across our public data projections in a complete private append-only archive. This page presents a bounded, balanced publication window with stable identifiers, observation times and original sources.

Complete private append-only archiveBounded, balanced publication windowObservation and effective time stay separateNo generative AI model
Chronicle generated 17 Jul 2026, 21:16Observed since 17 Jul 2026Revision 2Schema ki-signalchronik.public.v1
chronicle events8immutable observations in the complete private archive
newly observed1first appeared in a later projection
changed subjects1stable identifiers with at least one change
covered data lanes7active lanes and lanes limited to their baseline
The critical time boundary

Observed does not mean effective

The chronicle separates when we observed a change from an effective date stated by the source. If a record is missing in a later run, it is not automatically deleted or withdrawn.

observedAt ≠ effectiveAt
8 results
Append-only change feed

Every change remains in the complete archive

The private archive preserves every event immutably. This public page shows a bounded, balanced window grouped by observation time; open the record-level evidence or jump into the matching data tool to inspect the current state.

8 events
  1. NewIdentity review

    Medtronic GmbH ↔ Medtronic GmbH

    Observation state set
    Not observedobserved
  2. BaselineData coverage
    Data laneData coveragelane:coverage

    Baseline captured for Data coverage

    Observed facts set
    Not observed953
  3. BaselineFunding
    Data laneFundinglane:funding

    Baseline captured for Funding

    Observed facts set
    Not observed3,085
  4. BaselineResearch
    Data laneResearchlane:research

    Baseline captured for Research

    Observed facts set
    Not observed1,453
  5. BaselineKnowledge flow
    Data laneKnowledge flowlane:knowledge_flow

    Baseline captured for Knowledge flow

    Observed facts set
    Not observed1,133
  6. BaselinePublic procurement
    Data lanePublic procurementlane:procurement

    Baseline captured for Public procurement

    Observed facts set
    Not observed49
  7. BaselinePatents
    Data lanePatentslane:patents

    Baseline captured for Patents

    Observed facts set
    Not observed250
  8. BaselineIdentity review
    Data laneIdentity reviewlane:identity

    Baseline captured for Identity review

    Observed facts set
    Not observed10
Deterministic method

From a snapshot to an auditable event

The chronicle compares versioned projections by stable identifier. It describes differences without inventing reasons, impact or identities.

  1. 1
    Preserve the snapshot

    Every successful source state receives an immutable timestamp and revision.

  2. 2
    Canonicalize identifiers

    Subjects are recognized across runs only through published stable source identifiers.

  3. 3
    Compare fields

    Versioned rules distinguish the baseline, new records, updates, corrections and explicit withdrawals.

  4. 4
    Append the event

    The difference enters the chronicle with before, after, provenance and a direct link.

Collection, comparison, classification and presentation use no generative AI model.

Source snapshots

Every input retains its own timestamp

The chronicle run does not replace individual source dates. Open the original evidence when you need to inspect a change in context.

16 Jul 2026, 15:58AI patent families with German association signalsOpen source ↗

How to read the chronicle correctly

  • The baseline is a first observation, not a historical change.
  • A change may have become effective before it was first observed.
  • A missing record is not a withdrawal without an explicit source signal.
  • A correction describes a changed source state and does not automatically imply an error by the organization concerned.
  • Bounded data lanes are never extrapolated to a supposed full population.
  • Temporal sequence proves neither cause nor impact.

Frequently asked questions

Is the baseline a list of earlier changes?
No. It documents only what existed when longitudinal observation began. Real changes arise only from comparison with later successful snapshots.
Why are there two timestamps?
observedAt shows when our collector recognized a difference. effectiveAt contains an effective time only when the source itself supplies one. The two can be far apart.
Does a disappearing record mean it was deleted?
No. A record may be temporarily absent, a source may change its coverage, or a run may be bounded. A withdrawal appears only when there is an explicit and reliable source signal.
How are subjects recognized across runs?
Only through stable source identifiers or published exact identifier crosswalks. Matching or similar names never create an automatic connection.
Does the signal chronicle use generative AI?
No. Snapshotting, normalization, comparison, event classification and presentation follow versioned deterministic rules.

Want to track changes in public data reliably?

We build append-only data chronicles, collectors and evidence products with stable identifiers and verifiable provenance.

Discuss a data projectView all tools
Related tools

Useful next steps

These tools complement the current result.

View all tools
IndexPublic AI outcome ledgerFollow public AI funding from projects and organizations to reported outputs and citations. Every connection retains its evidence; patents and procurement appear only with exact identifier bridges.IndexPublic AI actorsSearch organizations across funding, research, procurement, influence and infrastructure. Every role stays tied to its source; matching names are queued for review only.IndexAI innovation pathFollow evidence-backed links from EU funding to reported research outputs—and see exactly where patents or procurement cannot yet be joined safely.