Information autonomy starts with meaning.
Keeping information on your own servers is one thing. Reading, connecting and using it independently is another. Martijn Aslander makes that distinction clear. We share the direction and examine what it requires of a platform for organisations.
Appreciation ↓An Indentia review of Martijn Aslander’s work, with attention to Fons Trompenaars. This is our interpretation, not a joint publication or an endorsement by the named authors.
What Aslander gets right.
Aslander’s contribution goes beyond a preference for Markdown. His book brings information architecture into the boardroom. His papers explore how to record which person, organisation or event information refers to, what a judgment is based on and how errors can be corrected.
Control and usability
Moving information from the public cloud to your own infrastructure does not automatically make it findable or transferable.
Meaning matters more than LLM model choice
Carefully connect references to the same people, organisations and events. A well organised knowledge base retains value when the language model (LLM) changes.
Expert judgment gets an address
Human in the Ground calls for judgments at an identifiable object, with reasons, provenance and a lasting record of challenges and corrections.
You can keep using Word.
The Managementboek interview says this explicitly. New information can be structured from the start; existing Word documents, PDFs and scans can be processed. Both routes need identification, quality checks and provenance.
Follow one claim through the process.
Fictional example · open the steps
1The source makes a statement
A report states: “Delta supplies the municipality of North.” Preserve the passage, document version and date. The mention alone does not establish the correct legal identity.
2Identity is reviewed
Two companies are named Delta. A candidate gets a stable identifier; the visible name remains a label. Without sufficient evidence, the link stays unresolved.
3A judgment gets grounds
An expert assesses a dependency risk using the contract and ownership structure. The judgment gets an author, reason and status. A derivation does not automatically inherit “human reviewed” status.
4The source changes
If the contract turns out to belong to a different company, affected conclusions and indexes must also be identified. Correcting the text alone does not repair the chain.
How Indentia develops these principles.
The value lies in explicit contracts for shared use. These are architecture choices and their acceptance questions. A design choice does not guarantee that every product path is already delivered.
| Question | Indentia approach | What must be demonstrated? |
|---|---|---|
| Who do we mean? | Stable identity separate from its label; candidate matching and review. Unresolved links remain visible. Architecture contract | Namesakes stay separate. Renaming does not change identity. |
| Who may know this? | Source permissions, tenant boundaries and delegated agent permissions in the access architecture. Existing foundation; validate each application path | Evidence, snippets, derivations and exports also respect permissions. |
| Why is this here? | Provenance connects source, processing, claim and human review. Architecture contract | A user can return to the right source version and passage. |
| Who established this? | Model proposals, human corrections and derived results retain distinct statuses. Review contract in development | Automatic reuse never silently becomes human approval. |
| What if a source is withdrawn? | A designed cascade follows dependencies, reassesses conclusions and updates indexes. In development | A withdrawn source does not silently continue supporting answers. |
| Can we continue without Indentia? | The proposed autonomy package connects documents, graph, schema and provenance in an export. Proposed guarantee; still to be demonstrated | An independent reader reconstructs content and relationships without our code. |
Where the argument exceeds the evidence.
Appreciation also requires precise limits. These notes concern published claims. They do not prove that a capability is absent from his own system.
Machine-readable does not mean reliable
Markdown cannot prevent an incorrect record, missing source or unclear concept. Word and PDF can also be processed by machines, with varying reliability. Format matters, but cannot fully explain IT problems.
Read the source ↗Repeatability is not correctness
A deterministic script can consistently miscalculate or search an incomplete dossier. “No payment found” is meaningful only if you know which accounts were searched and how payments were matched to invoices. Source quality, exceptions and assumptions still matter.
Read the source ↗A readable name does not prove identity
The emoji paper makes references inspectable. It also acknowledges that linking to the wrong namesake can remain invisible. Its write experiment gives models no lookup for unknown numbers. It does not establish general superiority over resolving, validating and then writing.
Read the source ↗A corpus scan and an experiment are different evidence
399,598 references belong to an integrity scan. The forty invisible failures come from a separate experiment with 320 write acts. The private snapshot limits independent reproduction. The result supports readability in that setup, not universal error prevention.
Read the source ↗A combination is not automatically a new foundation
Formal inference, entity resolution and provenance have a long history. OWL and PROV-O provide existing standards. The practical combination can be valuable; broad novelty claims require independent comparative research.
Read the source ↗Six lines to open the conversation.
Original Indentia wording inspired by this review. These are not quotations from Martijn Aslander or product guarantees.
You can keep using Word and still free your information.
A readable file is the beginning. Meaning makes it useful.
A name helps you read. An identity helps you connect.
Repeatability becomes useful when the assumptions are visible.
Correcting a source also requires repairing its consequences.
An autonomy promise becomes credible when the customer can continue without us.
Read the authors themselves.
Credit to Martijn Aslander for the book, papers and Life Lens System, and to Fons Trompenaars for the leadership perspective. The interview with Aslander and Trompenaars was published by Managementboek, crediting Ronald Buitenhuis and Mirjam van der Linden. The papers are preprints. The Indentia comparison is our own analysis.
- Informatieautonomie ↗Martijn Aslander · boek
- Informatie moet chefsache zijn ↗Ronald Buitenhuis / Mirjam van der Linden · Managementboek · 12 oktober 2026
- A reading guide ↗Martijn Aslander
- The Information Problem Was Never the Real Problem ↗Martijn Aslander · preprint v1.3
- If You Know Why What Is What ↗Martijn Aslander · preprint v1.0
- Peirceiving Turing’s Dream ↗Martijn Aslander · preprint v1.0
- Human in the Ground ↗Martijn Aslander · preprint v1.0
- The Unexpected Elegance of Emoji ↗Martijn Aslander · preprint v1.0
- Life Lens System ↗Martijn Aslander
- OWL 2 / PROV-O ↗W3C · standards
- PROV-O: The PROV Ontology ↗W3C
Make autonomy testable.
Bring your own dossier. Examine what becomes readable, which identity remains uncertain, who may see the evidence and what changes after a correction.
Discuss your own caseAll reports