# Informational intelligence
canonical_url: https://www.3.14financialcontents.com/en/solutions/informational-intelligence/
markdown_url: https://www.3.14financialcontents.com/en/solutions/informational-intelligence/index.md
language: en
content_type: solutions
status: published
description: Turns scattered sources, data, language and logic into knowledge that can be understood, reused and put in motion.

Making what an organisation knows usable.

An organisation knows far more than any of its applications can show.

Part of that knowledge lives in documents and databases. Another part is built into code, configurations, forms, tables, models, emails and historical decisions. People add context, vocabulary, exceptions and criteria that make it possible to interpret all of the above correctly.

Each medium preserves a different part of reality:

- data records facts and states;
- documents explain purposes, procedures and arguments;
- software executes rules and decisions;
- traces show part of what happened;
- language expresses nuance, relationships and uncertainty;
- people know the scope, the exceptions and the consequences.

Informational intelligence relates those pieces and turns them into a representation that can be consulted, discussed, reused and connected to the work.

We build systems so that an organisation can establish what it knows, where it comes from, what it means, which version is in force, what contradictions exist and what remains open. From that base it can research, produce reporting, localise content, prepare decisions, supply context to agents or make the logic of an operation examinable.

## From accumulating information to having usable knowledge

The quantity of information has stopped being an advantage in itself. Value appears when the organisation can use it with sufficient context.

That requires answering questions a conventional search engine leaves out:

- are two documents talking about the same entity?;
- which version contains the policy in force?;
- does a source really support the claim, or does it merely share similar words?;
- does one figure correspond to the same period and unit as another?;
- does the documentation match what the system actually executes?;
- does an observed rule represent business intent or a historical workaround?;
- what can be stated with sufficient evidence?;
- what should be presented as a hypothesis, a contradiction or an absence?;
- how should the same knowledge be expressed for another audience, language or format?;
- which part needs to reach a decision, an agent or an operation?

Informational intelligence turns these questions into design responsibilities. The system preserves the link between each result and the material that supports it.

## A common base for many forms of intelligence

The same informational base can feed very different results.

It can answer a question with evidence. It can generate a report. It can produce a narrative in text, voice or video. It can adapt content to another language or variety. It can help a person compare alternatives. It can supply context to an agent before it acts. It can document the logic of an ERP. It can preserve why a project took a decision.

That is why we treat informational intelligence as a core solution and not as a technical data-preparation stage.

Its role consists of holding six qualities together:

### Meaning

Concepts, entities and relationships keep a shared interpretation within the domain.

### Provenance

Every fact, rule or claim can be related to the sources, locations and versions that support it.

### State

The system distinguishes information that is observed, inferred, validated, contradicted, obsolete or still pending.

### Context

Knowledge preserves the scope, the period, the units, the conditions and the exceptions that determine how it can be used.

### Accessibility

People, search engines, applications and agents can retrieve the representation appropriate to their function.

### Evolution

New sources, corrections and decisions update the knowledge without erasing the history that makes the change comprehensible.

## The journey we build

### 01 — Take in sources without losing their original form

The system can gather documents, tables, messages, pages, databases, code, configurations, images, audio, video, traces and other relevant sources.

Ingestion preserves:

- the original;
- its origin;
- the version;
- the moment or period it belongs to;
- the access permissions;
- the relationship with other sources;
- the structure needed to locate each fragment later.

Preserving the original makes it always possible to return to whatever an extraction, a synthesis or an inference tried to represent.

### 02 — Extract structure and relationships

The systems identify entities, events, figures, units, dates, concepts, claims, conditions, decisions and relationships.

Extraction can combine:

- parsers and rules;
- language models;
- vision and document recognition;
- table analysis;
- code analysis;
- specialised models;
- domain validations.

Each result is stored with its source location and with the method that produced it. Structure makes retrieval and relation easier; provenance makes examination possible.

### 03 — Resolve entities, versions and scopes

The same object may appear under several names. Two similar names may correspond to different entities. A rule may be valid only for one unit, one product, one period or one type of case.

We design resolution to preserve identity and scope:

- aliases and designations;
- hierarchies and memberships;
- versions;
- period of validity;
- units and scales;
- temporal relationships;
- territories or language varieties;
- domains of application;
- conditions and exceptions.

Resolution does not consist of merging everything that looks alike. It consists of making explicit when two references can be treated as the same object and when they must remain separate.

### 04 — Represent semantics and domain knowledge

An organisation needs more than retrievable fragments. It needs a common representation of the concepts with which it thinks and operates.

We work with:

- taxonomies;
- lightweight ontologies;
- knowledge graphs;
- typed schemas;
- vocabularies and terminology;
- logical relationships;
- states and transitions;
- questions and decisions;
- constraints and invariants.

The representation is designed together with the people who know the domain. Its aim is to let people and systems share meanings sufficient for the work, without turning the knowledge model into a bureaucracy separate from the operation.

### 05 — Preserve evidence, contradiction and uncertainty

Finding a relevant source and demonstrating a claim are different tasks.

Every important claim can preserve:

- the evidence that supports it;
- the evidence that contradicts it;
- the exact fragment or datum used;
- the authority and scope of the source;
- the transformation applied;
- the validation state;
- the questions that remain open.

Contradiction is represented as knowledge. An answer can explain that two versions exist, show where they come from and point out which criterion is missing to resolve them.

Absence is represented too. When the available coverage does not support a conclusion, the system can abstain, formulate the pending question or propose which source would be needed.

### 06 — Retrieve according to the nature of the question

Not every question is resolved by the same mechanism.

We combine:

- exact search;
- lexical search;
- semantic retrieval;
- structured filters;
- relational queries;
- graph traversals;
- entity expansion;
- code and dependency retrieval;
- decision memory;
- reranking.

A query about a name, a figure or an identifier demands lexical precision. A conceptual question needs semantic similarity. A question about dependencies, versions or paths needs structure.

Hybrid retrieval selects and combines those routes, and then returns evidence located precisely enough for the result to be reviewed.

### 07 — Compose answers, reporting and content

Informational intelligence also covers the way knowledge is expressed.

Once the evidence is structured, the system can produce:

- cited answers;
- reports;
- executive summaries;
- explanations by level of knowledge;
- scripts;
- narratives;
- voice and subtitles;
- graphics and video;
- translations;
- localisations;
- NLG compositions;
- technical and operational documentation.

Composition starts from a shared informational structure. Facts, figures, entities and relationships are established before any drafting. Afterwards the selection, the order, the level of detail, the language, the register and the format are adapted.

This separation lets a change in the evidence update different versions without rebuilding each piece by hand.

### 08 — Deliver knowledge to decisions, agents and operations

Knowledge acquires more value when it can enter the place where a decision is taken or an action is performed.

We build interfaces so that:

- a person receives the relevant context before deciding;
- a predictive model uses consistent variables and definitions;
- an agent retrieves valid tools, policies and evidence;
- a workflow knows states, dependencies and exceptions;
- an ERP consults rules or documentation;
- a project preserves questions and decisions;
- a reporting system reuses the same informational base;
- a physical operation receives coherent objectives and constraints.

The informational representation works as a common layer between sources and uses. Each consumer receives only what it needs, with the corresponding scope and permissions.

## Business logic and operational knowledge

A particularly valuable part of an organisation's knowledge is built into its systems.

Code, stored procedures, configurations, tables, forms, spreadsheets and workflows execute decisions every day. Documentation explains another part. People know the exceptions and the agreements that keep the process running.

DocuLogic is our line of work for reconstructing that logic and turning it into examinable operational knowledge.

The system identifies:

- entities and attributes;
- states and transitions;
- conditions and thresholds;
- calculations;
- decisions and alternatives;
- permissions;
- dependencies;
- effects;
- exceptions;
- versions;
- pending questions.

Every rule preserves its evidence and its state. We distinguish between observed behaviour, inferred logic, validated knowledge, contradiction and organisational decision.

This distinction makes it possible to use existing software as a source without turning it into an absolute authority. An implementation may contain a historical workaround, an inconsistency or a behaviour the organisation wants to change.

Confirmed logic can feed:

- living documentation;
- impact analysis;
- migrations;
- ERP modernisation;
- testing;
- decision tables;
- assistants;
- agents;
- automation;
- auditing;
- optimisation.

## DocuLogic and KIR: knowledge readable by people and usable by systems

We have developed an architecture of our own to reduce the distance between the knowledge a person can write and review and the representation that applications, rules and agents need.

### DocuLogic

Besides reconstructing the logic, DocuLogic preserves it: it works as a layer for human authoring and review. It allows knowledge to be expressed in readable, versionable formats — Markdown and YAML, for example — together with:

- provenance;
- verification state;
- period of validity;
- scope;
- life cycle;
- attested calculations or checks;
- relationships with other pieces.

Its aim is to let a person read, discuss and modify knowledge without working directly on a database or an execution representation.

### KIR

KIR — Knowledge Intermediate Representation — is a typed, canonical intermediate representation.

It receives knowledge from DocuLogic, extractors, analysers and other sources, and turns it into a stable structure that can feed:

- relational databases;
- graphs;
- semantic indexes;
- rule engines;
- APIs;
- MCP;
- agents;
- Finaz;
- ERP-Bots;
- SuperQuestions;
- MagdalenOS;
- other 3.14 systems.

The architecture can be summarised like this:

```text
SOURCES
   ↓
EXTRACTION AND AUTHORING
   ↓
DOCULOGIC
   ↓
KIR
   ↓
IMMUTABLE KNOWLEDGE VERSION
   ↓
INDEXES, GRAPHS AND SEMANTIC PROJECTIONS
   ↓
SEARCH · REPORTING · APIs · MCP · AGENTS · OPERATIONS
```

This intermediate layer makes it possible to change a database, a model or an interface without losing the central representation of the knowledge.

### Semantic and logical technologies

Depending on the case, the structure can rely on:

- LinkML to define schemas;
- TypeDB or other typed graphs for complex relationships;
- PostgreSQL and pgvector for relational, lexical and semantic knowledge;
- Datalog for deduction;
- DMN for decisions;
- CEL for validations;
- Tree-sitter and LSP for code structure;
- hybrid search, embeddings and rerankers;
- language models for extraction, comparison and composition.

These technologies sit underneath the solution. The solution remains making knowledge comprehensible and usable.

## Language, translation, localisation and NLG composition

Language is part of knowledge and also part of its delivery.

An organisation may need to express the same informational base:

- in different languages;
- in national varieties;
- with specialised terminology;
- for audiences with different levels of knowledge;
- in text, voice or video;
- at different lengths;
- under editorial or regulatory constraints.

Localisation and translation belong to informational intelligence because they force a distinction between what may change and what must be preserved.

We work with:

- contrastive research;
- glossaries and memories;
- semantic invariants;
- terminology;
- register;
- relationships between concepts;
- controlled generation;
- adversarial auditing;
- human review.

Excorpora explores these capabilities in translation and localisation. Finaz applies them to multi-format, multi-audience reporting.

Natural language generation — NLG — makes it possible to turn informational structures into language through rules, templates, calculations and generative models. The deterministic part protects what must remain exact; the generative part adapts selection, explanation and style.

## What can change for an organisation

### Less time spent rebuilding context

People can reach a representation that relates sources, versions, entities and decisions instead of repeating the same research in every case.

### More examinable answers and reports

Every result can show what evidence it used, what transformation it applied and what uncertainty remains.

### Reusable knowledge

A single base can feed search, reporting, training, decisions, agents, documentation and operations.

### Logic that is easier to change

Rules and dependencies stop being visible only inside code or configurations and can be discussed before migrating, automating or optimising.

### Greater coherence across languages and formats

Text, voice, video, translations and localisations share facts, relationships and invariants.

### Better context for agents and models

Systems receive entities, policies, relationships and evidence with scope and provenance, instead of fragments without a common representation.

### More useful organisational memory

Decisions, questions and changes of criterion can be preserved in a form that allows the work to be continued and reviewed.

## Applications

### Research and corporate knowledge

Systems for consulting documentation, relating sources, examining evidence and keeping knowledge up to date.

### Business logic and modernisation

Reconstruction of rules, decisions and dependencies for ERP, migrations, integration and automation.

### Generative and multimodal reporting

Production of reports, scripts, voice, subtitles and video from a single informational base.

### Translation and localisation

Adaptation across languages, varieties, registers and audiences while preserving meaning and terminology.

### Intelligence for projects

Memory of questions, evidence, decisions, dependencies and work.

### Context for agents

Tools, policies, domain knowledge and states delivered to agents through controlled interfaces.

### Specialised service and support

Answers based on documentation, case context and procedures, with human handover when the knowledge or the authority requires it.

### Decision intelligence

Variables, definitions, evidence and context prepared for classification, prediction, scenarios or optimisation.

## The role of people

Informational intelligence needs human knowledge in several places.

People:

- establish which sources hold authority;
- define concepts and relationships;
- resolve domain ambiguities;
- validate inferred logic;
- decide between contradictory versions;
- determine what information may be shared;
- review meaning, register and interpretation;
- correct the system;
- decide what knowledge should be preserved, updated or forgotten.

The interface must concentrate their attention on real decisions. A specialist contributes more value by examining contradictions, high-impact inferences and doubtful scopes than by mechanically reviewing hundreds of fragments that are already checkable.

## How we evaluate

Evaluation is adapted to the use, but it can include:

### Coverage

How much of the relevant sources, entities, concepts, rules or questions is represented.

### Retrieval

Whether the system finds the right evidence for a query and leaves out material that is similar but insufficient.

### Support

Whether every citation or datum supports exactly the claim that consumes it.

### Semantic resolution

Whether entities, versions, scopes, units and relationships are interpreted correctly.

### Contradiction

Whether the system detects and represents incompatible evidence without hiding it inside an answer.

### Abstention

Whether it recognises when it needs more information, a different source or human judgement.

### Linguistic fidelity

Whether translations, localisations and compositions preserve meaning, terminology, relationships and invariants.

### Multi-format coherence

Whether text, voice, subtitles, graphics and video share the same informational core.

### Updating

Whether the arrival of new information modifies the right pieces and preserves the relevant history.

### Usefulness

Whether people find context sooner, review better, reuse knowledge and reduce reconstruction or rework.

## Conditions and limits

Informational intelligence organises, relates and makes examinable the knowledge available. The quality of the result depends on the sources, their coverage, their currency and the authority with which they can be used.

Domain contradictions, implicit intentions and organisational decisions still need people able to answer for them.

Models help to extract, relate and compose. Rules, schemas, tests and human review protect whatever requires exactness, stability or accountability.

Maturity is declared along the journey. A system may have solid ingestion and still be evaluating logic inference or generation in a particular language variety.

## Related systems

### DocuLogic

Documentation, code and logic turned into verifiable operational knowledge.

### KIR

Typed intermediate representation that projects knowledge towards each use.

### Finaz

Evidence and informational structure turned into reporting in text, voice and video.

### Excorpora

Research and evaluation of translation and localisation with semantic integrity.

### SuperSocrates

The system that asks: gaps, collisions between sources, decisions and assumptions, and the next useful question of a project.

### SuperQuestions

The product that represents and manages questions, evidence, decisions, dependencies and work as knowledge objects of a project.

## Closing

Informational intelligence lets an organisation better understand what it already knows, make visible the relationships that today remain scattered, and deliver reliable knowledge wherever it must inform, decide or act.

---

https://www.3.14financialcontents.com/llms-en.txt · https://www.3.14financialcontents.com/llms-full-en.txt