What is ASUME?

ASUME reasons only from information it is allowed to access and use. That information may come from publicly accessible sources, such as company websites, product pages, job posts, case studies, documentation, announcements, public registers, and other business sources.

It may also come from Customer-provided or Customer-authorised sources, such as sales notes, product documents, technical files, call transcripts, customer feedback, internal reports, private knowledge bases, CRM systems, and connected integrations.

Where a Customer provides or connects private information, the Customer controls what it makes available to ASUME and the permissions granted to the Service. Where ASUME obtains information independently from public sources, ASUME determines the relevant sources and retrieval process subject to applicable law, third-party rights, source restrictions, and ASUME’s compliance controls.

A Customer may also designate a source it owns or controls as a Customer-Controlled Source, which allows ASUME to process that source within the permissions granted under the applicable agreement.

Company information is the input to ASUME. It is not, by itself, an understanding of the company. ASUME processes this information to identify and structure evidence relevant to constructing, maintaining, and revising its understanding of the company.

Not all information available to ASUME is evidence. Information becomes evidence when it provides support for a proposition about some aspect of the company's state. In other words, evidence is information that supports a proposition about a company.

Evidence remains connected to the information from which it was obtained so that its provenance, context, and relationship to the proposition it supports can be inspected. For example, a company publishing fifteen vacancies in implementation roles is an observation. That observation may become evidence supporting a proposition that the company is actively investing in implementation capacity. It does not, by itself, establish that proposition as true.

At a simplified level: Company information → Evidence → Proposition

Multiple pieces of evidence may support the same proposition, and the same information may provide evidence for more than one proposition when the relevant relationship can be established. Evidence may also differ in how strongly it supports a proposition. Additional information may strengthen the proposition, introduce competing evidence, or make the proposition no longer adequately supported.

ASUME therefore does not treat evidence as proof. Evidence provides support for propositions from which a broader understanding of the company can be constructed.

Inference is the process of deriving a proposition from evidence, assumptions where required, and relevant prior knowledge. Inference allows ASUME to derive propositions about a company that are not directly stated in the information available to it.

For example, available information may show that a company has opened multiple implementation roles, introduced new enterprise features, and expanded its implementation documentation. These observations may provide evidence supporting propositions about changes in the company's implementation capacity or priorities, even if the company never explicitly states those conclusions.

Inference can take different forms, including:

• Deduction: deriving a conclusion that necessarily follows from accepted premises.
• Induction: deriving a probable conclusion from observations or recurring patterns.
• Abduction: deriving a plausible explanation for available evidence when the underlying state cannot be directly observed.

Company information is often incomplete, distributed across different sources, and created for purposes other than describing the company's complete state. ASUME can reason across evidence derived from those sources rather than limiting its understanding to propositions explicitly stated by them. Where a proposition required for an inference is not sufficiently supported by available evidence, an explicit assumption may be required.

The Assumptive Understanding Model (AUM) is the computational framework ASUME uses to construct, maintain, and revise a structured understanding of a company when its state cannot be fully observed.

Not everything that matters about a company is directly available in the information ASUME can access. Its needs, capabilities, constraints, priorities, intentions, relationships, and other properties may need to be inferred from evidence distributed across different sources.

AUM provides a structure for performing this reasoning while preserving the distinctions and relationships among the elements involved.

Within AUM:

• Evidence is information that supports a proposition about the company.
• Assumptions are explicit propositions used where the available evidence does not sufficiently support something required for reasoning.
• Inference derives propositions from evidence, assumptions where required, and relevant prior knowledge.
• Relationships represent how evidence, assumptions, propositions, and other elements depend on or relate to one another.
• Understanding organizes these elements and their relationships into a structured representation of what ASUME currently understands about the company.

At a simplified level:

Evidence + assumptions where required → Inference → Propositions → Understanding

These distinctions matter because different parts of an understanding can have different epistemic bases. One proposition may be directly supported by evidence, while another may depend on an inference that itself relies on an assumption. Keeping those dependencies explicit makes it possible to inspect how a proposition was reached and what other parts of the understanding depend on it.

When new information provides new evidence, an assumption is revised or rejected, or the basis for an inferred proposition changes, AUM can reconsider the affected propositions and revise the resulting understanding while preserving what remains supported.

AUM therefore does not require uncertainty to be eliminated before a company can be understood. It provides a computational structure for reasoning under incomplete information while keeping the basis and dependencies of that understanding available for inspection and revision.

The result is not a collection of documents, search results, or a generic company summary. ASUME organizes supported and inferred propositions about a company into a structured understanding of that company.

This may include propositions about the company's:

• Needs: problems, requirements, or outcomes relevant to the company.
• Capabilities: what the company can currently do, provide, or support.
• Constraints: conditions that may limit the company's available actions or decisions.
• Priorities: areas receiving current attention, investment, or strategic importance.

These are examples rather than an exhaustive definition of company state. Other properties may form part of the understanding when they can be supported or inferred from available information.

Understanding is not simply a collection of propositions. Its structure preserves how those propositions relate to one another and the epistemic basis on which they are held. This can include the evidence supporting a proposition, assumptions on which an inference depends, relationships between propositions, uncertainty, temporal scope, provenance, and relevant alternatives.

As a result, two propositions within the same company understanding may have different bases. One may be supported directly by available evidence, while another may be inferred from evidence together with an assumption required for the reasoning.

A simplified representation might look like:

Evidence + assumptions where required → Inference → Proposition

The proposition forms part of the company understanding together with the basis and relationships relevant to how it was reached.

ASUME can therefore distinguish what is supported by available evidence, what has been inferred, and what depends on assumptions rather than reducing these different epistemic states to generated text. The result is a structured representation of what ASUME currently understands about the company, how the elements of that understanding relate, and the basis on which they are held.

Companies change. Companies change. They launch products, hire people, enter markets, change priorities, develop capabilities, encounter new constraints, form partnerships, and make decisions that alter their state.

An understanding of a company therefore represents its state at a particular point in time and on the basis of the information then available.

ASUME maintains its understanding as the company changes and new information becomes available. At a simplified level:

Understanding₁ + new information → New evidence → Revision → Understanding₂

Not all new information necessarily changes the understanding. New information may provide evidence supporting an existing proposition, supporting a new proposition, or supporting a proposition that is inconsistent with something currently held. As a result, an existing proposition may become more strongly supported, need to be reconsidered, cease to represent the company's current state, or be replaced by a different proposition.

New information may also affect assumptions used in previous reasoning. An assumption may become supported by evidence, remain necessary, require revision, or no longer be appropriate. Revision does not necessarily require reconstructing the entire company understanding. Because AUM preserves relevant dependencies among evidence, assumptions, inferences, and propositions, a change can be evaluated in relation to the parts of the understanding that depend on it while other supported parts remain intact.

This is the difference between repeatedly reconstructing company context:

Question → Search → Generate → Start again

and maintaining company understanding:

Information → Understand → Maintain → Revise → Reuse

Structured and maintained describe two properties of the same company understanding. Structured describes how the understanding and its dependencies are represented. Maintained describes how that representation persists and is revised as the company and the information available about it change. This allows people, applications, and agents to reason from the same maintained understanding rather than independently reconstructing the company from original source material for every task.

Understanding a company and determining what matters about that company for a particular purpose are separate problems. An objective defines the purpose for which ASUME reasons over an existing company understanding. For example, under a Revenue objective, ASUME may reason over a company's understanding to identify commercial opportunities. Under a Partnerships objective, ASUME may reason over the same understanding to identify the basis for a potential strategic relationship. A Marketing objective may use the same understanding to derive conclusions relevant to a different purpose. The important separation is:

Understand the company first → Apply the objective second

Changing the objective does not change the underlying company understanding. Instead, the objective determines which parts and relationships within that understanding are relevant to the purpose, what additional reasoning should be performed over them, and what output should be produced. At a simplified level:

Company understanding + Objective → Objective-dependent reasoning → Output

This means one maintained company understanding can support multiple objectives without requiring the company to be independently researched and reconstructed for each one. The output is therefore not the company understanding itself. It is a result derived by reasoning over that understanding in relation to a defined objective.

ASUME's understanding of a company is not tied to a single interface or objective. The same maintained company understanding can be used by people, applications, and agents for different purposes. A person may use ASUME directly to ask questions, discover opportunities, compare companies, or inspect the basis of its understanding. An application may use ASUME programmatically as part of another product or workflow. An agent may use ASUME while performing a larger task that requires understanding one or more companies. In each case, the underlying company understanding remains separate from the purpose for which it is being used:

Company understanding + Objective → Objective-dependent reasoning → Output

Different interfaces can therefore access the same underlying system without requiring ASUME to construct a separate understanding of the company for each interface or task.

ASUME may perform the objective-dependent reasoning and return the resulting output. Where appropriate, its structured understanding can also provide a basis for further reasoning by another application or agent. This separates three things:

Company understanding → Purpose → Consumer

The company understanding represents what ASUME currently understands about the company. The objective defines what should be derived from that understanding for a particular purpose. The consumer is the person, application, or agent using the resulting understanding or output. The ASUME application is one interface to this system. Programmatic interfaces allow applications and agents to use the same underlying company understanding within their own workflows.

An understanding is more useful when the basis on which it is held can be inspected. Traceability preserves the connections among company information, evidence, assumptions, inferences, propositions, and the outputs that depend on them. For a proposition reached through inference, a simplified dependency may look like:

Evidence + assumptions where required → Inference → Proposition

The proposition and its relevant dependencies form part of the company understanding. When that understanding is used for an objective, another dependency is created:

Company understanding + Objective → Objective-dependent reasoning → Output

Traceability connects these structures so that the basis of a proposition or output can be followed backward to the evidence, assumptions, and reasoning on which it depends. It can also be followed forward to identify what depends on a particular piece of evidence, assumption, or proposition.

This matters when the basis of the understanding changes. If new information changes the evidential support for a proposition, or an assumption is revised or rejected, ASUME can identify the dependent propositions and outputs that may need to be reconsidered.

Verification uses this traceability to inspect the basis on which propositions, understanding, and outputs are held. Verification may confirm that the existing basis remains adequate, identify information that changes that basis, challenge an assumption or inference, qualify a conclusion, or cause affected parts of the understanding to be revised.

Traceability and verification do not make inferred propositions certain. They make their basis, dependencies, and resulting conclusions inspectable, challengeable, and revisable.

The concepts above form a single process for constructing, maintaining, and using company understanding.

Company information is the information available to ASUME.

↓

Evidence is information that supports propositions about the company.

↓

AUM reasons from evidence, using explicit assumptions where required, to derive propositions and organize them and their dependencies into a structured company understanding.

↓

Company understanding represents what ASUME currently understands about the company, how the elements of that understanding relate, and the basis on which they are held.

↓

Maintenance and revision preserve that understanding over time and revise affected parts when new information changes its evidential or inferential basis.

↓

Objectives define the purpose for which ASUME reasons over the company understanding.

↓

Outputs are produced through objective-dependent reasoning over that understanding.

The central separation is: ASUME constructs and maintains an understanding of the company independently of any single objective, then reasons over that understanding for different purposes. The same maintained understanding can therefore support different objectives and be used through different interfaces without requiring the company to be independently reconstructed for each task.