A practical guide for using Blyoum like a working manual.
This page gives users a simple starting point: what a Reasoner is, how to define its scope, how to add custom rules and evidence, what to ask, and how to verify the response before acting on it.
What the core product terms actually mean
If you are new to Blyoum, start here. These definitions explain the product model before you create anything in Studio.
The short version of the workflow
A Reasoner is not a generic AI chat. It is a reusable setup for a specific kind of work.
You begin by choosing one industry, then define the business domains where the system should operate, then add the work knowledge it should use. That structure becomes the frame for later answers.
You can also add Axioms to tell the Reasoner what rules it should always respect. These can encode internal policy, approval logic, compliance boundaries, and other custom operating constraints.
After that, you attach documents and reference material in the Knowledge Vault. When you ask a question, the system answers inside that bounded structure instead of pulling from an unstructured pile of context.
Every response should then be reviewed through its governance signals, including the evidence that supported it and any caution flags the system surfaced. The percentage bars now use a traffic-light read so weaker answers stand out faster: green for stronger signals, yellow for caution, and red for lower ones.
Use the definitions below to understand the product model first.
Create the Reasoner, add Axioms and sources, then ask practical questions.
Check evidence, governance signals, and surfaced cautions before acting.
A reusable reasoning instrument built for one industry, the business contexts you operate in, and the work it should know how to perform.
The top-level anchor for the Reasoner. It sets the vocabulary, operating context, and baseline assumptions the system works within.
The specific parts of the business where the Reasoner should operate, such as credit, underwriting, procurement, or field operations.
The methods or tasks the Reasoner is expected to handle, such as scenario analysis, policy review, forecasting, or investigations.
Custom rules you give the Reasoner directly. Use them for regulatory obligations, internal policy, definitions, approval logic, prohibited actions, and other non-negotiable guardrails.
The place where you attach documents, datasets, and reference material so the Reasoner can use them as bounded evidence.
Review cues like veracity, confidence, evidence coverage, and rule checks that help you inspect an answer before acting on it. Score bars use a traffic-light read: green for stronger results, yellow for caution, and red for weaker signals.
The basic operating flow
If someone clicks Documentation from the footer, this is the shortest path to understanding how the product should be used.
Create your Reasoner
Choose one industry, add the business domains you operate in, and select the work knowledge the system should use.
Set custom rules with Axioms
Add Axioms to give the Reasoner explicit operating rules, such as policy constraints, approval requirements, escalation triggers, and hard boundaries it should not cross.
Load supporting material
Upload playbooks, policies, spreadsheets, and reference documents into the right domain so answers stay properly scoped.
Ask practical questions
Use plain language. The system routes the request through its reasoning operators and returns a structured, inspectable answer.
Review the audit trail
Open governance, source evidence, and reasoning details to verify what supported the answer and where caution flags were raised. The score bars read like a traffic light: green means stronger, yellow means caution, and red means lower confidence or veracity.
The documentation areas users will care about most
These sections frame the page like a lightweight manual: setup, custom rules, source material, asking questions, and reviewing outputs.
User manual
Start here for the full workflow: composition, setup, axioms, uploading knowledge, and how to interpret answers in day-to-day use.
Document handling
Learn what to upload, where to attach it, and how document scope affects the evidence the Reasoner can rely on.
Prompting guide
Examples for asking comparative, investigative, and risk-oriented questions in a way that produces clearer outputs.
Axioms and governance
Understand how custom rules, veracity, confidence, evidence signals, and rule checks shape answers before they are acted on.
Common questions from new users
A small help section makes the page feel immediately useful even before you add deeper docs later.
What kind of files should go into the Knowledge Vault?
Upload the materials an operator would actually trust at work: policies, operating procedures, transaction documents, historical references, datasets, and internal guidance.
Why does document placement matter?
Files attached to the correct business or work context become bounded evidence for that scope. This helps prevent an answer from borrowing support from the wrong area.
What are Axioms used for?
Axioms let you give the Reasoner custom rules in plain language. They are useful for policy constraints, mandatory checks, approval requirements, escalation rules, definitions, and instructions the system should consistently follow.
When should I use an Axiom instead of a document?
Use an Axiom when you want to define a rule the system should apply every time. Use a document when you want to provide supporting material, evidence, examples, or references the Reasoner may need to inspect and cite.
How should I read the governance score bars?
Treat them like a traffic light. Green indicates a stronger signal, yellow suggests caution or mixed support, and red marks a weaker result that deserves closer review before acting.
Can I inspect how an answer was formed?
Yes. The product is designed to expose source evidence, rule checks, and reasoning details so users can audit the output instead of treating it like a black box.
Ready to try the workflow?
Create a Reasoner, add the rules and documents that define its scope, and start asking questions you can review with confidence.