By default, an AI model has broad general knowledge, but no understanding of what “net revenue,” “active customer,” or “renewed contract” mean at your company. So it fills the gaps with plausible assumptions, and no one notices. KesDone makes your AI more reliable by giving it access to validated, versioned, continuously maintained business context, so AI truly understands your business.
The same term can mean different things across departments. The AI picks the most common definition, not yours.
Exceptions no one wrote down
Edge cases live in people’s heads and conversations. Without them, an answer looks right but is not.
Decisions without a rationale
A rule changes and the history is lost. No one remembers why, or which version is authoritative.
Method
Business context evolves with your company.
A frozen reference quickly becomes wrong. When a definition, rule, exception, or source changes, the context is updated, validated, and versioned.
Understand
Understand the company, its knowledge, its rules, and its needs to identify the business context to build.
Structure and validate
Formalize definitions, rules, exceptions, and relationships, then have business owners validate them to build a reliable reference.
Maintain
Continuously version and maintain the business context to track changes in the company’s rules, knowledge, and implementations.
v1.0v1.1v2.0
Approach
A single source of business truth, built, validated, and maintained.
KesDone turns your company’s scattered knowledge into structured business context. That context becomes a shared resource for your AI systems, business teams, and Data and AI teams. It does not replace your source systems.
Captured knowledge
Tacit knowledge
What your experts know that is written down nowhere.
Documentation
Existing procedures, notes, policies, and documents.
Metadata, lineage, and observations
What your data and its transformations reveal about the rules in use.
Validated implementations
Rules already coded and confirmed in your systems.
KESDONE
Business Context Engineering Platform
StructureOrganize knowledge and its relationships.
FormalizeDefine business concepts, rules, exceptions, and decisions.
ConnectLink knowledge to its sources, dependencies, and supporting evidence.
ValidateHave business owners approve the knowledge and keep it traceable.
Version and maintainTrack how the context evolves and keep its history.
Consumers
AI systems
AI agent
Acts on the company’s validated rules.
AI assistant
Answers using the business definitions currently in force.
Business teams
Browse rules and definitions
A readable, shared, up-to-date reference.
Access to validated knowledge
Everyone knows which version is authoritative.
Data and AI teams
Use of the business context
One shared context instead of isolated interpretations.
Reference for AI projects and systems
A stable foundation to build and verify on.
A business application
Same question. Same data. Two different answers.
An AI assistant can calculate correctly and still be wrong, because it does not know your company’s rules. The figures below are not savings or gains: they show an answer that looks right and is not.
Finance assistantInternal AI assistant, example
HF
Head of Finance asksWhat was our net revenue in 2025?
AI
Assistant without business context
A generic definition calculates an incorrect net revenue without the company’s own rules.
No business context available. The assistant falls back on a generic definition of net revenue.
Its answer: net revenue 2025€1,060,000
Invoiced sales€1,150,000
− Discounts€50,000
− Credit memos issued in 2025€40,000
The math is right. The business answer is wrong.No warning is shown. The answer looks reliable, so nobody checks it.
What the assistant did not know
1€120,000 of intercompany sales were counted
2€30,000 of orders canceled before delivery were counted
3€20,000 of 2026 credit memos on 2025 sales were not deducted
AI
Assistant with business context
The answer relies on the company’s validated definitions, rules, and exceptions.
BUSINESS CONTEXT READ BEFORE ANSWERING
Net revenue · v1.0 · Validated by Finance
1Exclude intercompany sales
2Exclude orders canceled before delivery
3Deduct all credit memos, whatever their issue date
BUSINESS CONTEXT EXCERPTEXCERPT
Business definition
Amount of external sales delivered during the fiscal year, reduced by sales discounts and all credit memos relating to those sales, regardless of the credit memo’s issue date.
Business rules
Period: January 1 to December 31, 2025.
A sale belongs to the fiscal year by delivery date, not invoice date.
Sales discounts on eligible sales: deducted.
Exceptions and exclusions
Orders canceled before delivery: excluded.
Intercompany sales: excluded.
Credit memos issued in 2026 for 2025 sales: assigned to fiscal year 2025.
Governance and traceability
Business ownerFinanceApprovalApproved by FinanceVersion1.0StatusValidated
Its answer: net revenue 2025€890,000
Invoiced sales€1,150,000
1− Intercompany sales€120,000
2− Orders canceled before delivery€30,000
− Sales discounts€50,000
3− Credit memos on 2025 sales€60,000
Same data, same math. The difference is the company’s own rules, and the AI only applies them if it can read them.
CONTACT
Is your business context ready for the way you use AI?
We talk with Data and AI teams that want to better structure, share, and maintain the business knowledge their AI systems rely on.
The goal: understand your challenges, identify a concrete business problem, and explore how KesDone could help. You do not need to have defined a specific use case yet.