A semantic model is built once and then used everywhere. Once SemanticTwin.ai has turned your documents and code into a trusted model of your system, the same model can power AI agents, software development, documentation, analysis, reporting and compliance.
Each short video shows something SemanticTwin.ai already does. The link under it goes to the use case it explains.
AI and agents
Software development
Knowledge and documentation
Governance, risk and compliance
Business and people
Videos play from YouTube. Nothing loads from YouTube until you press play.
The examples use the BAI procurement system, where buyers request building materials, sellers quote, and an accepted quote becomes a purchase order. Watch the same three steps serve a different team in each area.
Buyers request building materials.
Sellers quote on the request.
An accepted quote becomes a purchase order.
Is prch_ordr a purchase order? How does it connect to the other tables?
| Table | What it holds |
|---|---|
prch_ordr | Unknown |
prch_ordr_offers | Unknown |
mvt_prch_ordr | Unknown |
prch_ordr and guesses.prch_ordr holds material requests. Sellers quote on them, and an accepted quote becomes a purchase order.
| Table | What it holds |
|---|---|
prch_ordr | Material request |
prch_ordr_offers | Seller quote |
mvt_prch_ordr | Purchase order |
SemanticTwin already does this.
Teams build it on top of the model. SemanticTwin provides the foundation: the meaning, connections, rules and proof that make it work.
Area 1 of 7 · Context for AI
AI is only as good as the context it gets. Without a semantic model, an AI assistant looking at a database sees prch_ordr and guesses. With one, it knows it is looking at material requests, and how they connect to quotes and orders.
MCP (Model Context Protocol) is the standard way to plug AI assistants into business systems. Through an MCP server, an assistant gets the meaning of every table and the right way to join them.
In SemanticTwin today
Built in An MCP server for published models. Teams can also build their own MCP tools that pair the model with their APIs.
“Which material requests became purchase orders last month?” The model gives the correct path through the tables, so the answer is right the first time.
In SemanticTwin today
Built in Ask in the Playground, and read-only queries checked against the model’s rules.
Answers come only from the model and its proof, and show where each fact came from.
In SemanticTwin today
Built in Ask.
Score how ready a model is for AI agents, with a list of gaps to fix first.
In SemanticTwin today
Built in Through the API.
An agent helping a seller needs to know that a submitted quote can’t be edited. The model’s rules and life stages tell it what it may and may not do.
In SemanticTwin today
Enabled Teams build it on top of the model.
Area 2 of 7 · A blueprint to build from
AI coding agents write code quickly, but they need to know what they are building. The semantic model gives them the blueprint: the business things, their details, the rules and the life stages.
Automated pipelines turn specifications into working software, and check their output against the model.
In SemanticTwin today
Built in An export format designed for spec-to-code pipelines.
Before making a link optional or removing a business thing, see which findings and rules are affected.
In SemanticTwin today
Built in What-if in the Playground.
New developers learn the system from the model, instead of piecing it together from the code.
In SemanticTwin today
Built in The Semantic Playground.
AI coding agents build new features using the model as their blueprint, so database tables, APIs and screens use the same names and rules everywhere.
In SemanticTwin today
Enabled Teams build it on top of the model.
Before moving an old system to a new platform, the team knows what every table and field means and which rules must survive the move.
In SemanticTwin today
Enabled Teams build it on top of the model.
Two systems call the same thing by different names. The model maps one to the other.
In SemanticTwin today
Enabled Teams build it on top of the model.
Each rule becomes a test. “A submitted quote can’t be edited” becomes a test that tries to edit one.
In SemanticTwin today
Enabled Teams build it on top of the model.
Area 3 of 7 · Documents tied to the facts
Most system documentation is out of date the day it is finished. Documentation generated from the model stays tied to the facts and their sources.
The business things, their connections and their proof form a graph you can explore, extend and share with other tools.
In SemanticTwin today
Built in The Knowledge Graph view, plus exports in the standard formats Turtle, JSON-LD, RDF/XML and N-Triples.
Twelve documents come from the model, including a domain overview, data dictionary, ER diagram, process flows, lifecycle diagrams and a business rules catalogue.
In SemanticTwin today
Built in
Sales, finance and IT use the same words for the same things, from one business glossary.
In SemanticTwin today
Built in The business glossary.
“Where do we store a seller’s short name?” Search the model instead of asking around.
In SemanticTwin today
Built in Search in the Playground.
New analysts, testers and support staff learn how the business works through the model.
In SemanticTwin today
Built in The Semantic Playground.
Every use case in this area is built in.
Area 4 of 7 · The answer up front
Analysts spend much of their time working out what a system does before they can improve it. The model gives them the answer up front.
An analyst inherits a system and explores it visually: the business things, what each one means, and where it is stored.
In SemanticTwin today
Built in The Semantic Playground.
Spot where the requirements say one thing and the database does another, or where a documented field doesn’t exist.
In SemanticTwin today
Built in Conflicts between sources are flagged, and a validation report lists the gaps.
See how work flows and the stages each thing moves through, such as a request’s nine stages from new to closed.
In SemanticTwin today
Built in Process flow and lifecycle diagrams.
Understand what an acquired company’s system does before deciding to keep, merge or replace it.
In SemanticTwin today
Enabled Teams build it on top of the model.
Find where several systems store the same business things, to plan consolidation.
In SemanticTwin today
Enabled Teams build it on top of the model.
Area 5 of 7 · One definition for every number
Reports disagree when two teams define the same number differently. A semantic model gives everyone one definition to work from.
Every field, with its meaning, its type, and whether it is required.
In SemanticTwin today
Built in The data dictionary.
“Orders dispatched this month” means the same thing in every report.
In SemanticTwin today
Built in The model can carry ready-made metric definitions.
For any fact in the model, see the document and the sentence it came from.
In SemanticTwin today
Built in The traceability matrix.
Business users build reports with business names, such as “material request”, instead of table names.
In SemanticTwin today
Enabled Teams build it on top of the model.
Required fields, allowed values and uniqueness become automatic checks. For example, a request’s status must be one of nine known codes.
In SemanticTwin today
Enabled Teams build it on top of the model.
Area 6 of 7 · Answers with proof
Regulators ask where personal data lives and who changed what. With a semantic model, those answers are already written down, with proof.
For privacy laws such as GDPR, list every field that holds personal information.
In SemanticTwin today
Built in Personal-data detection and the PII register.
Every change to the model records who made it, when and why, and every fact shows its source.
In SemanticTwin today
Built in History and proof on every fact.
Decide which roles may see which data, starting from the model’s list of sensitive fields.
In SemanticTwin today
Enabled Teams build it on top of the model.
When a rule changes, find every business thing, field and process it touches.
In SemanticTwin today
Enabled Teams build it on top of the model.
Area 7 of 7 · Knowledge that stays
The biggest benefit is often the simplest: knowledge that used to live in a few people’s heads now belongs to the whole organisation.
What a long-serving expert knows about a system is captured in the model, and stays when they move on.
In SemanticTwin today
Built in The published model.
Give an outsourcing partner or a new vendor a precise, sourced description of the system on day one.
In SemanticTwin today
Built in Generated documents and model exports.
Business owners and IT review and approve the same model, so what gets built matches what was asked for.
In SemanticTwin today
Built in Review and approval before publishing.
Every use case in this area is built in.
Every area, split into what SemanticTwin does today and what teams build on top of the model. Choose a use case to go to it.
| Area | Built in 23 | Enabled 11 |
|---|---|---|
| AI and agents | ||
| Software development | ||
| Knowledge and documentation | None | |
| System analysis | ||
| Data and analytics | ||
| Governance, risk and compliance | ||
| Business and people | None |
has a short video
A rule lives in the model with its source. Here is one rule from the procurement example, and four use cases that rely on it.
The ruleA submitted quote can’t be edited.
AI and agents
An agent helping a seller needs to know that a submitted quote can’t be edited. The model’s rules tell it.
Enabled AI agents that act safely
Software development
The rule becomes a test that tries to edit a submitted quote.
Enabled Designing tests from rules
Knowledge and documentation
The rule is listed in the business rules catalogue, one of the twelve generated documents.
Built in Documentation that writes itself
Governance, risk and compliance
Every change to the rule records who made it, when and why, and the rule shows its source.
Built in Audit readiness
MCP (Model Context Protocol) is the standard way to plug AI assistants into business systems. SemanticTwin includes an MCP server for published models, so an assistant gets the meaning of every table and the right way to join them.
Teams can also build their own MCP tools that pair the model with their APIs.
One semantic model supports 34 uses, across AI, software development, documentation, analysis, data, compliance and the business itself.
23 of them are built into SemanticTwin.ai today, including an MCP server for AI assistants, plain-English questions, the Semantic Playground, twelve generated documents and a register of personal data.
The other 11 are enabled: teams build them on top of the model, because it already holds the meaning, connections, rules and proof they need.
AI agents are the headline. Through MCP, an assistant can understand your system instead of guessing, and agents built on the model can act within its rules.
Build the model once with SemanticTwin.ai, then use it everywhere.
Watch SemanticTwin.ai build a semantic model from real documents.
Book a demo