How the Three Platforms Work Together

In One Sentence

The three platforms handle different stages of the same workflow and share one account system and one Credit ledger:

• MatrixOne puts data in the right place, keeps it governed, and makes it queryable;

• AI Studio builds on this foundation to make data AI-Ready and put it to work through agents;

• Genesis turns model supply into services that can be operated, metered, and offered externally.

There are three separate entry points so that different roles can find the right place to start—not because they are three unrelated systems. The sections below explain their specific connections along three main axes.

Axis 1: Data (MatrixOne ↔ AI Studio)

This is the strongest structural connection. It is not merely “collaboration,” but a storage and processing relationship involving the same data.

AI Studio: Connectors → Catalog (tables / volumes / vectors / models / operators) → Workflows / SQL / knowledge bases
                                    │
                   (tables, volumes, vectors, and SQL are all stored in or run on)
                                    ▼
MatrixOne: Hyper-converged engine (OLTP / OLAP / vector / full-text) + compute-storage separation + Git-for-Data

Handoff object	On the AI Studio side	On the MatrixOne side
Tables	Structured objects in the Catalog	Table storage in the engine
Volumes / files	Unstructured objects in the Catalog	Object storage and engine management
Vectors	Embedding results from knowledge bases	Vector search capabilities in the engine
SQL	SQL editor, workflow SQL nodes, and NL2SQL	Executed on the engine by query compute resources

Users can work entirely within AI Studio without being aware of the underlying engine. Alternatively, DBAs can create instances and write SQL directly. There are two permission systems: AI Studio’s functional-domain permissions control who can use which features and objects, while MatrixOne’s SQL permissions control who can perform which actions on which databases and tables. Both govern data, but they are configured separately.

Axis 2: Models (Genesis ↔ AI Studio, with Separate Ledgers)

Agent conversations, knowledge base Embedding, and workflow model nodes in AI Studio all invoke models. Model supply can share the same infrastructure, but the ledgers must remain separate. This separation is an intentional product design, not an oversight.

AI Studio needs to invoke a model (conversation / Embedding / workflow model node)
        │
        ├─ Use it in the platform UI → billed as "AI Services" (models may be hidden, with unified pricing)
        │
        └─ Use an API Key from an external site / your own app → billed as "Genesis" (choose the model, priced by model)

“Integrated” does not mean “one muddled ledger”: model supply is shared, but customers need clear billing records and a clear mental model.

Axis 3: The Flywheel (Runtime Data Feedback)

An agent does not remain unchanged after it is built. It becomes more accurate with use through runtime data feedback:

An AI Studio agent performs a task
   → Produces interactions, retrieval evidence, feedback, and success/failure records
   → Accumulates them as governable data (written back to the Catalog, on the MatrixOne foundation)
   → Improves knowledge base chunking, semantic configuration, retrieval, and prompt strategies
   → Produces a better agent

MatrixOne’s versioning (Git-for-Data) makes the data from every iteration reproducible and reversible. Genesis provides compute when model inference is required, while AI Studio is where the flywheel primarily operates.

Separate Ledgers

Ledger separation is the foundation of billing across the three platforms and must be respected in any content related to billing or usage. Which ledger a model invocation belongs to depends only on its entry point:

Billing item	What is sold	Who initiates it	Is the model exposed?
AI Services	Platform-packaged AI capabilities	A user working with agents / workflows / knowledge bases in AI Studio	No
Genesis	External-facing model services	An external site / your own app using an API Key	Yes; the user chooses it
BYOM	Integration services for user-owned models	A user integrating a self-deployed model into workflows or other features	The model belongs to you

Quick guide:

• Intelligence used through the UI (conversations, knowledge base Q&A, and so on) → AI Services

• Models invoked in code with an API Key → Genesis

• Your own deployed model is invoked → BYOM

A “Genesis” charge generally comes from a direct model API call. Intelligence used through the UI is not included in the usage shown in the Genesis console. For the complete billing rules, see Credits and Billing Items.

Unified Account

The three entry points share one registration, sign-in, and identity system, so users do not need to create multiple accounts:

• The same account can access each entry point, with product switching available in the top navigation bar;

• Credits are metered centrally at the account level, while billing details are separated by platform / billing item;

• The primary way to obtain a Genesis API Key is through a service account issued by the account system (UC), making the account system another connection spanning all three platforms.

What Is Missing If You Use Only One?

If you use only…	What is missing
MatrixOne	You have a database, but lack multimodal governance, a knowledge base collaboration interface, agent building, and a closed feedback loop
AI Studio (assuming there is no underlying foundation)	There is no hyper-converged compute and storage or versioned foundation on which the “all-in-one” experience can operate
Genesis	You can invoke models, but lack AI-Ready enterprise private data and an agent workbench
AI Studio + externally assembled databases / vectors / gateways	You return to fragmentation and stitching together multiple systems—exactly what MOI is designed to eliminate

Three Typical Combinations

Use case	How the platforms connect
Enterprise knowledge agent (most common)	Ingest through AI Studio → parse with workflows → build a knowledge base → create an agent; data is stored in MatrixOne; use Genesis only if an external application also needs to connect directly to a model
Q&A over structured + unstructured data	Use MatrixOne + Studio SQL / workflows to prepare wide metric tables, prepare documents in a knowledge base, and have the agent query data before retrieving documents
Primarily using model APIs	Start with Genesis Keys / models / usage / logs; when private-domain augmentation is needed, use AI Studio + MatrixOne for RAG

Related Documents

• AI Studio · MatrixOne · Genesis

• Product Introduction · Overview

• Guide overview and domain entry points: Guides

• Credits and Billing Items — details of the boundaries among the three ledgers