Grounded answers
Facts come from the layer itself rather than from a copy that may be days old.
AI & Intelligent Systems
The data layer an AI client can reason over.
An AI feature is only as good as what it can read. When the context has to be assembled from four systems, the feature inherits the seams.
An agent asks; the layer answers with records, windows and documents in one request.
Retrieval, reasoning and generated output grounded in the operational data.
Intelligent features need their context close to the data: records for facts, time series for what is happening, documents for policy, relationships for structure. PLOMID keeps those shapes in one layer so retrieval is an access path, not an export. Retrieval is designed as an access path over the same layer, not an export.
An agent asks; the layer answers with records, windows and documents in one request.
Environment
The agent
An AI client that needs grounded context before it can produce an answer.
SQL · JSON / documents
Two estates, drawn.
The same shapes, held two ways. The difference is not the storage, it is where the agreement between them lives.
Separate systems Copies kept in step
One layer One plan · one contract
What this workload moves.
Every shape below is a surface of the layer, not a format to be converted into one. They are read and written together.
- SQL Records, keys and joins
- JSON / documents Documents and nested objects
- Time series Measurements and events in time order
- Vector Similarity and semantic retrieval
- Graph Relationships and traversal
How the workload reaches the data.
The models this workload names, the path a request takes through them, and the surfaces that speak to the layer.
AI & Intelligent Systems · architecture
Workload
The shape of the work itself.
- Context
- Retrieve
- Ground
- Act
Data models
The models this workload names.
- SQL
- JSON / documents
- Time series
- Vector
- Graph
The layer
One path from a request to the data it names.
- Planning Predicates narrow the work before it runs
- Execution Rows, fields and windows answered together
- Transactions Readers and writers do not block each other
Surfaces
How the workload reaches the layer.
- SQL surface The query language the layer is documented in
- Applications Services and jobs reading and writing as they run
- Analytics & AI clients The same layer, the same access path
The work, stage by stage.
Choose a stage to read what happens there, or a shape to see every stage that handles it. Nothing on the map is a private interface.
PLOMID · AI & Intelligent Systems context · retrieve · ground · act Select a stage
Stage
Context
Records, windows and documents identified for the request
SQL · Time series · JSON / documents
01
Context
Records, windows and documents identified for the request
- SQL
- Time series
- JSON / documents
02
Retrieve
Similarity used to narrow candidates
- Vector
03
Ground
Facts read back from the operational layer
- SQL
- JSON / documents
04
Act
Output written as a record the system can audit
- SQL
- Events
What changes when the data is in one place.
Stated as properties of the system rather than as results we cannot measure for you.
One place to govern
Retrieval does not create a second estate of data with its own access rules.
Auditable output
What a feature decided can be written back as a record beside the data it used.
Mixed context
A single request can hold structured filters, a time window and a similarity search.
What this is built on.
Retrieval quality depends on provenance. These are the parts that keep the context attached to its source.
- Retrieval as an access path → Similarity search arrives as another way into the same data rather than a second copy to keep in step.
- One data layer → Rows, documents and time-ordered events live in one system, so a question is asked once instead of once per store.
- Every client, one layer → Applications, AI clients, services and background jobs reach the same data through the same interfaces.
- A public surface from the start → The SQL surface, the documentation and a browser playground exist today. Nothing described here needs a private API.
Workloads that sit beside this one.
Where this shows up.
Derived from the industry pages that reference this solution, so the two directions of the relationship always agree.