GenticDB

The database your agents share.
The data you own.

Give every agent your brand builds one place to read, write, and search across commerce data, company knowledge, and agent-created work. Start with Gentic-hosted storage or connect your own Cloudflare R2 bucket so your structured and vector data files live in storage you control.

Open ParquetSQL + semantic searchCloudflare R2 BYOB
Business systems, knowledge & agent work
storefront · ads · analytics · reviews · docs · agent outputs
Gentic MCP
one consistent interface
GenticDB
the shared agent data layer
Every agent your brand builds
Gentic Computer · in-house agents · apps · Claude / ChatGPT
rowsembeddingsmarkdownimages/videoscodeparquet
The problem

Your agents do not have a model problem. They have a data problem.

Brand data is fragmented across storefronts, ad platforms, analytics tools, reviews, documents, asset libraries, and SaaS databases. Every system speaks its own dialect, and nothing holds the whole picture.

Vendor MCP servers expose inconsistent, isolated interfaces; they don't create shared business memory. Agents can answer questions, but they struggle to reliably understand the brand, complete work, and preserve what happened for the next agent.

The shared interface

One shared data layer for every agent your brand builds.

Every agent reads and contributes through the same consistent interface. No separate memory and no fresh data copy for every new agent you stand up — they all work against one store, so what one agent learns is there for the next.

Gentic Computer, in-house agents, apps, Claude, ChatGPT, and other MCP clients all read, write, and search through the shared Gentic MCP surface.

Business systems, knowledge & agent work
storefront · ads · analytics · reviews · docs · agent outputs
Gentic MCP
one consistent interface
GenticDB
the shared agent data layer
Every agent your brand builds
Gentic Computer · in-house agents · apps · Claude / ChatGPT
rowsembeddingsmarkdownimages/videoscodeparquet
What agents can work with

One layer for commerce data, knowledge, and the work your agents create.

Commerce and operational data

Products, customers, orders, reviews, campaigns, analytics events, research results, agent outcomes, and the derived tables your workflows build on top.

Structured data + embeddings

Familiar relational data alongside text and image embeddings — precise SQL and semantic search in one store, instead of a separate vector database bolted on.

Brand knowledge and memory

Guidelines, briefs, documents, research, transcripts, Brand Brain, and a brand wiki your agents keep current as work happens.

Creative and experiences

Make images, video, and landing pages agent-findable through durable records and references. (These bytes are tracked through Gentic — they don't live in your R2 bucket today.)

Open formats

Own your data in a format that does not need Gentic.

GenticDB stores structured and vector data in open Parquet files. Query them with DuckDB. Export CSV, Parquet, or JSON. Use standard object-storage tools. With bring-your-own-bucket, those files live in your Cloudflare R2 bucket, where you can inspect them and revoke access.

Gentic manages the catalog and operating layer. With BYOB, your structured and vector data files live in your R2 bucket.
DuckDBIllustrative
-- query your open Parquet directly
SELECT product, sum(revenue) AS rev
FROM read_parquet('orders/*.parquet')
GROUP BY product ORDER BY rev DESC LIMIT 3;

┌────────────────┬──────────┐
│ product        │ rev      │
├────────────────┼──────────┤
│ Kitchen Bundle │ 48,210.00│
│ Travel Mug     │ 31,884.50│
│ Spice Set      │ 22,019.00│
└────────────────┴──────────┘
your-bucket · Cloudflare R2Illustrative
$ aws s3 ls s3://your-bucket/genticdb/ --recursive
  orders/part-0001.parquet
  customers/part-0001.parquet
  embeddings/products.parquet
  reviews/part-0001.parquet
open Parquet · your storage · revocable
Operating modes

Hosted or bring your own bucket.

Gentic-hosted

Gentic manages storage and operation so your team can start quickly — no infrastructure to stand up. Live in production today.

Your Cloudflare R2Early access

Connect credentials scoped to your bucket through Gentic. Once your org is provisioned onto it, your structured and vector data files live there in open Parquet — inspect them, export them, or revoke access at any time.

Coming soon

Already hosted with Gentic? Moving your existing historical data into your bucket is coming soon.

Isolation and trust

Designed so one brand can never become another brand.

Every brand is bound to its own catalog namespace, storage location, and scoped credentials. GenticDB fails closed if a connection does not match the brand it belongs to, and customer-bucket connectivity is verified before going live. This is the technical proof behind the ownership promise.

Per-brand catalog namespace
Per-brand storage location
Scoped credentials
Fail-closed on mismatch
Technical architecture

DuckDB for speed, DuckLake for table state, open Parquet for ownership.

Agents
Gentic MCP
embedded DuckDB / DuckLake
Gentic-managed catalog
stays with Gentic
Open Parquet in R2
your storage (BYOB)

DuckDB gives agents fast SQL and semantic retrieval, DuckLake manages the lakehouse table state, and R2 stores the open data files. The catalog metadata stays Gentic-managed; your structured and vector data files are what live in your bucket.

Compounding memory

Every action makes the next agent smarter.

Work doesn't evaporate when an agent finishes. Each outcome lands back in the shared layer, so the next workflow starts from everything that came before.

Research insight
Creative brief
Generated asset / landing page
Performance outcome
Brand Brain / wiki update
Better next workflow
The platform

Three layers, one platform.

Gentic Computer

Keeps agents working until the job is finished.

Gentic MCP

Gives agents the tools to act.

GenticDB

Gives every workflow shared memory.

Gentic Computer thinks and finishes the job. Gentic MCP acts. GenticDB remembers.

Build any agent. Give it the data to succeed.

Stop rebuilding integrations, memory, and storage for every agent your brand creates. Give them one shared data layer built for e-commerce operations.

FAQ

Questions, answered.

What data can agents access through GenticDB?

Commerce and operational data, brand knowledge and memory, and the outputs agents create — read, written, and searched through the shared Gentic MCP interface. Structured data and embeddings live together, so the same store answers SQL queries and semantic search.

What does “own your data” mean?

GenticDB stores structured and vector data in open Parquet files. You can query them with DuckDB, export CSV / Parquet / JSON, and use standard object-storage tools. With bring-your-own-bucket, those files live in your Cloudflare R2, where you can inspect them and revoke Gentic's access.

What lives in my Cloudflare R2 bucket?

Your structured and vector data files, in open Parquet. Once Gentic provisions your org onto your connected bucket, that's where they live. The Gentic-managed catalog and media/file bytes are not in your bucket.

Does GenticDB replace Shopify or my operational systems?

No. GenticDB is the shared data and memory layer your agents read, write, and search across — not a replacement for Shopify or your source systems of record.

Can I query or export my data without Gentic?

Yes. The data files are open Parquet. Point DuckDB at them, export CSV / Parquet / JSON, or use standard object-storage tooling — no Gentic runtime required to read your files.

Can I move existing Gentic-hosted data into my bucket?Coming soon

Coming soon. Today, connecting a bucket means new structured and vector data is written there going forward. Migrating an existing hosted org's historical data into its bucket is in progress and not yet generally available.