The platform

Between your data and production, one agent-run platform

Sources in, pipelines out. The agent handles the middle.

How it works

You describe the outcome. The agent writes the pipeline.

Plain language in, version-controlled pipeline code out.

1
Chat or canvas
Type what you want, or drag nodes on the visual builder. Both stay in sync.
2
Real execution
Pipelines compile to Spark and Kafka jobs on the cluster your analysts already query.
3
Your models, your keys
Local Ollama or vLLM, including weights we built for you, or a hosted API on your own keys. No lock in, no per seat AI tax.

orders_cdc.pipeline

# generated by the agent, editable by you
source postgres_prod {
  table    = "public.orders"
  mode     = "cdc"        # change data capture
}

transform dedupe {
  by   = ["id"]
  keep = "latest"
}

sink warehouse {
  engine = spark
  to     = "analytics.orders"
}

What is inside

Everything you need to ship a pipeline

One platform instead of a stack of tools.

✦

AI pipeline builder

Go from a sentence to a running pipeline. Chat or drag nodes on the visual canvas, the agent keeps both in sync.

⛃

Connect anything

Databases, REST APIs, files, and Kafka streams. 50+ connectors with cross database ETL and CDC out of the box.

⚡

Runs on real infra

Pipelines execute on Spark and DataFusion with Kafka streaming, the same distributed compute your production jobs use.

◇

Agents and tools

Named, versioned agents with tool calling and the Model Context Protocol. Pick the right model per agent.

❖

Built-in memory

Agents remember your schemas, fixes, and decisions across sessions and workspaces, so they get better as you work.

▸

Deploy and monitor

One click deploy, live execution, quality scoring, and bottleneck detection with context aware fix suggestions.

Bring your own model

Studio serves whatever weights you already own

Studio points at an endpoint you control. No model vendor, no lock-in.

1
A model we built for you
If you commissioned one from Bonacci Foundry, this is where it goes to work: your weights on your vLLM or Ollama, driving the pipelines that supply it. The two are sold separately and neither requires the other.
2
An open checkpoint
Llama, Qwen, Mistral, or whatever you already run. Good enough for a great deal of pipeline work, and it costs you nothing but the GPU it sits on.
3
A hosted API, if your data is allowed out
OpenAI, Anthropic, Gemini, Groq, on your own keys. Not everyone is air-gapped, and Studio does not pretend otherwise. No per-seat AI tax either way.

application-onprem.properties

# the real config. no hidden endpoints.
deployment.mode=onprem

# inference points at your box, not ours
OPENAI_BASE_URL=http://vllm.internal:8000/v1
OPENAI_MODEL=acme_9b   # your weights

# no billing calls, no licence server
razorpay.enabled=false
paddle.enabled=false

# your directory, your mail, your storage
LDAP_ENABLED=true
MAIL_HOST=smtp.internal

Built for real work

From a quick load to streaming in production

ETL

Database pipelines

Cross database moves, change data capture, dedupe, and transforms. Postgres, MySQL, and the warehouse, wired in minutes.

Streaming

Real-time on Kafka

Consume events, transform on the fly, and sink continuously to the warehouse. Backpressure and recovery handled for you.

Files and APIs

Ingest from anywhere

CSV, Parquet, and JSON files, plus REST endpoints on a schedule. Paste, upload, or let an agent fetch it for you.

Demos

See it in action

Real, interactive walkthroughs that run right in the browser. No signup required.

studio.bonacci.thinkingdbx.com/demo · full product flow

Connect sources, design on the canvas, run the agent, and ship.

Design partner program

Run Studio on your real pipelines. Help shape what we build.

We are onboarding a small group of design partners: teams who use Studio on production work, meet with the founders monthly, and get lifetime free access. Limited seats.

For the security review

The questions your architecture team will ask

Answered up front, so the review starts from facts instead of a questionnaire. The full security page goes further, including what we do not claim. Send it to your architecture team.

Where does it run?

A Docker Compose stack on infrastructure you own: VPC, private data center, or an isolated network. Images can be mirrored into your own registry.

What talks to the outside?

In a self-hosted deployment, nothing in the install path requires a call to us. Inference defaults to a local endpoint, and payment processors are disabled entirely.

Where does the AI run?

Against your model server: Ollama or vLLM on your GPUs. If you choose to permit egress, you can point it at a commercial API with your own key instead.

How does licensing work?

A signed licence key validated offline against a signature on your machine. There is no licence server, no activation call, and no usage metering.

Who can log in?

Your directory. LDAP / Active Directory / OIDC single sign-on, with local accounts optional and self-registration disableable.

What about our data at rest?

Your PostgreSQL and your object storage, on your infrastructure, under your encryption, backup, and retention policy.

Pricing

Two ways to run Studio

Self-host on an annual licence, or try it in our cloud.

Recommended for regulated teams

Sovereign Self-Hosted

Runs entirely on your infrastructure. Your data, schemas, prompts, and model weights never leave your perimeter.

from$60,000/year

Annual licence · priced on deployment scale and support tier

Book a deployment call

✓
No quotas. Unlimited pipelines, connections, and executions. It is your hardware.
✓
Local inference. Ollama or vLLM on your own GPUs. No third-party model in the data path.
✓
Offline licence. Signature-validated on your machine. No licence server to reach.
✓
Your identity provider. LDAP / Active Directory / OIDC single sign-on.
✓
Your storage and SMTP. Object storage and mail stay on infrastructure you control.
✓
Security review support. We work directly with your architecture and security teams.
✓
Named support engineer and an agreed response SLA.

Or evaluate in our cloud

Hosted by us on our infrastructure, with self-serve plans from free.

See cloud plans

Self-hosted licences are invoiced annually. No payment processor, no metering, and no billing callbacks run inside your network.

See it live

Watch an agent build a real pipeline in 5 minutes

No slides. We describe a pipeline and ship it to Spark while you watch.

Book a deployment call Try the hosted version

0:30

Describe the pipeline

Plain English. Source, transform, destination.

2:00

The agent builds and tests it

Generates the pipeline, runs it against sample data.

4:30

Deployed and querying

Running on Spark and Kafka, rows landing in the warehouse.

The agentic data engineering
platform that runs inside your walls

Between your data and production, one agent-run platform

You describe the outcome. The agent writes the pipeline.

Chat or canvas

Real execution

Your models, your keys