Skip to content

How Carta Works

Carta is built around a four-stage workflow pattern that adapts to what your organization needs. The stages provide a consistent framework for how data and work move through the platform — but the specific tools, configurations, and integrations at each stage vary from deployment to deployment. Understanding this pattern helps you see how the individual components connect into a complete system.

The Four Stages

Stage 1 — Capture Intent

A participant with knowledge of the process fills out a form in Collections, recording the parameters, materials, instructions, or configuration for the operation. This isn't just data entry — the form captures provenance: who specified the operation, as part of which workflow, and in relationship to which prior and subsequent records.

What this stage answers: What needs to happen, and who said so?

Stage 2 — Convert Intent into an Artifact

When the form is submitted, a Dispatcher (or other post-submit automation) automatically transforms the form content into whatever artifact the next stage requires. Depending on the workflow, this might be:

  • A run-control file for a lab instrument
  • A configuration payload for a software system
  • A barcode or label for physical sample tracking
  • A routing instruction for the next step in the workflow

This transformation happens automatically and invisibly. The researcher fills out a form; the Dispatcher handles the translation between human intent and machine input — no manual file preparation required.

What this stage answers: What specific artifact does the executor need?

Stage 3 — Deliver to the Executor

The operational artifact is delivered to whoever or whatever will perform the work — a person, a machine, a piece of equipment, or a software system. DEXR handles this delivery securely, even when the sender and receiver don't share infrastructure, credentials, or data schemas.

The executor doesn't need to know anything about what happened before it received the artifact. The executor receives an encrypted reference, typically a presigned URL pointing to the artifact in client-controlled storage, and retrieves the artifact directly. The actual data never passes through DEXR.

What this stage answers: How does the artifact get to the right place?

Stage 4 — Return Results

After the operation is complete, results are returned to the coordination layer. Files and data artifacts are registered in CartaFS; structured outcome data is recorded back in FormsDB or your organization's configured data backend. The results become part of the searchable, traceable record — linked to the original intent, the operation, and the executor.

What this stage answers: What happened, and where do the results go?

A Concrete Example

Here's how the four stages play out in a materials testing environment:

  1. Capture: A materials engineer opens Collections and fills out a form specifying a tensile test — sample ID, testing parameters, and preparation notes.

  2. Convert: Submitting the form triggers a Dispatcher, which automatically generates a run-control file in the exact format the testing machine expects.

  3. Deliver: DEXR delivers the run-control file to the testing machine's control system using end-to-end encryption, without requiring the machine to be connected to Carta's infrastructure.

  4. Return: The testing machine completes the test and uploads its results. Raw output files are written to CartaFS; structured results are captured in FormsDB (or the organization's configured data backend) — both linked back to the original sample record.

The engineer never had to manually translate a form into a machine file, copy data between systems, or manage the handoff. Carta handled the coordination; the engineer handled the science.

Built to Adapt

No two Carta deployments look exactly alike. The four stages are a framework, not a script. Which components are used at each stage — and how they're configured — depends on your organization's infrastructure, workflows, and goals.

A deployment in a clinical research environment looks different from one in an advanced manufacturing facility. The Capture stage might use Collections, Workflow Manager, or a custom integration. Conversion might involve a Dispatcher or a direct service call. Delivery might go through DEXR, an internal system, or a direct API. Results might flow back into FormsDB, an existing database, or both.

Specific deployment scenarios and examples will be covered in a dedicated section of this documentation.

Why This Pattern Matters

The four stages are not specific to materials science. They describe the structure of any federated operation across any domain — clinical trials, industrial automation, multi-institution research collaborations, AI-assisted analysis workflows.

The key design property is decoupling: each stage is independent. An organization can integrate at whichever boundary is relevant — connecting a single instrument, plugging in a data pipeline, or deploying the full platform — without adopting the entire stack at once. The executor at Stage 3 doesn't need to know or care about what happened in Stages 1 and 2; it just needs the artifact. Similarly, Stage 4 returns results without requiring the executor to know where they'll end up.

This decoupling is what makes Carta composable across organizational and technical boundaries without requiring shared infrastructure at every seam.