Skip to main content
From tool call to process

Make ERP actions follow the business process, not a single tool call.

DataTether workflows chain generated tools, conditions, transforms, user inputs, review screens, and isolated code so each step has an owner and a boundary.

Request guided access

Workflow path

Execution sequence

review point
01

Input

02

Read ERP context

03

Branch policy

04

Show review UI

05

Execute approved action

Nodes turn policy into an execution path.

Each node has a job: read records, transform data, branch on policy, ask for input, show a review surface, or execute an approved action.

Tool

Call a generated OData, composite, or MCP tool.

Condition

Branch on field values, computed state, amount, owner, or risk.

Transform

Reshape data between steps without moving into code.

Input

Collect values from the user before a step runs.

LLM

Format context for model-assisted steps.

UI node

Show a dashboard, form, or review surface and wait.

Code

Run isolated JavaScript for procedural reshaping.

Authoring flow

01Create a workflow for an instance
02Add nodes and edges on the canvas
03Bind tool nodes to entity or function operations
04Configure mappings, conditions, transforms, and UI sections
05Enable the workflow as an MCP tool

MCP connection flow

01 CallMCP client invokes the workflow tool.
02 ResolveWorkflowEngine resolves inputs, node order, and mappings.
03 ExecuteTool nodes call underlying OData operations.
04 BranchCondition nodes choose the next path.
05 ReturnStructured content and optional app resources come back.

Real workflow examples

Workflows that look like work your SAP team already does.

Blocked orders

Fetch blocked records, show a review UI, capture an approval note, prepare a controlled update.

Procurement update

Read supplier context, branch on risk, request confirmation, execute the approved change.

Customer risk action

Join customer and invoice context, summarize risk, route the next step to a reviewer.

Review happens before the write step runs.

A pending change is held with an expiry window. The reviewer sees exactly what will run — the operation, the target, the arguments, and any risk context. The platform verifies that nothing changed before executing.

No-code

Assemble tool, condition, input, and UI nodes on a canvas.

Low-code

Configure field mappings, OData filters, transform blocks, and conditions.

Code

Use isolated code nodes for procedural reshaping and calculations.

One binding model, two surfaces

Dashboards and workflow review screens share the same widget binding model. A UI built once works as a standalone dashboard and as a node inside a running workflow — no duplication, no divergence.

Automate one workflow

Turn a manual process into a governed, repeatable AI workflow your team can trust.

One workflow. Governed from step one.

Pick one approval chain, exception queue, or multi-step operation. We help you build it as a visual workflow with scoped steps, human gates, and a full audit trail — live by the end of the week.

Months of dev time eliminated

Auto-generated tools replace weeks of custom MCP server development.

No infrastructure to run

Fully managed. Nothing to install, host, or maintain.

Instant dashboards and UI

Live tables, KPIs, and review screens — no frontend developer needed.

One setup, every AI client

ChatGPT, Claude, Manus, custom agents — all from the same connection.

A good first step is small: one service, one user outcome, one visible control boundary.

Request guided access