MindStudio

Invoke your published MindStudio agents from FlowRunner flows, synchronously or via async callback, passing launch variables in and returning structured output for downstream steps.

3 actions API key available

MindStudio website ↗ Platform Documentation ↗ Capability data verified 2026-07-27

A batch of content briefs reaches ready status in the tracking table

Load Workspace verifies the key is connected to the expected MindStudio workspace

Run Agent executes the copy agent on one sample brief, returning output and billing cost

The flow projects total batch cost from the sample and counts the queue

The owner reviews the sample output and approves the projected spend before the batch runs

Run Agent Asynchronously fans out the remaining briefs with a callback URL

Callback results write back to the source table as they complete

The completion summary posts with run counts and total cost

What This Integration Enables

This is the connector that makes FlowRunner's thesis concrete: agent builders and orchestration are different layers. MindStudio is where your team builds and publishes agents; FlowRunner is where those agents become steps in governed business processes, invoked with launch variables from upstream data, returning structured output that downstream steps act on, wrapped in the retries, approvals, and audit trail an unattended process needs. Companies will run agents from many builders. The coordination problem, which agent runs, on what input, at what cost, with whose approval, is the layer above, and it is the layer FlowRunner owns. The relationship is complementary by design: a well-built MindStudio agent is exactly the kind of component worth orchestrating, and no agent should be the one governing its own spend. - Run published MindStudio agents inline and use their structured output in downstream steps - Kick off long-running agents asynchronously and receive results at a callback URL instead of blocking - Pass dynamic launch variables, topic, tone, customer data, from earlier workflow steps - Target a specific named workflow inside an agent when one app holds several - Track billing cost per run and gate batch spend before it happens - Verify workspace identity before any run, so the wrong key fails loudly at the start

Without FlowRunner

Agents live where they were built The MindStudio agent works beautifully in its own tab and touches nothing else in the business

Batch work means babysitting Someone runs the agent input by input, pasting results into the tracker between meetings

Agent spend is discovered on the bill Runs accumulate across the team with no checkpoint between one execution and a thousand

With FlowRunner

Agents become workflow steps Published MindStudio agents run inside FlowRunner flows, fed by upstream data and feeding downstream systems

Batches run themselves Async runs fan out with callbacks, and results file into the systems of record without a human courier

Spend is a gate, not a surprise A sample run prices the batch, and a person approves the projection before the fan-out

Use Case Scenarios

The content batch that ships itself

Briefs accumulate in [Airtable](/integrations/airtable). When the batch is ready, the flow runs one brief through Run Agent, posts the sample and the projected cost, and on approval fans out the rest through Run Agent Asynchronously. Callbacks write each result back to its source row with the thread ID for traceability, and the editor reviews finished drafts in the table they already work from, not in a separate tab.

The channel assistant with a specialist behind it

A request lands in [Slack](/integrations/slack). The FlowRunner flow classifies it, and requests matching the research agent's specialty pass to Run Agent with the question and context as launch variables. The structured result posts back to the thread with sources. The MindStudio agent stays a focused specialist; FlowRunner handles the routing, the fallback when the agent errors, and the escalation to a person when confidence is low. When the answer comes back thin, the flow says so and pages a person with the thread ID, rather than letting a confident-sounding non-answer stand.

Row in, verdict out, on a schedule

New rows in [Google Sheets](/integrations/google-sheets) carry inbound partnership applications. Each row's values pass as launch variables to Run Agent, whose published workflow scores fit and drafts a recommendation. Results write back beside the input, and applications scoring above threshold open a review task for the partnerships lead. Load Workspace runs at flow start, so a rotated key or wrong workspace stops the run before the first application is touched.

Human-in-Loop Highlight

The costly operation in this connector is the fan-out. Run Agent Asynchronously accepts work as fast as a flow can submit it, every accepted run bills against the MindStudio workspace, and a loop over a big table is a spend commitment made in seconds that cannot be recalled once the runs execute. The failure mode is mundane: the filter that was supposed to select 40 ready briefs selects 4,000 rows, and the agent dutifully processes all of them. So FlowRunner batch flows price before they run. The agent executes exactly one representative input through Run Agent with the billing cost included in the response, multiplies by the queue size, and posts both the sample output and the projected total. The owner judges two things at once, is the quality worth it and is the number right, and only an approval releases Run Agent Asynchronously on the rest. The thousand runs are machine work; the decision to buy them belongs to the person who answers for the invoice.

Agent processes routinely

Detects exception requiring judgment

Clear match Continues automatically

Ambiguous Routes to human via preferred channel

Human decides

Agent resumes with decision

Agent Capabilities

3 actions

Execution

Run Agent Runs a published MindStudio agent synchronously and returns its structured output, with launch variables mapped to the agent's defined inputs, an optional named workflow target, and an optional billing cost in the response. Returns the result plus a thread ID and full thread object for follow-up. Blocks until the agent finishes, so long runs belong on the async path.
Run Agent Asynchronously Starts a published agent run and delivers the final structured output, result, thread, and optional billing cost, to your callback URL as JSON when it completes. Returns immediately on acceptance. The fan-out engine this page's human gate prices before it runs.

Workspace

Load Workspace Loads the workspace context for the configured API key, returning the organization ID and name. The preflight that proves the key is valid and pointed at the expected workspace before any run bills against it.

Frequently Asked Questions

What can FlowRunner do with MindStudio?

FlowRunner agents can run Run Agent, Run Agent Asynchronously, and Load Workspace in MindStudio.

Does connecting MindStudio to FlowRunner require OAuth?

No. MindStudio connects to FlowRunner with an API key, no OAuth flow required.

Can MindStudio trigger a FlowRunner workflow automatically?

MindStudio doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.

Start building with MindStudio

$100 in credits. No card required. Connect in minutes.

Start Building Free Talk to us