Transloadit

Kick off Transloadit Assemblies that import media, run it through encoding Robots to resize, transcode, or convert, then poll for the finished files and their URLs. Agents also manage the reusable Templates that hold Assembly Instructions, with every request HMAC-signed.

7 actions Custom auth available

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

A new product with raw imagery is published in the storefront

Get Template pulls the saved product-image Assembly Instructions so the agent runs the versioned workflow, not an improvised one

Create Assembly imports the source image by URL and runs the resize and optimize Steps

Get Assembly polls until the status reads ASSEMBLY_COMPLETED and collects the output URLs

The agent compares output dimensions and file sizes against the spec for the storefront theme

A reviewer approves the rendered set before it replaces the live product imagery

The finished URLs and encoding results post to the team channel with the assembly ID for the audit trail

What This Integration Enables

Transloadit turns media processing into declarative jobs: an Assembly imports files, runs them through encoding Robots, and stores the results. The operational problem it leaves open is orchestration, deciding when jobs run, verifying they finished, and keeping the instructions consistent across every flow that encodes media. That is the layer FlowRunner supplies. Agents kick off Assemblies from real business events, poll them to completion, route the output files onward, and treat saved Templates as the single source of truth for how your media gets processed. - Resize, crop, and optimize images imported from public URLs, storing results in your cloud storage - Transcode video and audio into web-friendly formats as part of a larger workflow - Centralize encoding logic as saved Templates and invoke it by ID from any flow - Poll asynchronous jobs to completion and hand the output URLs to the next system - Audit encoding activity by listing and filtering past Assemblies Encoding is metered work, so FlowRunner's [human-in-the-loop](/concepts/human-in-the-loop) gate stands in front of the runs that are big enough to show up on an invoice. Routine single-file jobs flow autonomously; the library-wide runs wait for a person who has seen a sample of the output and the size of the bill.

Without FlowRunner

Media processing as tribal knowledge The resize settings live in one engineer's head and a script nobody wants to touch

Encoding jobs vanish into the void A job gets kicked off, and whether it finished, failed, or is still burning credit is unknowable without logging in

Batch re-encodes are a leap of faith Someone points a script at the whole media library and hopes the bill and the output are both fine

With FlowRunner

Workflows are Templates Assembly Instructions live as named, versioned Templates that any flow invokes by ID

Every job is tracked to completion Agents poll Get Assembly, capture result URLs and metadata, and log the outcome

Big runs get priced and approved The agent sizes the batch, a person approves the spend, then the Assemblies fire

Use Case Scenarios

Product images that ship themselves

A new product lands in [Shopify](/integrations/shopify). The agent calls Create Assembly referencing the saved product-image Template, which imports the source image and produces the storefront's required renditions. Get Assembly polls until completion, and the output URLs, dimensions, and file sizes come back grouped by Step. The agent verifies the set against spec and posts it for a quick approval before the listing goes live with optimized imagery.

The media pipeline with an archive

When an Assembly completes, the agent walks the results object and hands each output URL to [Amazon S3](/integrations/aws-s3) for archival in your own bucket. List Assemblies runs as a weekly reconciliation, catching any job that errored without a notification arriving. The Transloadit job record, the assembly ID, and the final storage keys all land in a log row, so a year from now the question "where did this rendition come from and what settings produced it?" has an answer: this Template version, this Assembly, this date.

The spreadsheet-driven batch re-encode

A sheet in [Google Sheets](/integrations/google-sheets) holds five hundred legacy video URLs that need transcoding to the new format. The agent reads the rows, resolves the transcode Template with List Templates, and sizes the job: five hundred Assemblies against metered encoding. That summary goes to the owner for approval. On the yes, the agent creates Assemblies in controlled batches, polls each with Get Assembly, retries the stragglers, cancels any that hang with Cancel Assembly, and reports progress to [Slack](/integrations/slack) until the sheet is fully processed.

Human-in-Loop Highlight

Create Assembly spends metered encoding capacity the moment it fires, and a batch run multiplies that spend by every row in the source list. Point a re-encode at your full media library with the wrong Step parameters and you pay twice: once for the bad run, once for the corrected one, and Cancel Assembly can only save the jobs still executing. So FlowRunner agents treat any batch beyond routine single-file processing as a proposal, not an instruction. The agent reports the file count, the Template it will reference, and a test result from one representative file, then waits. The owner checks the sample output, approves, and the batch runs at machine speed with the agent watching every job to completion. The spend happens once, on purpose, with a person's name on the decision.

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

7 actions

Assemblies

Create Assembly Starts an encoding job by submitting Steps inline or referencing a saved Template, importing source files from public URLs. Returns immediately with an assembly_id and executing status. The metered operation this page's human gate exists for.
Get Assembly Retrieves an Assembly's current status and, once complete, the results object with output file URLs, dimensions, and metadata grouped by Step. The polling step that turns an async job into a verified outcome.
List Assemblies Lists Assemblies most recent first, filterable by keyword, outcome, and created-date range. Used for auditing encoding activity and reconciling what ran against what was expected.
Cancel Assembly Cancels an Assembly still executing, stopping in-progress work and further billing for that job. The brake an agent pulls on hung or mistaken runs.

Templates

List Templates Lists the reusable Templates saved on your account with pagination and keyword search. Used to resolve the right workflow before creating Assemblies.
Get Template Retrieves a Template's name and full Assembly Instructions. Used to inspect exactly what a workflow will do before invoking or updating it.
Create Template Creates a named, reusable Template from an Assembly Instructions body. Names must be unique. The step that turns a proven one-off job into centralized, versioned encoding logic.

Frequently Asked Questions

What can FlowRunner do with Transloadit?

FlowRunner agents can run Create Assembly, Get Assembly, and List Assemblies in Transloadit, plus 4 more actions.

Does connecting Transloadit to FlowRunner require OAuth?

Transloadit uses a custom authentication method to connect to FlowRunner.

Can Transloadit trigger a FlowRunner workflow automatically?

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

Start building with Transloadit

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

Start Building Free Talk to us