Long-running jobs for apps and agents

Deploy reliable application logic and parallel workloads without managing queues, worker pools, or custom retry logic.

Get started Read the docs

A better way to build reliable

agentLoop()7.4

execute()3.7

execute()4.8

synthesize()2.5

const agentLoop = task({ name: 'agentLoop' },
  async (userInput: string, context: object) => {
    const actions = await callLlm(userInput, context)

    // Spawn a parallel task run for each planned action
    const results = await Promise.all(
      actions.map((action) => execute(action))
    )

    // Combine results into a coherent response
    return synthesize(userInput, results)
  }
)

const execute = task({ name: 'execute' },
  (action: { name: string }) => {
    const handler = getHandler(action.name)
    return { action: action.name, result: handler(action) }
  }
)

const synthesize = task({ name: 'synthesize' }, ...)

Turn functions into composable, async tasks

Define tasks and logic in code

Make any function an async task that runs in its own ephemeral container.

Retry failed steps automatically

Define retry logic declaratively to mitigate flaky steps and deliver more reliable experiences.

Checkpoint state (coming soon)

Make crashes inconsequential by resuming workflows right where they left off.

Scale without managing queues or workers

Sub-second start times

Render spins up new task workers in milliseconds to keep your experiences responsive.

Thousands of concurrent tasks

Run massively parallel workloads without managing worker pools or standing up queuing systems.

Powerful workers

Support compute-intensive tasks with high-spec machines and multi-hour run durations.

Observability & debugging

Every task. Every run. Full visibility.

See a unified history of active, queued, and completed task runs.

Analyze performance and reliability across time spans.

Inspect runs to optimize and debug.

“Render Workflows completely solved our scaling and simultaneous usage issues. They give us a reliable execution layer outside our core application for compute-heavy jobs. It’s been invaluable to have great visibility into runs and failures without building custom orchestration.”

Sebastian Franco, Software Engineer