Frequently Asked Questions

Tilebox is a distributed computing and data workflow framework for space and geospatial applications. It combines a searchable metadata catalog with workflows executed by runners in environments you control. Payload files can remain in their original storage while Tilebox stores their metadata and references.

Run compute near your data to reduce unnecessary transfers and process large datasets where they already live. Start with one workflow, then add runners to process independent tasks in parallel as your workload grows.

Tilebox is built for architects, engineers, and data scientists building and managing data pipelines in space, defense, and geospatial intelligence. Teams use the same task-and-runner model on a laptop, in the cloud, and on premises. You can develop locally and scale across environments without redesigning how your workflow is coordinated.

Tilebox is a transatlantic team based in Europe and the United States. We work remotely across these regions.

Yes. Labs is $0/month with no credit card required. It includes one user, 100 MB of catalog metadata, 1 GB of compressed telemetry, 24 task-hours per month, and unlimited parallel runners. Labs has a 1 GB monthly query-egress limit and seven-day job retention, with no usage overages.

CRON and storage-event automations require a paid plan. A standalone edge-compute license is a paid-plan add-on. You remain responsible for the compute and storage resources you use. See the pricing page for the full comparison.

No. If you can write Python or Go, you can build space data pipelines. Tilebox abstracts the complexity of infrastructure and data management so you can focus on building.

You can add Tilebox alongside your existing system. Use the catalog to organize your existing data and intermediate results, and find inputs for the next step. Use Tilebox Workflows to coordinate task dependencies, run work across environments, and inspect logs and traces in one place.

Start with one workflow where those capabilities would help. Keep the parts of your Airflow, Argo, Kafka-based system, or local scripts that already meet your needs. Expand when the results justify it.

Yes. Ingest and query metadata with the datasets SDK while your existing scripts or orchestrator continue processing files. For example, catalog raster layers by footprint and acquisition time, with custom fields such as processing version, cloud cover, or quality scores. Query matching records, then read their referenced files from your own application. You do not need a Tilebox workflow runner to query a catalog.

You maintain the ingestion code, schema, and storage credentials. Catalog metadata is separate from the imagery itself, and catalog-only adoption does not imply a separate pricing plan. See how cataloging your holdings works and the plan allowances.

You can introduce Tilebox for a single workflow without moving all your payload files into Tilebox. The catalog stores metadata and references to files in external storage.

Start with code that already works, even a script processing a small region. Onboard your agent with the Tilebox CLI and skills. The skills give your coding agent Tilebox-specific guidance to turn that code into tasks, split independent work for parallel execution, configure runners, and test the workflow. Start with a small input, check the results, then expand to more data or processing steps.

Yes. Start independently with the documentation and sample workflows, or work with our engineers on setup and architecture. We also offer fixed-price migration projects, with the scope and price agreed before work begins. Tell us about your existing pipeline to discuss a migration.

No. Introduce Tilebox for a catalog, a single workflow, or one part of a pipeline. Keep your existing system responsible for the rest. Agree which system owns submission, retries, and completion status before connecting two orchestrators.

You can run all workflow execution on infrastructure you control. Runners connect out to the Tilebox API to claim tasks and report results; Tilebox does not need inbound network access to your runners. Imagery and output files can remain accessible only from your environment, using credentials you manage.

With the hosted service, Tilebox stores the catalog metadata and file references you ingest, workflow state, task inputs, logs, and traces. You choose what to include in those records.

For Python workflows, release runners simplify deployment: publish and deploy a release in Tilebox, and your runners pick up deployment changes automatically. Tilebox stores the code artifacts you publish for those releases. You can also use direct runners with your own CI/CD pipeline and deploy code yourself, without uploading workflow artifacts. Compare the two runner modes.

On-premises deployments and a headless daemon are also available. These let you deploy Tilebox beyond the hosted service. Talk through your deployment with us to choose the setup for your network, data-residency, and security requirements.

You can fix a failed workflow and retry the job without rerunning the tasks that already completed. Use the logs and traces to find the problem, deploy a compatible code fix, and restart the failed work. For a long processing job, keeping completed results saves both time and the compute cost of doing that work again.

Runners send regular heartbeats. If a runner stops responding, Tilebox can retry its task on another compatible runner, subject to retry limits. Make file writes and other changes safe to repeat, since a task may run more than once. See how to retry with a code fix.

Keep your current schedule and delivery steps. When it is time to process data, your application submits a Tilebox job through the SDK. Tilebox runners do the processing and write results to your chosen storage. Your application checks when the job is done and uses those results in the next step.

You can start by wrapping one existing script in a Tilebox task. Decide which system submits the job and handles retries so work is not submitted twice. When a customer starts a job from your app or API, this is the Customer-Triggered Workflows pattern in our six workflow archetypes: processing runs in the background while your app stays responsive.

Start with one workflow and a result you want to improve: less time finding data, easier recovery from failures, faster processing, or less time operating the pipeline. Use your existing code and a representative set of inputs to test that result with Tilebox.

Compare output quality, runtime, operator effort, and total cost with your current approach. The workload cost example separates Tilebox fees from compute, storage, and transfer costs. Our team can help scope the evaluation and a fixed-price migration.

Your processing code and source files can stay under your control. Keeping the processing code separate from Tilebox task definitions makes it easier to reuse. If you later change systems, you can reuse that code while adapting the task definitions, catalog queries, and telemetry integrations.

Tilebox gives your team a system that keeps improving as you build on it. We develop and maintain the catalog, storage clients, workflow engine, and developer tools together, so your team can spend more time on its processing methods and products.

Good data-access practices are built into the tools. The storage client selects a compatible file location, supports streaming large files, and lets you read a region from a Cloud Optimized GeoTIFF without downloading the whole image. You get these capabilities through one interface across supported storage providers.

Workflows provide task scheduling, parallel execution, retries, and logs. Start with a small region or a few files, then add compatible runners to process independent tasks in parallel as the workload grows. Your team can build on these patterns without designing and maintaining its own storage and coordination layers.

Tilebox connects finding data, reading files, running your processing code, and inspecting the results. For example, query scenes over an area and time range, read the relevant image regions with the storage client, and process each scene as a workflow task. Logs and traces show how the processing ran.

Write the outputs to your storage, then catalog their metadata so your team can find and reuse them. Your application can use those outputs in a map, report, or API. You bring the processing methods; Tilebox provides the data access and coordination around them. Start with the data catalog and connect it to a repeatable workflow.

Small teams can start with public satellite data and the free Labs plan. Search the catalog by area, date, and fields such as cloud cover to find observations relevant to your research. This gives you a practical starting point for environmental monitoring, climate research, and public-interest reporting.

The Sentinel-2 image guide shows how to find a low-cloud scene and read only the part of its image bands you need. You can build on that example with your own analysis rather than downloading a whole archive first.

Explore the available datasets and start testing your analysis with the free Labs plan. Tell us about your research, the data you need, and what you want to produce. We've helped NGOs, think tanks, and non-profits move their projects forward and can work with you to find a practical way to get started.

Your workflow can grow with the data it finds. A task can query a catalog, create one processing task for each matching scene, and schedule a final step after those tasks finish. You express that logic in code instead of listing every scene in advance.

Tilebox tracks the dependencies and distributes ready tasks across compatible runners. Add runners to process independent tasks in parallel, or route work to a cluster with the data access or hardware it needs. The parallel workflow guide shows how the same workflow runs with one runner or several.

Keep the code, storage, and services your team already uses. Python and Go tasks can call your existing libraries, databases, and APIs. The Python storage client provides one interface for assets in local files, S3, Google Cloud Storage, Azure, and HTTP locations.

Run the processing on your laptop, cloud VMs, Kubernetes, or on-premises machines. Install the libraries your code needs and supply credentials through your deployment system. Write outputs where your application already reads them. See the deployment guide for direct runners, Docker, and Kubernetes.

Yes. Your AI tool can work directly with Tilebox datasets and workflows. For a coding agent with terminal access, install the Tilebox CLI and skills. The agent can build a workflow, run a test job, inspect logs, and improve the code using what happened in the run.

For a chat client without a terminal, use the Tilebox MCP server to query datasets, work with workflows, and search the documentation. Sign in with your Tilebox account and select the team to connect.