At the Nextflow Summit, we shared what's new in the latest release of Seqera Platform (version 26.2). The release was built around one goal: moving data through bioinformatics faster while keeping results traceable and controlled. Here's a recap of what we announced.
End to End Automation
The journey from data landing in the cloud, to processing with Nextflow, to interactive and collaborative analysis has many handoffs. Today, results get stuck in many places:
- →Runs waiting to launch. The path from data landing to a samplesheet to a pipeline launch can be long.
- →Manual debugging. Pipeline issues often need to be triaged and solved by hand.
- →Results outpacing tracing. Outputs pile up faster than they can be passed downstream.
Even as individual steps get optimized and automated, automating just one piece only moves the bottleneck. Results can always outpace your ability to analyze them unless you look at the journey as a whole. This Seqera Platform release brings updates across every stage of the data journey, from the moment data lands, through pipeline runs and data access, to the analysis of results.
The journey starts when data lands. Too often it then waits for someone to notice it, build a samplesheet and launch the right pipeline. In this release, we expanded Actions so that work can be started on new data immediately.
Event-Driven Actions
Actions can launch a pipeline or hand off to an AI agent automatically in response to events like changes in cloud storage, schedules, finished pipeline runs, repository pushes or API calls. When an Action hands off to an agent, the agent can create datasets, run pipelines or pass data to a Studio. Process new data as it arrives, run pipelines on a schedule, and let agents respond to run outcomes, without manual intervention. There's no external scheduler to maintain and no one needs to watch the bucket.
File, Dataset, and Schedule Triggers
Triggers are the events that start an Action. With a bucket event, an Action fires when a marker file arrives in cloud storage, so new data is processed as soon as it lands. With a schedule, an Action runs on a recurring cadence. Triggers based on a dataset change or update are coming soon.
A finished run is rarely the end of the work. It might need to feed the next pipeline or be opened up for analysis. If it failed, it needs to be debugged and relaunched. In this release, Actions and Agents handle that next step automatically, so you can move from run to run, and from run to analysis, without manual handoffs.
Pipeline-Status Triggers
With a pipeline run event, an Action fires when a run reaches a terminal state, such as succeeding or failing. Use it to chain pipelines without glue scripts or an external orchestrator. When Pipeline A succeeds, Pipeline B starts automatically. When it fails, the Action can hand off to an agent that debugs the problem and restarts Pipeline A, so every outcome has a defined next step.
Pipeline Run Actions
Actions don't stop at the next pipeline. They can also take you from run to analysis. When a pipeline succeeds, its results can automatically stand up in a Studio, so interactive analysis starts as soon as the run completes. No one has to wait for a notification, find the outputs and set up an environment by hand.
Scheduled Automations
Some work doesn’t start with an event. It runs on the calendar, reference databases as they refresh, waiting for data from multiple sources to land, or aggregating work into a daily review. For work like this, Action can now run on a recurring cadence you define, directly in Seqera Platform. It can launch a pipeline, such as rebuilding an annotation index weekly or re-running a cohort analysis as new samples accumulate. It can also hand off to a Co-Scientist Agent, for example to review the last 24 hours of runs, flag failures and cost outliers.
Co-Scientist Agents
Agents are reusable, named sets of instructions that Co-Scientist can run on your behalf in your workspace. Shared across the workspace, agents capture repetitive tasks so anyone can run them without rewriting the prompt. Agents run fully autonomously with no need to manage via the UI and can be linked to their own unique identity and permissions, providing full auditing integration and peace of mind for handling sensitive operations.
Users and Agent Get Specific Permissions
As you move faster, researchers need the right access in real time. As pipelines chain together and results flow straight into analysis, it matters more that every user and every agent has the right access. Access covers two things: the data people can reach and the features they can use. This release makes it easier than ever to get both right.
User-Level Data Permissions
Previously, each workspace leveraged shared credentials, and teams created workspaces to control data access. Now, when a user logs in, Platform uses their enterprise identity to get short-lived credentials specific to that user and workspace from the underlying data provider. Those credentials apply to everything the user does with the data, whether in Data Explorer, a pipeline run or a Studio. Every action is logged in Seqera Platform and per user on the cloud provider side, giving you auditing on both. Every user can get full access to features across Platform and these can also be controlled via fine-grained roles.
Project View
Not every user needs the full platform. Platform’s new Projects group the pipelines, datasets and runs that belong to a single piece of work together, giving scientists a simplified view without the deeper configuration, optimization, and troubleshooting. This makes it easy for more scientifically focused users to go directly to the analysis they need and self-serve analysis wherever possible. Set one up by tagging resources with a label, and scientists can focus on their work without the noise of the rest of the workspace, while keeping the same fine-grained data controls.
Service Accounts
Platform’s new Service Accounts now give agents and automation their own non-human identity, so automated work is attributed to the automation rather than to a person. Service Accounts are managed at the organization level and belong to the organization, not the user who created them. They have no password and cannot sign in, but can be defined with fine-grained access controls within a workspace. Looking ahead, we're working on more granular data access for agents, including controlling access to data, credentials, and other external services.
MCP Access
Many users want to work in their own tools. We've extended the Seqera MCP, enabling users to make their preferred AI tools a control center for Seqera Platform. The MCP connects deeply to Platform while preserving the same control over feature access and data access. Seqera MCP works with tools like Claude, Codex, and Benchling, so users can connect it and do their work with the right access, entirely in the interface of their choice.
Enterprise Ready Data Handling
Access is only useful if you can find the right data. And as teams move faster, they need to keep the security and validation that matter to them, whether their work is R&D or clinical. This release makes the right data quicker to find and brings interactive analysis inside validated environments.
Lineage Search
Data Lineage automatically captures Nextflow provenance records for every run, task and output file. Start anywhere: trace an output back through its task and run to its inputs, or follow an input forward to everything it produced.
New in this release, Lineage Search lets you find and trace these records across every workspace you have access to, instead of digging through workspaces one by one. Type a file name, run or pipeline, with suggestions as you go, or narrow results by type, label or workspace, then go straight into analysis.
Expanded GxP Support
Regulators need to be able to replay exactly what happened, under 21 CFR Part 11 and Annex 11. In this release, we've broadened our GxP support. GxP installations now support Studios, so interactive analysis can sit inside the validated environment. Moderna runs a GxP-validated deployment of Seqera Platform, implemented with ZS Associates, with full IQ, OQ and PQ qualification. This is what it looks like when the core runs at speed and stays provable.
Available Now on Seqera Platform
This release smooths the whole journey from data to analysis, not just one step of it. The result is a platform where data moves faster from landing to analysis, while every result stays traceable and controlled.
| Feature | What it enables you to do |
|---|---|
| Event-Driven Actions | Launch a pipeline or hand off to an AI agent automatically when something happens, with no external scheduler to maintain. |
| File, dataset and schedule triggers | Start Actions when a file lands in cloud storage or on a recurring schedule, with dataset triggers coming soon. |
| Pipeline-status triggers | Chain pipelines together when a run succeeds or fails, without glue scripts or an external orchestrator. |
| Pipeline Run Actions | Send results from a successful run straight into a Studio, so analysis starts as soon as the run completes. |
| Co-Scientist Agents | Save repeatable tasks as shared agents that run autonomously, such as debugging and relaunching failed runs. |
| User-Level Data Permissions | Give each user short-lived, identity-based credentials, with auditing in both Platform and your cloud provider. |
| Project View | Give scientists a simplified view of the pipelines, datasets and runs for a single piece of work. |
| Service Accounts | Give agents and automation their own non-human identity with fine-grained access controls. |
| MCP Access | Work with Platform from tools like Claude, Codex and Benchling, with the same feature and data controls. |
| Lineage Search | Find and trace provenance records across every workspace you have access to. |
| Expanded GxP Support | Run Studios inside GxP-validated environments, so interactive analysis stays within a full audit trail. |