Sample management sits at the operational heart of any high-throughput lab. When it works seamlessly, science moves. When it doesn't, data gets stranded, errors accumulate, and scale becomes a liability rather than an advantage.
In a recent joint webinar, Russell Green and Crystal McKinnon (HighRes) joined Alan Parkin (Cenevo) to walk through how the integration of Cellario OS™ and Mosaic is changing what's possible for modern sample management operations (Video 1).
Video 1. This recent webinar with HighRes and our software partner at Cenevo shows how the integration of Cellario OS and Mosaic transforms sample management in high-throughput labs, enhancing efficiency and data integrity.
The Problem: Three Failure Modes in Sample Management Labs
Russell opened by clustering the core challenges he sees across labs into three categories. The first is fragmented handovers: manual transitions or brittle custom scripts that fill the gaps between sample management software and physical execution. These gaps don't just slow things down; they erode data provenance and strip away the metadata that increasingly matters for training AI models on experimental outcomes.
The second challenge is the absence of a true single source of truth. In practice, sample data gets copied into planning spreadsheets, transfer lists, and instrument-local files. Small discrepancies compound into major traceability problems over time.
The third, and perhaps most pressing, is scale. Labs are managing larger and more diverse sample collections than ever before, handling hundreds or thousands of concurrent requests across multiple instruments, with shifting priorities. Without intelligent scheduling and orchestration, that kind of workload is simply unmanageable.
Cellario OS: The Orchestration Layer
Russell introduced Cellario OS as the connective tissue between digital intent and physical execution, a cloud-based orchestration platform that communicates directly with standalone devices (via a portfolio of over 500 drivers), integrated systems (through Cellario Scheduler™), and partner platforms like Mosaic by Cenevo. Every function in Cellario OS is built API-first, meaning anything visible in the UI is also callable through a RESTful API.
For sample-driven labs specifically, Cellario OS enables real-time sample tracking at the plate and well level, intelligent workflow selection based on request type (cherry pick, dose-response curve setup, and more), and data continuity from job submission through execution and back, including resilience against network interruptions between cloud and physical lab environments.
Live Demo: End-to-End Cherry Pick Workflow
Crystal walked through a live demonstration of the full integration in action, starting from an open order in Mosaic and following it all the way through to well-level results.
From Mosaic, she selected a high-priority cherry-picking order and created a machine job, assigning it to the HighRes instance of Cellario OS. The job immediately received a unique identifier (Job 281) that carries across both platforms throughout the entire workflow. Moving to Cellario OS, all sample-level source and destination information had already been passed from Mosaic automatically. No manual data entry required. Crystal scheduled the run on the lab calendar, reserved the time slot, and initiated execution on the physical system through Cellario Scheduler.
Back in Cellario OS, the run tracked in real time: percent of plates processed, percent of wells completed, failed transfer counts, and a live feed of individual transfer events. Once the run finished, the completion trigger pushed all results,including file attachments and transfer logs, back to Mosaic automatically. Within Mosaic, the order updated to completed, with full plate maps and well-level metadata (sample amount, concentration, solvent, and more) accessible without any manual intervention.
Looking Ahead: Agentic Labs
Both Alan and Russell pointed toward where this integration is heading. Alan previewed Mosaic's development of MCP (Model Context Protocol) APIs, which will allow natural language interaction with Mosaic from any connected platform, submitting sample requests, querying inventory, initiating orders, and creating housekeeping tasks through conversational interfaces or third-party agent frameworks.
Russell demonstrated a preview of HighRes' Cellario Lab Assistant™, a natural language interface that already integrates with the Mosaic MCP server. In the demo, Lab Assistant fielded queries like "I found this plate. What is its provenance?" and "Create a list of failed transfers for reprocessing," delegating each to the Mosaic MCP server and returning structured responses, all without leaving Cellario OS.
Russell framed this as the first stage of a broader trajectory toward AI-driven scientific intent, where agents assist with experimental design, translate intent into executable workflows, and enable intelligent orchestration, ultimately driving closed-loop experimentation where each outcome informs the next experimental iteration.
Interested in seeing the integration in your own lab? For more information on how HighRes can support your lab orchestration needs, get in touch with us.
