Category Archives: Sophia AI

Security is one of the first topics that comes up in almost every demo, evaluation, and conversation we have with labs considering agentic AI. The moment an AI can act, not just answer questions, a different question takes over: “Is it safe?”

It’s the right question to ask. Laboratory systems hold sensitive operational data and support workflows where permissions, traceability and accountability matter. Introducing AI into those workflows requires more than useful capabilities. It requires clear boundaries around what AI can access, what it can do and how its actions are recorded.

Sophia AI is built to operate within the existing security and governance model of QLIMS. Its interactions with the system are governed by authentication, user permissions and audit trails, helping laboratories explore AI-assisted workflows without treating security as an afterthought.

Here’s what that means in practice. 

Authentication & Access Control

• Zero credential exposure. You never provide a username or password to Sophia. Authentication is handled automatically and securely in the background through tokenized sessions.
• Role-based access control (RBAC). Every action Sophia performs is governed by your existing QLIMS permissions. She can only do what your account is authorised to do, nothing more.
• Session-scoped. All API interactions are tied to your authenticated session, preventing any unauthorised access.

Data Security

• No data storage. Sophia does not store, log, or retain your laboratory data, sample information, or credentials between sessions.
• Encrypted communication. All interactions with the QLIMS Action API occur over encrypted channels (HTTPS/TLS).
• Audit trails. Every action performed in QLIMS, whether by a person or by Sophia, is fully logged and traceable in the system’s audit trail, ensuring complete accountability.

Compliance & Governance

• 21 CFR Part 11 / GxP ready. QLIMS is built for regulated laboratory environments, so every action Sophia takes is auditable and traceable.
• No autonomous overreach. Sophia follows strict directives. She never creates, modifies, or deletes records without your explicit instruction, and she always asks permission before creating new identifiers or entities.
• Transparent actions. Before executing any significant operation, Sophia walks you through exactly what will be done.

Security is part of making AI useful

AI has the potential to reduce administrative work and help laboratory teams interact with their systems more efficiently. But capability alone isn’t enough. Laboratories also need confidence in how AI accesses information, performs tasks and leaves a traceable record of its activity.

Sophia AI brings AI-assisted workflows into QLIMS’s existing security and permission framework, helping teams explore what AI can do while keeping access, control and accountability central to the process.

Curious how Sophia AI could support your laboratory workflows? Contact us for a personalised walkthrough and explore how its capabilities could fit your lab’s needs.

We’ve previously announced that Model Context Protocol (MCP) is now available in Sophia AI. And we also mentioned it’s an open standard that allows AI applications to connect with external systems and information.

But if you work in a laboratory, your first question is probably: “What does this actually mean for my lab?”

What matters is what greater AI connectivity could mean for the way your team works with laboratory information.

In this article, we’ll look at what AI connectivity means in practical, and how Sophia can help your team not only find information, but take action in QLIMS.

What does AI connectivity mean for your lab? 

AI tools have become increasingly good at understanding questions, summarising information and generating responses.

But there is an important limitation: an AI model cannot automatically know what is happening inside your laboratory. Your laboratory data still lives in the systems your team uses every day, including your LIMS, instruments, databases and other software.

This is where MCP, or Model Context Protocol, comes in.

For a laboratory, that means AI can move beyond simply talking about information and start working with the information your team already uses. 

      1. What if you could spend less time searching for information? 

Laboratory teams work with large amounts of information every day. Finding one answer can sometimes mean moving through records, filters, screens and different parts of a system before you have what you need.

What if you could simply ask?

“Which samples are still waiting for review?”

“Show me the batches associated with this order.”

“Which methods were used for these samples?”

Instead of changing how your laboratory stores information, the goal is to make that information easier to access.

With Sophia, you can ask questions in natural language and get information from your QLIMS data without having to manually search through every record. But finding information is only part of the story.

       2. What if Sophia could understand the context around your question? 

Laboratory information rarely exists on its own. A sample can be connected to an order. An order can be connected to a batch. A method can be associated with testing. Materials are used in laboratory processes. Instruments support the work being carried out.

So the useful answer is often not just one piece of information. It is the context around it.

That is where connected AI can become more useful in a laboratory.

Instead of simply asking for a record, you could ask questions such as:

“Which samples are part of this batch?”

“Which method is associated with these samples?”

“What materials are connected to this work?”

“Which instruments are involved in this process?”

The aim is to make it easier to move between the information that already exists in your laboratory system and understand how it relates.

      3. What if you could ask Sophia to actually do something? 

This is where everything gets interesting. Sophia doesn’t only answer questions, it can also take actions in QLIMS. So what could that look like in a real laboratory?

“Create samples from this information”

“Create a new method for this test”

“Register this new material”

“Create a worklist for these samples” 

“Show me the information related to this OOS investigation”

“Read my attached handwritten notes, and first confirm with me, then create those samples”

Imagine being able to ask these questions, get the information you need, and take the next step without navigating through multiple screens or repeating manual tasks. The real value of AI connectivity is the time and effort it can save your team.

And of course, acting on laboratory data comes with responsibility. Sophia’s actions are built within the same security and governance model that protects your lab in QLIMS, with actions and changes recorded through the QLIMS audit trail.

So, how much time could your laboratory save by simply asking?

Book a demo and see what Sophia AI can do for your lab.

Until recently, the conversation about “AI in the lab” was mainly focused on natural language search. However, new trends in AI are taking everything to the next level, so the QLIMS team thought: why not laboratories? 

We’re excited to announce a major addition to Sophia AI, the agentic AI built into QLIMS: MCP (Model Context Protocol) is now available.

MCP is an open standard, created by Anthropic, for connecting AI to the tools and systems it works with. It acts like a universal “USB-C port” for connecting artificial intelligence models to external data sources, tools, and workflows. In practice, it’s what turns Sophia from an assistant that tells you things into an agent that does things, acting directly and securely inside your lab environment.

The integration of the Model Context Protocol (MCP) into QLIMS marks a major shift for laboratory informatics. By connecting QLIMS directly to AI agent architectures, OnQ Software enables true Lab-in-the-Loop workflows and sets the foundation for fully autonomous scientific labs.

From Search to Action

For most of the labs, “AI” still refers to asking a question and getting an answer, however, labs need more than getting answers. Imagine that you could ask your AI assistant to create the samples for you, and Sophia AI goes even beyond it with reading your attached handwritten notes. 

With MCP integration, Sophia now can perform real laboratory actions on your behalf, right from a conversation. 

What Sophia Can Now Do

Through MCP, Sophia can act across the entire lab workflow:

• Sample & Order Management: Create and manage orders (batches) and samples from scratch, or directly from documents/OCR (Optical Character Recognition) input.
• Analytical Management: Create, approve, and duplicate methods and sample types.
• Inventory Control: Register new materials, track physical containers, and record reagent consumption.
• Instrumentation: Monitor hardware health and decommission faulty instruments.
• Execution & Worklists: Build batch runs (worklists), assign samples to positions, and enter test results in bulk.
• Compliance & Audit: Manage OOS (Out-of-Specification) investigations and retrieve full audit trails for any record.

Security is one of the biggest topics we’re asked about in demos, and in evaluations. Sophia’s ability to act is built on the same security and governance model that already protects your lab in QLIMS. 

Curious what Sophia could do in your lab? Step into the laboratory of the future today and contact us for a personalised walkthrough. Every lab is different, and we’d love to show you how Sophia adapts to yours.