In Simple Terms
Adding AI to a lab won’t fix problems if the information it needs is scattered across disconnected systems. When lab tools share information and workflows connect smoothly, people spend less time searching and re-entering details. That gives AI a clearer picture of what’s happening—and makes its help more useful and easier to trust.
Why Disconnected Systems Can Hold Labs Back
There is plenty of talk about bringing AI into laboratories, but installing a new tool does not automatically improve the work. Before an intelligent system can offer useful help, it needs to understand what is happening around it: which step has been completed, what information relates to that step, and what should happen next. If that picture is scattered across separate programs and procedures, AI may be technically present but disconnected from the context that would make its output useful.
That is the focus of the third episode in a series on the subject, on the GEN podcast. Denise Bell, vice president of product management and marketing at LabVantage, discusses ways to reduce operational fragmentation and build connected, scalable laboratory processes. The point is not simply to add more digital tools. It begins with how systems connect and how the steps involved in everyday scientific work flow from one to the next.
When the Picture of a Workflow Is Spread Across Systems
In a laboratory, information is rarely just an isolated number. It may be tied to a measurement result, a sample, or a procedure that came before it or will follow it. To make sense of the information, someone needs to know where it came from and where it fits in the workflow. When different systems store separate pieces of that picture, it becomes harder to bring them together and understand how they relate.
The discussion describes this problem as information silos: data exists, but is effectively trapped in systems that do not share enough context. That does not necessarily mean the data has disappeared. It may already be stored, but finding it and connecting it to work happening at another stage does not happen automatically. As a result, people moving between steps may have to search again or transfer details by hand instead of finding them within the workflow itself.
This separation affects both what data means and how easily it can be used. A result shown in one place may not, by itself, reveal what came before it or how it relates to a procedure underway somewhere else. The more people have to gather pieces of the picture from separate sources, the more they must rely on coordination between systems and colleagues to understand the full situation. Connecting systems is therefore more than a technical cleanup; it helps preserve scientific and operational context as work moves along.
Manual Handoffs: A Small Step That Can Disrupt the Workflow
One sign of fragmentation is the manual transfer of a task or information from one stage to another. That handoff may seem simple, but it creates another point that requires human attention: the right information must reach the right place and be understood in the context of the next step. As the discussion notes, the more a process depends on these handoffs, the greater the chance of delays or problems transferring details.
This does not mean every human intervention is undesirable, or that every activity can or should be automated. The point is to identify repetitive tasks that separate systems without adding scientific value in themselves. If people have to re-enter information that already exists or manually look up the status of an earlier step each time, connecting workflows can ease that burden and make the movement of work clearer.
With less reliance on manual handling, a workflow can become more consistent and easier to follow. But that improvement does not come simply from adding an AI feature. First, a lab needs to identify where handoffs occur, what information must move, and which systems need to share it. Without that understanding, a new tool may be placed at the heart of a process that is still fragmented. The interface may have changed, but the underlying problem remains.
Four Kinds of Systems in One Execution Environment
The discussion considers bringing four kinds of systems together in a unified environment for laboratory work. The first is a laboratory information management system, or LIMS, which supports the organization of laboratory information and processes. Next is the electronic laboratory notebook, or ELN, a digital space for documenting scientific work. A laboratory execution system, or LES, focuses on carrying out procedural steps, while a scientific data management system, or SDMS, handles scientific data.
Each type of system has a different role. Bringing them into one environment does not mean their functions become identical or that one system makes all the others unnecessary. The idea is for the components to work together, rather than forcing users to treat them as isolated islands. When work documentation connects with the information, execution steps, and data associated with it, it becomes easier to see how the parts of a process fit together.
The value of this approach is that work context does not have to stop at the boundary of a particular program. Information needed by a workflow can be linked to the relevant step, while knowing the status of a process can help guide what comes next. This describes the direction discussed in the source; it is not a promise that every connection will automatically eliminate errors. The quality of the organization, design, and implementation still matters in any digital environment.
That is why it is important to coordinate workflows across different laboratory functions. Rather than viewing each system as a standalone tool, the questions become: How do tasks follow one another? What does each part need to know? Where should information move so its meaning is preserved? Answering these questions helps labs picture a connected execution environment and identify where they can reduce manual transfers or repeated searches.
Why AI Needs Context
An AI tool can work with the information available to it, but the value of its help also depends on how clearly that information is linked to what it describes. If data reaches the tool without a clear relationship to the workflow, its ability to offer relevant support may be limited. When the data is linked to the steps and functions that produced it, the tool is better placed to understand the work it is being asked to support.
That is why the discussion links operational infrastructure to the potential benefits of AI. Scientific context explains how information relates to the work that produced it, while operational context places it within the sequence of tasks. Together, the two give AI tools a better foundation for helping, rather than leaving them to work with scattered information whose connections are unclear.
Trust matters especially here. A laboratory needs more than an answer that sounds intelligent; it needs assistance whose basis can be understood in the context of the process where it is used. The source provides no numerical measures of accuracy or trust and describes no specific trial results. Instead, it puts forward a general principle: connecting systems and workflows supplies the context that can make AI assistance more usable and easier to scale.
Scalability, meanwhile, does not simply mean using a tool for one task or at one site. When the underlying work is connected, it becomes easier to consider extending its use to other functions or workflows, because information is less dependent on temporary workarounds between systems. Still, connection alone cannot guarantee that any application will succeed. It creates a more suitable environment, while choosing the right use and deciding how to implement it remain part of the work.
Where Should a Lab Begin?
The discussion considers practical starting points for adopting AI, but the available description does not provide a specific list of steps. The problem itself suggests a general starting point, without implying that the guest offered detailed recommendations: start by understanding existing workflows, rather than assuming that buying a new tool will automatically fix whatever is holding operations back.
A lab can map how a task moves from beginning to end and ask where information is recorded, where it is re-entered, and where context gets lost as work moves from one system to another. This kind of review helps distinguish between different problems: Is information missing? Is it difficult to access? Or is the connection missing between existing data and a step that depends on it? Pinpointing where work gets stuck makes conversations about connecting systems more practical.
Next, it helps to look at systems by the functions they perform, not just by their names. What information does the laboratory information management system hold? How does the electronic notebook connect to execution steps? Where does scientific data live, and how can it remain understandable within the process? There is no single answer that fits every lab, but asking these questions clarifies what systems need to share to close the gaps.
It is also useful to distinguish between two related but different goals: improving a workflow and adding AI. Reviewing the process may reveal that some operations first need to be better organized or connected. Once the relationships between data and steps are clear, it becomes easier to assess where intelligent tools could genuinely help, rather than using them simply because the technology is available. This approach makes AI part of an understandable process, not a separate layer sitting on top of it.
Connection Is a Foundation, Not a Magic Shortcut
The phrase unified environment might make the solution sound simple, but unity does not mean merely putting everything in one place. What matters is that information remains traceable and understandable as it moves between tasks and systems. If tools are brought together without addressing how they exchange data and context, fragmentation can persist under a new arrangement.
The episode does not promise that AI will eliminate every error or shorten every procedure. Instead, it focuses on what must come before AI can be useful: reducing fragmentation, coordinating work, easing manual handoffs, and connecting the laboratory’s core tools. These are preparatory conditions that bring digital assistance closer to the reality of laboratory work, but they do not replace the need to understand a lab’s needs and assess what will suit it.
Preserving context benefits people as well as tools. Laboratory staff also need a clearer view of how information relates to each step and a workflow that reduces the need to move between separate sources. When a process is better connected, it can become more straightforward to track work and understand where each piece of information belongs. In this sense, AI is part of a broader shift in how operations are organized—not a substitute for organizing them well.
The Bottom Line
The central message of Denise Bell’s discussion in the third episode of the GEN podcast is that laboratories need to put the right foundations in place before asking AI to help. Disconnected systems can leave information scattered and stripped of some of its context, while manual handoffs add time and create potential points of failure. The discussion suggests that coordinating workflows and connecting LIMS, ELN, LES, and SDMS in an integrated laboratory execution environment can provide clearer scientific and operational context.
The source offers no figures or quantitative results and does not lay out detailed implementation steps. But it makes a clear point: thinking about AI should not start and end with the tool itself. Understanding workflows, reducing information silos, and making it easier for data to move are all groundwork for making AI assistance more useful and scalable. When that foundation is built carefully, the key question is no longer simply whether AI is in the lab, but whether it has been placed where it can understand and support the work.