AI often sounds like a big-picture strategy conversation. In practice, its most useful starting point may be a task that is slow, repetitive, and overdue.
During a roundtable discussion led by Adam Steinert and James Smagala of Yahara Software at the 2026 Lab Manager Leadership Summit, one attendee shared a simple but effective example: chemical inventory. Instead of typing information from each bottle into a spreadsheet, the team took photos of chemical shelves using a phone and used ChatGPT to pull key details from the images, including catalog numbers, expiration dates, and CAS numbers.
The process was not perfect. A few details needed correction. Some outputs required review. But the time savings were significant. A task that used to take up most of a workday was cut down to roughly an hour.
That example illustrates one of the most useful ways lab leaders can begin thinking about AI. The first win does not need to be complex, expensive, or tied to a major system overhaul. It needs to solve a real problem, reduce friction, and give the team a safe way to learn how these tools behave.
Start where the work is slow and repetitive
Laboratories are full of processes that are essential but time-consuming: inventory updates, SOP review, data summaries, document searches, staff feedback analysis, report drafting, and project tracking. These tasks often do not require deep scientific interpretation, but they still consume valuable staff time.
That makes them good candidates for early AI use. They are close enough to daily work to matter, but low enough in risk that teams can experiment without handing over critical decisions.
In the roundtable, participants described several practical uses already taking shape in labs. One attendee discussed using AI to help prepare for an ISO 17025 assessment by creating an “ISO lab advisor” GPT loaded with relevant standards. The tool helped generate templates and support documentation that the team could review and refine.
Another lab leader described using AI to analyze anonymized stay interview data from a staff of 85 people. The leader and deputy asked different questions of the same data and received different types of summaries. When they compared the outputs, the combined results gave them a broader view of staff satisfaction.
The value here was not just speed. AI helped the leaders look at the same information from multiple angles, provided they treated the output as a starting point rather than a final conclusion.
AI prompting is becoming a management skill
Several participants noted that the quality of an AI output depends heavily on the quality of the prompt. A vague question will likely produce a vague or misleading answer. A more specific prompt can produce a far more useful response.
This makes prompting a practical skill worth developing. Staff need to learn how to describe the task, provide context, set expectations, and explain what the tool should not do.
For example, a team using AI for inventory could prompt the tool not only to extract information, but also to flag any label it cannot read. That small change makes the tool more useful because it asks the AI to acknowledge what it cannot interpret instead of generating an answer from incomplete information.
Lab Quality Management Certificate
The Lab Quality Management certificate is more than training—it’s a professional advantage.
Gain critical skills and IACET-approved CEUs that make a measurable difference.
The same principle applies to less visual tasks. A lab manager using AI to summarize staff feedback might ask it to identify recurring themes, separate operational concerns from leadership concerns, and flag any comments that appear outside the main patterns. A manager working on project tracking might ask AI to suggest categories, dependencies, or missing information before building a project management system.
Another useful approach discussed during the roundtable: ask the AI tool to ask questions first. Rather than trying to write the perfect prompt from scratch, a user can provide a brief description of the task and ask the tool what additional information it needs. This helps reduce the blank-page problem and can reveal assumptions the user has not yet considered.
Keep people in the loop
The strongest early AI applications still depend on human judgment. In the inventory example, the team reviewed the extracted information and corrected errors. In the staff survey example, leaders compared outputs and interpreted what the results meant for their team. In the ISO preparation example, AI helped generate material, but the lab still owned the final documentation and assessment readiness.
Interested in lab leadership?
The human role is still essential. AI can make a process faster, but speed alone does not make a process better. Lab leaders need to decide where review is required, who owns that review, and what level of accuracy is acceptable for the task.
A low-risk use case, such as organizing meeting notes or summarizing internal feedback, may only need a basic review. A higher-risk use case, such as analyzing regulated test data or supporting quality decisions, requires a much more structured approach.
This distinction can help managers avoid two common mistakes: rejecting AI because it is not perfect, or adopting it without enough oversight because it appears confident.
Use early wins to reveal bigger opportunities
Small AI projects can also help labs spot larger operational problems. If a team saves hours on inventory by using photos and extraction tools, that may raise bigger questions about how inventory is tracked, where the data lives, and whether current systems support the lab’s needs.
The same is true for historical data. Several roundtable participants raised questions about valuable information locked inside PDFs, Word documents, old reports, or disconnected systems. AI may help extract or analyze some of that information, but the larger issue is whether the lab has a reliable data foundation.
Early AI use can expose those gaps without forcing the lab to solve everything at once. A team may start by using AI to search SOPs or summarize old reports, then realize it needs clearer metadata, better document structure, or a more consistent approach to data capture.
Choose the right first project for AI
For lab managers looking for a starting point, the best first AI project should meet a few basic criteria. It should address a real pain point, involve information the team can safely use, have a clear human review step, and produce an outcome that staff can evaluate.
Good candidates might include summarizing non-sensitive meeting notes, drafting project plans, organizing SOP questions, extracting basic information from images, analyzing anonymized staff feedback, or generating first-pass templates for review.
The goal is not to prove that AI can do everything. It is to help the team build familiarity, reduce low-value manual work, and learn where the tool fits into existing lab operations.
Small wins also make AI less abstract. Staff can see how it helps with work they recognize. Managers can observe where it performs well and where it struggles. Leaders can begin setting expectations based on direct experience rather than hype.
For labs under pressure to do more with limited time and resources, that practical approach may be the most useful place to begin. AI adoption does not have to start with a major leap. It can start with one shelf, one spreadsheet, one SOP, or one workflow that no longer needs to take as long as it used to.
