Blog

What a Toolbox Talk is and what it is not A toolbox talk is a short crew briefing on a specific hazard, delivered before the work that exposes the crew to that hazard. It is not formal training and it is not a pre-task-plan. The PTP identifies and controls hazards at the task level. The talk briefs the crew on one hazard and confirms shared mental model. Treating the talk as a substitute for either is the most common program defect. Step 1 - Topic selection Weight topics in this order: 1.

Brady Keene

May 6, 2026·3 min read

At a glance * BLS projects roughly 19,200 safety openings each year. US colleges produce about 1,100 safety graduates annually. * One in four current safety professionals are expected to retire in the next decade. * There is no single source of talent that closes this gap. * Companies that combine multiple pipelines, intentional development, and modern tools will be better positioned to staff the next decade. The ones waiting on one clean fix will not. The math does not work The demand f

Brady Keene

April 24, 2026·4 min read

Recently, I was asked to write a piece on the trajectory of AI in construction safety for the Connecticut Construction Industries Association (CCIA) CONNstruction Magazine. Below is a short summary. If you are looking for more insight, check out the full article. A few takeaways: AI in construction safety is not falling short because of the technology. In many cases, it is because the focus has been on speed rather than how work actually happens. Early efforts improved reporting and data coll

Brady Keene

April 23, 2026·1 min read

At a glance * Energy is present in every task, whether anyone is looking at it or not * Most serious injuries happen when energy moves in a way no one controlled * Energy based safety is a way of seeing work, not a new program to buy * A short list of energy sources shows up on almost every job * Three simple questions change how a team thinks about a task * The shift is small on paper and large in the field Why we are talking about energy Most safety training starts with rules. Wear t

Brady Keene

April 17, 2026·4 min read

At a glance: Not all energy is equal. Some will bruise, some can kill. High energy is the kind that drives serious injuries and fatalities. Direct controls are what actually stop that energy from reaching a worker. PPE, procedures, and training help, but they are not direct controls. Most SIF events share the same story. High energy was present, and no direct control was in place. Knowing the difference changes how you plan, observe, and respond. Why this matters If you read the last post on

Brady Keene

April 17, 2026·4 min read

At a glance: AI agents in safety should not feel like software that people operate. They should feel like the best coworker you have ever had on a job. Like the one who carries memory, understands context, and helps decisions happen in the flow of work without adding friction. The Real Problem With Safety Tech Today A lot of safety technology still behaves like a rulebook taped to a hard hat. It can be correct and still be unhelpful when the plan changes and the crew is improvising. Real work

Brady Keene

February 4, 2026·3 min read

At a glance: AI in safety only scales when the entire system works, not just the AI. Data quality, governance, cybersecurity, and operational fit determine whether outputs remain reliable as conditions change. Without that foundation, even impressive tools struggle to survive real world use. Artificial intelligence (AI) is rapidly entering the Environmental Health and Safety profession, but scaling AI in safety environments introduces challenges that many vendors underestimate. Reliable AI sys

Brady Keene

February 3, 2026·3 min read

At a glance: Artificial intelligence is entering Environmental Health and Safety fast, but much of the market is filled with shallow tools rather than meaningful operational innovation. Many so-called AI first products prioritize demos over reliability, eroding trust and slowing adoption in real work. The goal is not to chase AI. The goal is to apply it with purpose so it strengthens decision making and risk management. Artificial intelligence is rapidly entering the Environmental Health and Sa

Brady Keene

January 24, 2026·4 min read

The fastest way to build an AI agent users love using? Stop thinking about AI. Pretend you're hiring unlimited humans to do the job instead. Don't think about LLMs. Don't think about tool limitations. Don't think about cost. Just ask: if I could hire as many people as I want to do this job, what would I want them to do? What would they need to do it well? And what would success look like? Say you're building an agent to help field workers identify and control hazards. If you hired a human for

Houman Farokhzad

December 25, 2025·2 min read