From general to specialised: choosing the right AI for EHS

AI is no longer a futuristic buzzword — it’s increasingly integrated into our professional lives. For EHS leaders, AI offers a compelling promise: to revolutionise risk management, streamline compliance and, ultimately, save lives. But as AI tools become more accessible, one critical distinction is often overlooked — not all AI is created equal. As EGOR NAZAROV, Senior Head of Marketing at Soter Analytics, explains, understanding the difference between general-purpose AI models and specialised, industry-specific ‘vertical AI agents’ is essential for EHS professionals aiming to make informed, strategic decisions.

The current AI landscape is dominated by powerful ‘large language models’ (LLMs) like ChatGPT and Gemini. These general-purpose models are incredibly versatile, capable of drafting emails, summarising documents and even generating creative content. An EHS manager might use an LLM to outline a toolbox talk or summarise a lengthy regulatory update.

While helpful, these applications only scratch the surface of what AI can achieve when tailored to the unique complexities of the EHS domain. The true transformative power lies in leveraging vertical AI agents — systems specifically designed, trained and optimised to address the unique challenges, workflows and data of EHS.

Defining the divide: general-purpose vs industry-specific AI

To appreciate the value of vertical AI, it’s important to distinguish between two approaches:

General AI models

These foundational LLMs are akin to highly educated generalists, trained on vast, diverse datasets across countless domains. Their strength lies in broad versatility, not deep specialisation. For EHS use cases, they often require users to perform extensive prompt engineering or custom fine-tuning to produce meaningful outputs.

Vertical AI agents

Purpose-built for industries like EHS, these models follow a ‘narrow and deep’ design philosophy. They are trained on curated, high-quality datasets — regulatory texts, industry standards, safety data sheets (SDSs) and historical incident records, allowing them to understand EHS-specific language, context and logic with much greater accuracy.

General AI vs vertical AI for EHS: a comparison

The differences between these AI approaches become even clearer when we examine their core attributes in the context of these four EHS challenges:

1. Precision in a high-stakes field (scope and purpose)

In EHS, precision isn’t optional. General models can provide basic overviews but require highly specific prompting to interpret scenarios like confined space entry. Vertical AI, on the other hand, is built with these specific workflows in mind, delivering targeted, context-aware support out of the box.

2. The power of relevant data (training data)

General models are trained on internet-scale datasets that may lack the rigour, accuracy and domain specificity required for EHS. Vertical AI is fine-tuned on authoritative, industry-specific data, enabling it to ‘speak EHS’ natively and accurately.

3. Expertise that understands nuance (knowledge and expertise)

A general LLM might correctly define the ‘hierarchy of controls’ but struggle to apply it in context, such as addressing an ergonomic hazard. In contrast, a vertical AI trained on ergonomic assessments can analyse task data and suggest appropriate controls, grounded in regulatory and best-practice knowledge.

4. Seamless integration vs heavy lifting (customisation and implementation)

General AI models typically require significant customisation and engineering to align with specific EHS workflows — like coding a system to classify near-miss reports using your internal risk matrix. Vertical AI solutions often come with built-in connectors, compatibility with popular EHS software and pre-configured understanding of EHS data structures, reducing implementation time and complexity.

Why this matters

Distinguishing between general and vertical AI is more than an academic exercise — it’s a strategic imperative. In high-risk, compliance-driven fields like EHS, the benefits of vertical AI extend far beyond automation. These systems offer predictive insights, enhance regulatory alignment, and empower safety teams to prevent incidents proactively rather than reactively.

While general AI tools offer broad utility and convenience, the future of EHS transformation lies in vertical AI — deeply knowledgeable, workflow-aware and purpose-built for the environments they serve. By adopting vertical AI, organisations can move from simply managing safety data to actively shaping safer, more compliant workplaces.

Image credit: iStock.com/gorodenkoff. Stock image used is for illustrative purposes only.