Automation and AI in high-hazard operations, built into your management system

Where does AI change your safety management system first?

• AI is moving from advice to control. A model that writes set points, or that operators now follow without checking, is a change to the process. AI management of change, AI in process control and the DCS and autonomous operations.
• Your board asks which AI standard you follow. IOGP 817-1, DNV-RP-0671 and ISO/IEC 42001 each cover part of the answer, and none covers your own safety management system. AI standards gap assessment.
• Operators are losing hands-on skill. When AI runs the plant, the team still has to run it without the AI. Operator competence when AI runs the plant.
• AI is in use outside the approved route. An inventory of AI use, run under your legal or compliance lead, shows where governance needs to reach. AI use inventory.
• Pilots that never reach operations. Why AI stalls.

Behind all five sits one movement. Operators are climbing the WIB/NAMUR autonomous operation maturity levels, from 0, all manual, towards an end point that for many sites will never be level 4 or 5. At each step, governance and the management system have to be adapted, and regulators are checking that they are. The autonomous operation maturity levels.

Havtil’s 2026 audits of Vår Energi and Aker BP found the gaps in technology qualification and risk assessment, not in the technology. What regulators found.

Dates and events also drive the work: EU AI Act and Machinery Regulation dates, a regulator letter or order on AI, and an AI-related incident or near miss.

Where does XO Autonomous Operations bring value?

Industry is increasingly harnessing AI and technology to improve agility and efficiency. Many high-hazard operators keep AI at pilot stage, or deploy it without the controls their safety case assumes. XO is independent of AI and automation vendors and sells no software. It does not install a new system beside yours. It builds AI and digital systems into your management system and existing processes, so that they reach operations and stay safe. Your own aims for AI and digital are the main yardstick, alongside the law and good practice, and the work happens inside production, under management of change, with the barriers that prevent major accidents in view.

Implementation

When you bring in automation, digital systems or AI, I work through the management system you already run and adapt it where they change the picture: management of change, risk assessment, competence, procedures, barrier management, monitoring and records. Findings land with owners who already have a route to act, a budget and a regulator who recognises the standard.

Assurance

Before an AI system enters a decision that affects a barrier, I check what it can and cannot see: its inputs, its training data, the operating domain its vendor declares, and how it behaves when inputs degrade. I also check the stop function and the fallback when the system is unavailable.

Audit

I audit AI management systems and AI risk assessments, and join audit teams as the AI specialist for a specific audit. How audit support works.

Compliance

Each action is traced to the legal requirements that apply to your operation, starting with the major-accident and process safety duties you already work under (the Seveso III Directive, COMAH, OSHA process safety management, Havtil’s regulations in Norway) and, where it applies, the EU AI Act. The legal scope is agreed for each engagement. Each action is also traced to recognised good practice: ISO/IEC 42001, IOGP Report 817-1, the NIST AI Risk Management Framework (NIST AI 100-1) and DNV-RP-0671.

How do I work? The M-OPS service range

M-OPS is the XO service range. It takes you from an initial conversation to AI running inside your operation.

1. Initial conversation

   You describe the AI or digital system and where it touches your operation. I tell you whether and how I can help.

2. Autonomous operations briefing

   One day, on site or online, with your operations, process safety or technical authority lead and the owner of the AI system. We take one automation or AI change you run or plan and walk it through the M-OPS OMI framework: the requirements and opportunities that apply, where they sit in your management system, the first gaps and blockers, and what to fix first. Within a week you receive a two-page note with the gaps, the priorities and a proposed scope for the assessment.

   Fee: EUR 3,500 excl. VAT, credited in full against the M-OPS OMI assessment. On site, travel and subsistence are charged at cost, agreed in advance.

3. M-OPS AssessAssessment, gaps and a prioritised plan

   Your management system is assessed against the legal requirements in scope, industry good practice and, most importantly, your own aims for AI and digital. Where those aims are not yet set, I help you develop them, starting from the largest gains your current business and systems allow.

   You receive a report with each gap and blocker, ranked, and a plan to close them, with owners and actions, built into your existing procedures and governance.

   Fee: quoted for each engagement, with the scope and fee agreed in writing before work starts.

4. M-OPS IntegrateAdvice and coaching during implementation

   I work with your teams while the plan is carried out, including changes to governance, and coach as far as your organisation needs, across cultures.

   Fee: quoted for each engagement, with the scope and fee agreed in writing before work starts.

5. M-OPS AssureA check before the AI system goes live

   I check what the AI system can and cannot see: its inputs, its training data, the operating domain its vendor declares, and how it behaves when inputs degrade, as well as the stop function and the fallback when it is unavailable.

   Fee: quoted for each engagement, with the scope and fee agreed in writing before work starts.

6. M-OPS ReviewAudit once it is in use

   I audit AI management systems and AI risk assessments, or join your audit team as the AI specialist for a specific audit. How audit support works.

   Fee: quoted for each engagement, with the scope and fee agreed in writing before work starts.

Which situations does XO help with?

Four typical situations. They are illustrative: the organisations are composites, not clients.

• Operator · AI in productionAI deployed, no gains on the bottom lineAI tools have run in maintenance and production for some time. The costs are visible; the gains are not. M-OPS checks the tools against your own aims and finds where the management system stops them changing decisions: no owner, data not used, procedures unchanged. You get a prioritised plan to capture the gains.
• Operator · AI not yet startedUnder pressure to adopt AI, not sure where to startThe board and competitors are moving, and nobody wants to miss out or make a costly mistake. M-OPS helps you set your aims for AI and digital from the largest gains your current business and systems allow, shows what your management system must change first, and helps you choose a first use.
• Operator · AI use inventoryAI in use outside the approved routeGeneral-purpose AI assistants can reach procedures, permits, risk assessments and reports without passing through the site's approval rules, so M-OPS starts with an inventory, run under your legal or compliance lead, of where AI is used, which decisions it reaches and which of those uses the document control, management of change and competence rules already cover.
• Gas processing · Worst caseAn AI-supported decision and a major accidentA predictive maintenance model supports a deferral that every reviewer accepts, and a cold box fails outside the model's view. The Corlinda case shows how, and what an integration review would have asked.

What is M-OPS OMI?

M-OPS OMI (Operations Management Integration) is XO’s proprietary framework, behind every M-OPS service. It maps your AI and digital systems onto the management system you already run, to identify quickly the gaps and blockers that prevent delivery. It then builds a structured plan to close them and makes that plan operational. How M-OPS OMI works, and illustrative cases of the gaps it is designed to find.

What does XO bring?

• 30 years in oil, gas and chemical operations
• 1,000+ people in one operating organisation I managed
• 50+ management system audits and improvement plans delivered in over 20 countries worldwide
• 48 apprentice technicians in a national operations team I built

I have led major hazard operations and stood up the organisations that run new facilities, and I have designed, audited and improved the management systems behind them. The person who scopes it is the person who does it, and has run a site. I conduct and lead every mission myself. About the founder.