The anaerobic digestion sector is finally starting to embrace artificial intelligence. Across the UK and Europe, operators are investing in digital platforms that promise better monitoring, earlier warnings, and remote plant visibility.

That’s a positive step.

But there’s a misconception emerging in the market that I think we need to address:
AI monitoring is not the same thing as AI optimisation.

And the difference matters.

Most modern plants capture a wide range of process information:

• Gas yield and composition
• pH and temperature
• Feedstock loading data
• Pressure and flow rates
• VFA and ammonia measurements

Systems such as SCADA already provide good visibility of what is happening inside a plant.
The challenge isn’t collecting data.
The challenge is turning that data into better operational decisions.

Many of the AI tools currently entering the market focus on monitoring and anomaly detection.
They analyse sensor data to identify patterns that might indicate process instability.

For example, they may flag early warning signs of:

• Ammonia inhibition
• VFA accumulation
• Acidification risk
• Feedstock-related instability

That capability is genuinely useful. Earlier warnings mean operators can intervene before problems escalate.
But monitoring platforms still leave a key question unanswered.

Monitoring systems can tell you something might go wrong.

What operators actually need to know is:

• What should we change right now?
• How should loading rates be adjusted?
• Which feedstock blend will maximise methane yield?
• What will today’s operational decision do to the process in 48 hours?

Those are optimisation questions, not monitoring questions.
And answering them requires a different type of AI.

This is something that often gets overlooked when digital tools are discussed. AD is not simply a mechanical process. It’s a living microbial ecosystem.
Small changes in feedstock chemistry or loading strategy can have delayed and sometimes unpredictable biological effects.

To optimise performance, AI needs to understand:

• Microbial behaviourfeedstock chemistry
• Nutrient balance
• Inhibition thresholds
• Process dynamics over time

That’s where the next generation of platforms is starting to focus.

One of the biggest untapped opportunities in AD is feedstock strategy.
Different substrates introduce different:

• Nitrogen loads
• Degradability rates
• Micronutrient requirements
• Methane potentials

Most plants still manage feedstock blending based on experience and historical performance.
AI optimisation platforms can help operators design feedstock strategies that maximise methane yield while maintaining process stability.
For many plants, the economic impact of getting that right is significant.

I believe the industry is moving through four stages of digital maturity:

1. Manual operation
2. Digital monitoring
3. Predictive analytics
4. AI-driven optimisation

A lot of plants today are moving from stage two to stage three. The real transformation will come when optimisation platforms become widely adopted. Because ultimately the goal isn’t just to monitor digesters more effectively.

The goal is to run them better.