Digital Energy Journal- Baker Hughes and autonomous well construction

Join | Existing Member Login

You are Home » News » View Article

Baker Hughes and autonomous well construction

Tuesday, May 26, 2026

Baker Hughes' Kantori digital solution for well construction is designed to autonomously do well planning, monitoring and optimisation as well as drilling. Dr Matthias Gatzen explained how it works

In January, Baker Hughes launched Kantori, described as a "unified digital autonomous well construction solution", covering well construction planning, execution (drilling), monitoring and optimisation. It includes a mixture of AI, physics-based models and real time data analytics, to support operations and specific workflows.

It can work at a well or a field level. It integrates the Corva solution for real time analytics and predictive intelligence in drilling.

Kantori is designed to help customers achieve drilling costs that are the same as the AFE (Authorization for Expenditure), says Dr Matthias Gatzen, executive director, Digital Drilling Business, Baker Hughes.

Historically, digital tools in the industry for drilling have been 'quite disjointed,' he says. There were many different applications doing something different for different stakeholders. What is new specifically about Kantori is that all the relevant tools are connected together. 'Our customers are looking for a solution, they don't want an app.'

The goal of the system is to support rapid decision making, reduce the amount of human intervention necessary, reduce non-productive time of drilling rigs and variability. So, drilling can be safer and more predictable.

There have been times the software has helped achieve a greater than 40 per cent improvement in rate of penetration, he says.

The autonomous technology has helped steer drill bits within a 2-metre window in the subsurface, 'which is very challenging,' he says.

There have been improvements of 'weight to weight times' (total time taken during a drilling connection) of 30 per cent.

The software tool also helps operators do much more with their level of skills.

The digital tool also enables drilling to be done with more consistency. Where expertise is provided as part of the advice in the software, it is like 'you get the best engineer on the best day every single time of your operation,' he says.

There are still drillers, a person or a crew responsible for the operation. But they are supported by the autonomous system. 'They are there to drive that final maximum performance out of the system,' he says.

There is also a team of top engineers sitting in a remote 'solution room,' who determine what to do when an 'exception' happens.

Kantori is one of three major digital solutions from Baker Hughes, the others being Leucipa for field production and Cordant for asset performance management.

Putting these tools together, it becomes possible to incorporate well construction as part of a broader value chain, such as for delivery of gas volumes at promised levels.

A scalable solution

Kantori is provided as a 'scalable solution,' where you can get deeper and deeper in your use of it, until getting to full autonomous operations.

The first layer is where it provides monitoring. Once companies are comfortable using digital tools for monitoring, they can go to the second layer, with orchestrating the operations. Once that scales, you can go towards the third layer of full autonomous operation.

Baker Hughes has a systematic change management process, to help customers get to where they want to be, gradually getting used to each step.

To understand what it means to get used to a level of autonomy, Dr Gatzen cites semi-autonomous technology on cars, which can automatically change lanes when driving down a highway.

'The first time I drove one of those cars it wasn't driving the way I would drive,' he says. 'It was changing lanes more rapidly than I do. I had to get used to it.'

What autonomous drilling means

Autonomous drilling can be thought of as analogous to autonomous aviation. Even though modern planes can fly by themselves, the pilot in a plane is still responsible for making sure the plane takes off and lands.

'It is similar in our [drilling] space. Our 'pilots' are overseeing the system, making sure the system is getting the maximum performance,' he says.

Going from 'automated' to 'autonomous' means a shift from working with the technology in separate tasks and following rules, to where the technology works out the best way to achieve your overall goals while operating within preset boundaries, he says.

The engineer provides instructions to the system, such as the expected rate of penetration, and the system works out how to achieve it and executes without the engineer's interaction. 'That's an autonomous closed loop operation.'

A well construction project begins with making a plan. The digital tools offer 'autonomous planning.' This involves making multiple simulations of the drilling until it finds the best plan. The output is an engineering plan which can be sent to the rig site.

During drilling, the software can work with real time data from the drilling rig. 'It is constantly recalculating all the parameters within the given boundaries to find the optimum solution,' he says.

Data from the drilling rig needs to be quality checked, and then assessed, to determine how the operation is doing.

The software monitors the drilling activity, including rate of penetration of the drill bit, and ensuring the well will have the maximum contact with the reservoir.

The system can autonomously change parameters like mud flowrate or weight on bit, impacting the rate of penetration. For example, drillers are keen to avoid stuck drill pipe. They keep the pipe from getting stuck with optimized drilling windows flow around it and optimising the drilling speed.

With autonomous technology it is often possible to see much earlier that the system is changing towards the point where you need to make an alteration to the parameters.

It can identify possible problems, ensuring that drilling is as predictable as possible, so that overall costs meet expectations. 'We don't like surprises,' he says.

In autonomous drilling, the digital tools also seek to reduce the amount of human input required in 'transaction type activities,' Dr Gatzen says.
This means small adjustments, or repetitive tasks.

For example, it can automate the steering of the drill string. 'Historically the engineers would manually calculate the next steering command to send, up, down, right or left. You are also spending a lot of time making sure the command gets sent downhole. Now it all gets done autonomously.'

'In general, people don't like doing repetitive tasks,' he says. 'You become an engineer because you want to work with the customer and deliver a superior result.'

'You can run the operation in a state where you're not touching anything. The engineers can sit back and think even more about what's coming next.'

Physics and AI models

Inside Kantori, there are a number of different modules doing different tasks.

There are physics-based algorithms doing certain calculations, which were done manually in the past.

There are AI modules which 'process fast data streams, decipher data and make decisions,' he says. These modules utilize both machine learning and reinforcement learning methods, incorporating large amounts of incoming to support real-time estimates and decisions. .

There are generative AI tools for providing technical support, such as giving advice on maintenance, troubleshooting equipment failures, or answering technical queries, 'what do I need to do.' If the company has a large manual, it can help you find the right process in the manual.

The generative AI tools can be used for analysing documents and looking for commonalities. Sometimes multiple reports are really about the same thing.

AI tools can search through corporate data archives, to see if the well you are planning to drill is similar to wells drilled in the past, such as in terms of rock formation sequence.

It can analyse the drilling performance for the well drilled in the past and see if drilling performance could have gone better if today's knowledge had been available.

When considering AI, you should first consider what outcome you are looking for, he says. You should ask 'why you need it, what you are trying to accomplish.' Not use AI for the sake of saying you are using AI.

For example, you need AI when you have vast amounts of data and want to make decisions with that data. 'That's where it is valuable.'

AI to access data

The AI tools in Kantori have advanced tools to search for drilling data and interact with the results.

For example, you can search and select data from offset wells by depositional environment, stress orientation and formation types, not with preset menus.

You can select and refine your choice using natural language and interact with the results.

This leads to faster and better results which lead to better plans and service delivery.

Baker Hughes refers to this type of AI as 'amplified intelligence', because it puts the capability and control directly in the hands of the subject matter expert.

Agentic AI

There are agentic AI and agent frameworks that can run complex workflows including well trajectory and drilling programs.

There are follow-up agents that can watch ongoing operations and report back when actual results deviated from the plan.

These capabilities come together to provide a customized framework, uniquely tailored to each user, but still aligned and maintained as a central framework, Dr Gatzen says.

Associated Companies
» Baker Hughes (BHGE)

CREATE A FREE MEMBERSHIP

To attend our free events, receive our newsletter, and receive the free colour Digital Energy Journal.