Stephen Graham, Vice President Product & Technology, Production Software Division, Hexagon, explains why the most meaningful applications of AI in CAM will be measured by more reliable programmes, shorter prove-out, and greater confidence at the machine
MTN: Has the industry misunderstood where AI actually creates value in machining?
I think the industry is seeing the most visible part of AI first.
A lot of the current conversation is focused on the interface: copilots, prompts, assistants, and natural-language interaction. Those things are useful, and they will make software easier to use. But they are only one part of the picture.
In machining, the deeper value comes when AI starts to support the decisions that determine whether a programme will run correctly on a real machine tool, with real posts, tooling, setups, and production constraints.
The hard part is rarely creating a toolpath in isolation. It is building a process that reflects the part, the material, the machine, the workholding, the tooling, and the realities of production.
That is where AI becomes more meaningful. It moves from helping someone use software faster to helping manufacturers make better decisions before the first cut.
MTN: Where does that challenge show up most clearly?
It shows up wherever programming complexity is high and the cost of getting it wrong is significant.
Mill-turn and Swiss-type machining are good examples. In those environments, the programmer is managing far more than geometry. They are thinking about synchronisation, tool access, machine limits, channel sequencing, collision risk, and how each operation affects the next.
Small decisions at that stage can have a large effect later. A tool choice, an approach strategy, or the way operations are sequenced can determine whether the programme runs cleanly or needs intervention at the machine.
That is where uncertainty enters the process. It often starts before production, but it shows up during production, when the options are more limited and the cost of delay is higher.
MTN: Is that where AI starts to make a difference?
Yes, but the important point is that AI is not replacing the programmer.
The value is in making proven decisions easier to repeat, and in helping engineers manage the trade-offs they already make every day. In machining, there is rarely a single perfect answer. Programmers are constantly balancing speed, tool life, surface finish, machine availability, part quality, set-up constraints, and risk.
Experienced programmers carry a huge amount of knowledge about those trade-offs: what strategies work, where problems tend to occur, how a specific machine behaves, and which approaches are safest for certain families of parts.
The challenge is that this knowledge is not always available at the right moment, to every person, on every job. It may sit with one expert, one site, or one shift. Under pressure, even experienced teams can make decisions differently from one job to the next.
AI becomes useful when it can help capture those proven approaches and apply them in context. That means supporting the programmer with knowledge that reflects real manufacturing practice, rather than leaving every new part to start from a blank screen.
MTN: What does that look like in practice?
Automated process planning is one of the clearest examples.
Instead of asking the programmer to manually define every step from the beginning, the system can analyse the part, recognise features, and suggest machining strategies based on approaches that have worked before. The key is that this should be grounded in real machining context, not generic automation.
With solutions such as ProPlanAI, the goal is to bring proven manufacturing knowledge into the programming workflow. That can reduce programming time significantly, but the bigger benefit is predictability. Manufacturers can create programmes that are more consistent from the start, with less dependence on individual interpretation.
In practical terms, it helps turn expert knowledge into a shared capability. That is especially important as manufacturers face skills pressure, shorter lead times, and more complex parts.
MTN: Where do tools like EDGECAM Copilot fit into this?
Copilots are part of the story, but they are most valuable when they are connected to real manufacturing context.
A copilot that simply helps someone navigate software faster can be useful. But in machining, the bigger opportunity is to help users make decisions with more confidence. That means connecting assistance to the machines, tooling, strategies, and workflows that manufacturers actually use.
With EDGECAM Copilot, the aim is to reduce friction in the programming process. It helps users find the right commands, understand options more quickly, and move through the workflow with greater confidence. For less experienced users, that creates a stronger starting point. For experienced programmers, it can remove some of the repetitive effort that slows them down.
The important point is that this is a practical route into AI. It does not ask manufacturers to change everything at once. It builds on the way programming teams already work and helps them improve incrementally.
MTN: Where does CAM fit into this shift more broadly?
CAM is where manufacturing intent has to become machine reality.
It is one thing to create a toolpath. It is another to know that the programme will behave correctly on a specific machine, with a specific setup, using the right post-processor, under real production conditions.
That is why machine-aware programming, simulation, and post-processing are so important. In complex machining, confidence comes from knowing that what has been programmed and validated in CAM will match what happens on the machine.
Platforms such as ESPRIT EDGE are designed around that principle. The goal is to close the gap between programming and execution, so manufacturers can reduce prove-out risk and move into production with more confidence.
Q: A lot of vendors are positioning AI as a “copilot”. Is that enough?
It depends what problem the copilot is solving.
A copilot can make software easier to use. This matters, especially as manufacturers bring new people into programming roles and ask experienced teams to handle more complex work. But in machining, ease of use is only one part of the challenge.
There is a wider debate around whether AI puts CAM programming at risk. I see it differently. CAM programming is built on experience, judgement, and engineering trade-offs that are difficult to separate from the realities of production. A programmer is rarely choosing between a right answer and a wrong answer. They are balancing tool life, cycle time, surface finish, machine availability, part quality, setup constraints, and risk, while making trade-offs against the cost, performance, and production requirements of the part itself.
That is where the real distinction lies. AI can be extremely helpful when it supports those decisions, removes repetitive effort, and helps programmers apply proven strategies more consistently. The engineer still needs to know when to accept the recommendation, when to challenge it, and when the real machine, material, or job context demands a different decision.
So copilot is useful, but it is limited on its own. The more meaningful question is whether AI is connected to the manufacturing context behind the decision: the machine, the tooling, the post-processor, the setup, the process history, and the constraints of the job.
That is where AI becomes genuinely valuable in CAM. It does not replace the craft of the programmer. It helps make that craft easier to apply, easier to share, and more reliable under pressure.
MTN: What changes for manufacturers if they get this right?
They remove uncertainty from the process.
Programmes become more repeatable across different programmers, teams, and shifts. Prove-out time can reduce because fewer assumptions need to be tested at the machine. Operators spend less time resolving avoidable issues and more time making parts.
That has a direct impact on productivity, but it also affects confidence. Manufacturers can respond faster, quote with more certainty, introduce new work more smoothly, and protect valuable machine capacity.
For many companies, that stability matters as much as speed. They are not only trying to programme faster. They are trying to make production more predictable.
MTN: So what is the real role of AI in machining going forward?
The role of AI is to make good manufacturing decisions easier to repeat.
That does not mean removing human judgement. It means supporting it with better context, better access to proven knowledge, and more consistent application of best practice.
In machining, AI will not be defined only by what users see on the screen. It will be defined by what changes in the process: fewer avoidable errors, less rework, shorter prove-out, more consistent programmes, and greater confidence that parts will run correctly.
That is why AI should not be seen as something that sits on top of manufacturing. The real value comes when it sits inside the workflow.
MTN: What should manufacturers be focusing on now?
They should look for the points in their process where variability still exists.
If two programmers approach the same part in very different ways, there may be an opportunity to standardise proven practice. If programmes regularly need adjustment at the machine, there may be a gap between programming and production reality. If expert knowledge is concentrated in a small number of people, there may be a need to make that knowledge easier to capture and reuse.
The goal is not to automate everything. The goal is to make the process more reliable, so quality and productivity do not depend too heavily on who is available, how much time they have, or how much knowledge they personally carry.
And then AI starts to deliver real manufacturing value: by helping teams make better decisions earlier, apply them more consistently, and run parts with greater confidence.
FAQ:
FAQ 1: What is AI in machining?
AI in machining refers to the use of artificial intelligence to support decisions in CNC programming, CAM workflows, process planning, simulation, and production preparation. In Hexagon’s view, the real value is not only faster software navigation, but helping manufacturers make better decisions before parts reach the machine.
FAQ 2: How can AI improve CAM programming?
AI can improve CAM programming by helping programmers repeat proven strategies, reduce manual process planning, recognise part features, suggest machining methods, and reduce variability between programmers, teams, and shifts.
FAQ 3: Does AI replace CNC programmers?
No. AI is not replacing the programmer. Its role is to support human judgement, reduce repetitive work, and help engineers apply proven manufacturing knowledge more consistently.
FAQ 4: Why is AI useful before the first cut?
AI is useful before the first cut because many machining problems begin during programming and process planning. Better decisions around tooling, sequencing, machine limits, simulation, and setup can reduce prove-out time, avoid errors, and improve confidence at the machine.
FAQ 5: What is automated process planning?
Automated process planning uses software to analyse a part, recognise features, and suggest machining strategies based on proven manufacturing approaches. In the article, Hexagon presents this as one of the clearest examples of how AI can support CAM teams.
FAQ 6: What is EDGECAM Copilot?
EDGECAM Copilot is described as a practical route into AI for CAM users. It helps users find commands, understand options, and move through the programming workflow with greater confidence.
FAQ 7: Why does machine-aware programming matter?
Machine-aware programming matters because a CAM programme needs to work on a real machine with real tooling, posts, setups, and production constraints. Simulation, post-processing, and machine-specific validation help close the gap between programming and production.
FAQ 8: What benefits can manufacturers expect from AI in machining?
Manufacturers can expect more consistent programmes, shorter prove-out, fewer avoidable errors, less rework, better use of machine capacity, and greater confidence when moving from programming into production.
For more information visit: Hexagon
For more background on how AI is already changing CAM workflows, read our earlier feature on AI-powered CNC programming with CloudNC CAM Assist 2.0.
The discussion around Hexagon’s EDGECAM Copilot also connects closely with our Q&A on AI-driven CAM and programmer confidence.
For a wider view of the market, see our comparison of AI CAM software in 2026, including hyperMILL, NX, Mastercam and Fusion.
The article also links well to our analysis of AI machining assistants and whether programmers are ready to trust them.
For practical advice on programming productivity, read our guide to faster CAM programming.
You can also explore more stories in our AI in CNC section.
