Every year, I leave ACE inspired by the Aras community's enthusiasm for problem-solving and openness to sharing solutions. This year's event was one that will stand out in my memory for a long time though. We had a mind-blowing keynote from CERN (yes, that CERN), an amazing turnout for the Tech Summit with six community demos, and the first-ever Aras Innovation Challenge on the main stage!
Inspired by hackathons and the traditional "rapid-fire demos" at ACE, the Innovation Challenge provided community members an opportunity to show off the creative and impactful solutions they're building on the Aras Innovator® platform. Participants submitted projects in one of three topic areas: real-time shop floor data, digital thread, or sustainability.
And of course, no competition would be complete without some stakes. Participants competed for a speaking slot on the main stage in Boston, complimentary ACE registrations, a joint workshop with Aras, and glory (that is, to get their name etched on the Innovation Challenge hammer, the "Stanley Cup" of the Aras Community).
Four finalists presented their solutions to a bustling crowd of more than 250 attendees at the end of the first conference day.
After each finalist made the case for their solution on the main stage, the audience voted for their favorite solution. The next day our CEO, Roque Martin, announced the community's choice at the "Aras in the Round" session.
Congrats to Tom Smith, the first winner of the Aras Innovation Challenge!
Missed the session or want to learn more about Tom's solution? I was curious about how he chose the topic and how he implemented it, so I followed up with him after the event to chat about it.
Tom's a senior integration engineer at SMC, one of the largest pneumatics manufacturers in the world, so it's no surprise that he was drawn to the real-time shop floor data theme. The opening moments of his presentation made it clear that some shop floor challenges aren't exclusive to manufacturing spaces though:
Who can't relate to the experience of having an idea, but no idea what to do with it?
Tom's solution addresses three common challenges teams face when improving shop floor operations.
Shop floor feedback is often recorded manually, leading to delayed responses or lost ideas. Tom addresses this pain point with a super simple form, making it easy for users to submit feedback.
When feedback is reported, it's typically unstructured, which makes it hard to interpret and act on. Tom's solution makes it easier for teams to search and filter by using AI to categorize input.
AI also helps Tom summarize and recommend actions based on the feedback. These insights are delivered via weekly email, ensuring feedback is shared in an easily digestible format. This strategy enables faster issue resolution, supports data-driven decision-making, improves team communication, and ultimately aligns stakeholders around continuous improvement.
Tom's solution begins in Aras Innovator, where he created an ItemType to store all the shop floor feedback. To collect the input, he built a lightweight web form with three simple fields: Employee ID, Job ID, and Description. The submit button sends the form data to Aras Innovator via the OData REST API.
Next, an onAfterAdd server event sends the description to a large language model (LLM) for categorization. Tom uses Ollama 3.2 LLM, which is open-source and runs locally to ensure that no sensitive data leaves the company's systems. The code prompts the model to categorize feedback into three key areas: Cost, Delivery, and Quality. The server method updates the feedback item's category property with the result.
To close the loop, Tom uses the Aras scheduling service to trigger a weekly server method that uses the LLM to summarize the latest feedback and suggest actionable tasks. The method then emails the insights directly to managers.
Tom's solution demonstrates that leveraging AI in PLM doesn't need to be a "big bang" effort to have a tangible impact on the business. I think the key is identifying the right opportunity and using AI as a tool rather than the whole solution. By letting the LLM handle the tasks it does best (like categorization and summarization) and keeping the rest of the implementation simple, you can quickly deliver a valuable solution that's ready to iterate and improve.
What do you think? Could you incorporate LLMs, generative AI, or machine learning in your future Aras projects? Are you already doing something similar today? Let us know in the comments below!
Thanks again to Tom and all of the Innovation Challenge finalists. Thank you for reading along and supporting your fellow builders!
Inspired by hackathons and the traditional "rapid-fire demos" at ACE, the Innovation Challenge provided community members an opportunity to show off the creative and impactful solutions they're building on the Aras Innovator® platform. Participants submitted projects in one of three topic areas: real-time shop floor data, digital thread, or sustainability.
And of course, no competition would be complete without some stakes. Participants competed for a speaking slot on the main stage in Boston, complimentary ACE registrations, a joint workshop with Aras, and glory (that is, to get their name etched on the Innovation Challenge hammer, the "Stanley Cup" of the Aras Community).
Meet the finalists
Four finalists presented their solutions to a bustling crowd of more than 250 attendees at the end of the first conference day.
Yogesh & Vilas
AI Advisor for Engineering Design
Yogesh Kulkarni and Vilas Sarangdhar from TCS presented an AI-powered solution that facilitates the import of unstructured data into Aras Innovator. The tool processes engineering drawings, extracts data characteristics, and creates corresponding item structures.
Tom Smith
Enhancing Shop Floor Feedback with Aras & LLMs
Tom from SMC developed a solution to collect and analyze shop floor feedback. The solution uses AI to categorize and summarize input to help teams consume and respond to feedback.
Konrad Golińczak
Document Comparison Tool
At Demant, Konrad built a solution that compares versions of Word documents in Aras Innovator. The tool uses AI to make intelligent comparisons and summarize the differences, helping users quickly understand changes.
Gilbert Delabrousse
Aras Lite MES: Streamlined Shop Floor Management
Gilbert at Inensia shared a lightweight manufacturing execution system (MES) built on Aras Innovator. The solution helps operators follow and log work instructions while managers monitor assembly line status in real time.
And the winner is...
After each finalist made the case for their solution on the main stage, the audience voted for their favorite solution. The next day our CEO, Roque Martin, announced the community's choice at the "Aras in the Round" session.
Congrats to Tom Smith, the first winner of the Aras Innovation Challenge!
Missed the session or want to learn more about Tom's solution? I was curious about how he chose the topic and how he implemented it, so I followed up with him after the event to chat about it.
Why shop floor data?
Tom's a senior integration engineer at SMC, one of the largest pneumatics manufacturers in the world, so it's no surprise that he was drawn to the real-time shop floor data theme. The opening moments of his presentation made it clear that some shop floor challenges aren't exclusive to manufacturing spaces though:
"Every manufacturing environment thrives on continuous improvement, but collecting and making sense of real-world shop floor feedback can be a challenge.
Imagine you're on the shop floor in a manufacturing environment, and you have an idea. Where does that feedback go? Do you write it on a sticky note and put it on your manager's monitor? Or maybe you write it up in an email that gets buried in with a thousand other emails. Or worse, maybe it goes nowhere at all."
Who can't relate to the experience of having an idea, but no idea what to do with it?
Where to start?
Tom's solution addresses three common challenges teams face when improving shop floor operations.
1. Feedback collection
Shop floor feedback is often recorded manually, leading to delayed responses or lost ideas. Tom addresses this pain point with a super simple form, making it easy for users to submit feedback.
2. Categorization
When feedback is reported, it's typically unstructured, which makes it hard to interpret and act on. Tom's solution makes it easier for teams to search and filter by using AI to categorize input.
3. Summarization
AI also helps Tom summarize and recommend actions based on the feedback. These insights are delivered via weekly email, ensuring feedback is shared in an easily digestible format. This strategy enables faster issue resolution, supports data-driven decision-making, improves team communication, and ultimately aligns stakeholders around continuous improvement.
How does it work?
Tom's solution begins in Aras Innovator, where he created an ItemType to store all the shop floor feedback. To collect the input, he built a lightweight web form with three simple fields: Employee ID, Job ID, and Description. The submit button sends the form data to Aras Innovator via the OData REST API.
Next, an onAfterAdd server event sends the description to a large language model (LLM) for categorization. Tom uses Ollama 3.2 LLM, which is open-source and runs locally to ensure that no sensitive data leaves the company's systems. The code prompts the model to categorize feedback into three key areas: Cost, Delivery, and Quality. The server method updates the feedback item's category property with the result.
To close the loop, Tom uses the Aras scheduling service to trigger a weekly server method that uses the LLM to summarize the latest feedback and suggest actionable tasks. The method then emails the insights directly to managers.
Takeaways
Tom's solution demonstrates that leveraging AI in PLM doesn't need to be a "big bang" effort to have a tangible impact on the business. I think the key is identifying the right opportunity and using AI as a tool rather than the whole solution. By letting the LLM handle the tasks it does best (like categorization and summarization) and keeping the rest of the implementation simple, you can quickly deliver a valuable solution that's ready to iterate and improve.
What do you think? Could you incorporate LLMs, generative AI, or machine learning in your future Aras projects? Are you already doing something similar today? Let us know in the comments below!
Thanks again to Tom and all of the Innovation Challenge finalists. Thank you for reading along and supporting your fellow builders!
Published 5 months ago
Version 1.0