By using MathWorks’ low-code AI tools, engineers with minimal coding experience are saving time while reaping the benefits of AI.
Without a strong background in software development, engineers wanting to integrate AI into their workflow may find it challenging to do so.
AI is complex to learn. It often requires extensive programming and an in-depth understanding of coding.
“A lot of engineers come into engineering without much knowledge or depth of software development experience,” says Ruth-Anne Marchant, Senior Team Lead at mathematical computing software company, MathWorks.
Fortunately, there’s a solution that comes in the form of a software development technique known as low-code AI.
“This is a graphical approach to developing software,” says Marchant. “It uses a point and click, drag and drop approach instead of a manual coding approach which would need to be carried out by a traditional software developer.
“Low-code AI lowers the bar and makes it easier for engineers to adopt AI and experience its many advantages.”
Achieve time savings
Danone’s Auckland-based site recently began using MathWork’s low-code AI tools to develop an automated system for its manufacturing plant.
The AI algorithm enables the system to identify bags, based on its artwork, and compare it with an applied label to ensure the right bags are being loaded and used in their production process.
“It’s checking that what is coming through is what it expected to receive,” says Marchant. “If it’s not the right item, the production line stops and this ensures the right recipe is followed and avoids wastage.”
Before Danone began using low-code AI, the bags were manually checked.
“That’s now all automated so it’s saving the company a huge amount of time and resources that can instead be redeployed elsewhere.”
Compared to traditional AI, low-code AI creates major time savings in other ways, too.
“When you don’t need to manually program AI, you save time because you’re not spending a significant amount of time on software development,” says Marchant.
“You can also rapidly iterate on different designs, because you don’t have to code each of these different designs. You can instead use the point and click, drag and drop tools to quickly put together a model, run some simulations, test them, compare the different results against each other. So it allows for rapid design iteration.”
Four-stage workflow
For engineers hoping to integrate low-code AI into their workflow, Marchant advises beginning with a clear objective in mind and building the process out from there.
“I’d first suggest thinking about what question you want to answer,” says Marchant. “Then identify a small component of your project that aligns to that question and start building up your AI algorithms using the low-code tools. As your capability builds up, you can expand the application across your projects, systems and company.”
To begin this process, she advises following a four-step AI workflow:
While this is a useful process to follow, it can be iterative and doesn’t always flow in a stepwise manner.
“You might need to go back to a stage if, for example, you realise you need to acquire more data or need to understand how your system behaves in a particular way,” says Marchant. “You can constantly iterate on the model’s development.”
Marchant also suggests that leaning into your engineering domain knowledge can provide the right direction for developing a low-code AI model.
“It’s okay if you’re not an AI expert or a computer programmer… The low code workflow can really help to get you started and guide the way.”
Learn more about MathWorks’ low-code AI tools to integrate low-code AI into your workflow.
