A practical guide for AI-driven hardware engineering

There’s no shortage of mechanical engineering AI products promising the world. However, what they all seem to lack is an integrated plan for how to prepare for AI, apply AI and build a realistic business case for short-term and long-term results. So, we did the hard work those other companies won’t do.

Prepare for AI: Choose the right use case Prepare for AI

The reason so many AI companies promise a lot, but deliver on very little is because they don’t have a well-defined use case for their AI product. To start with AI, you need to know where to start. In hardware engineering, no use case could be simpler than one of the building blocks of product development: design reviews. 

Your most important product decisions are made every day in design reviews. Yet teams still struggle with mundane, admin tasks when it comes to tracking design review feedback. So, then teams struggle to apply the knowledge from past reviews to future reviews. Voila! A perfect use case for AI: tracking design review data and applying it to future reviews.

Apply AI: Build the right infrastructure to feed the AI model Apply AI

This is the key. Before applying AI at scale, you must build the infrastructure to gather the right data. Here’s what that should look like for design reviews:

  • Enable the right experts to provide quality design feedback. This is best done using a Design Engagement System, which enables engineers and suppliers to view and interrogate 2D drawings or 3D CAD models in their browser without downloads or special software. This allows the right experts to give feedback asynchronously and in a single platform.
  • Track qualitative feedback as a byproduct of reviewing the design. Today, when an expert leaves feedback, it’s usually done in a meeting or a PowerPoint deck. Then, someone has to manually track that in a spreadsheet. While, simultaneously, all these discussions about the same issue are happening over email, Teams and phone calls. Instead, the right infrastructure should capture and track all that feedback automatically. A Design Engagement System does this by allowing reviewers to pin feedback right on the CAD geometry. So, not only is the qualitative feedback saved, but so is the CAD view state where that feedback was left. This context is key for training the AI models.
  • Gather other design review feedback sources. Now, just because your other design review data lives in decks, spreadsheets and emails doesn’t mean it’s not valuable. As long as you can tie the comments to the respective CAD, the data is still useful context for AI.
  • Use AI to simplify design review workflows. Now that you have the right data, you can start using AI just like the generative AI tools you’re used to. In CoLab, simply pull up a model, click the ReviewAI icon and ReviewAI will:
    1. Generate lessons learned from previous similar projects.
    2. Complete basic checks, like part number verification, wall thickness and draft angles.
    3. Generate AI comments based on knowledge from past reviews. 
      ‍


Realistically, you can start using ReviewAI within a few weeks with the data you have. Then, use CoLab at the same time as your future infrastructure for enabling high-quality input data.

Now that you have a system for gathering data, it’s time to build the short-term and long-term business case for practical AI applications. 
‍

‍

5 reasons you think you’re not ready for AI (and why you should start anyway) Objections to AI

  1. AI is not ready to do my job. You’re right! AI can’t do everything you can do, and it probably shouldn’t. What it can do is the tasks that steal your time from more important tasks. AI can documents and surface data very quickly. It can perform routine checks based on input data and it can generate high-quality comments on basic DFM and drawing checks.
  2. AI will eventually replace me. Based on most AI predictions, this is simply not true for complex engineering jobs. Even in the next 3-5 years, AI experts predict only the most basic jobs, like data entry and administrative work will be fully replaced by AI. In more optimistic predictions, AI should make engineering vastly more enjoyable as engineers delegate the most mundane, time-consuming tasks to AI.  
  3. We don’t have good data to train AI. This is valid, as AI thrives on high-quality, well-structured data. The good news is that not all AI applications require that level of input. In fact, many modern AI tools can work with: unstructured or messy data (e.g., PDFs, emails, notes), small datasets, and human-in-the-loop systems that learn as they go.
  4. I can’t put my company data into this! It’ll be used to train other models. CoLab will never use your company’s data to train other AI models. All data used to feed ReviewAI stays in CoLabʼs environment on the AWS platform, and stays within the company’s space.
  5. My team isn’t ready. We have too much on our plates as it is. Unlike other SaaS companies who throw a tool at you and expect you to do the hard work, CoLab is different. We provide white glove service to every customer. That means CoLab and ReviewAI integrate with your current workflows without disrupting other projects. 

Short-term AI business results AI business results

Engineering teams who apply AI to future design reviews benefit from many short-term business gains. With CoLab + ReviewAI, engineering teams benefit from:

  • Faster design cycles
  • More informed decision-making
  • Less late-stage rework
  • Preventing repeat errors
  • Preserving institutional knowledge
  • Improving engineering quality-of-life

Long-term AI business results

As your team continues to use and apply AI, you’ll also be feeding ReviewAI high-quality data from your team’s qualitative feedback. This feedback loop means you’ll benefit from exponential business results.

Using a combination of AI case studies and proprietary CoLab data, we predict teams who use AI will:

Complete design cycles 4x faster

2x faster with a DES because:

  • Reviews take place in parallel, not in sequence
  • Automatically document and track design feedback
  • Source: CoLab Proprietary Customer Benchmarks

2x again with AI

  • Automate repetitive tasks
  • Cognitively demanding tasks benefit more from AI use
  • Source: A review of 3 published studies comparing empirical data from hands-on use showed that AI increases worker productivity by 66-126%

Reduce COPQ by 30% or more

15% lower COPQ in Y1 by fixing lowest performing review type

  • Earlier reviews pull risk forward
  • Design feedback from more internal and external stakeholders
  • Source: CoLab Proprietary Customer Benchmarks

Capture another 15%+ by expanding to more use cases and leveraging AI

  • Eliminate late-stage errors across more review types
  • Accelerate delivery of sustaining engineering projects
  • Source: 72% of late stage errors could be prevented (2023 engineering leadership survey; n = 250)

Reduce costs by 8-9 figures with engineering-led VA/VE

2x-3x idea pipeline value with a DES

  • 100% more cost reduction ideas per event
  • 6x number of events per year
  • Source: CoLab Proprietary Customer Benchmarks

4x-6x VA/VE program savings with DES + AI

  • $1M in additional savings realized per event with DES alone
  • Potential to double speed and effectiveness with AI
  • Source: CoLab Proprietary Customer Benchmarks

See 4x-8x shorter lead times 

2x-4x faster external collaboration 

  • Eliminate manual sharing and delays
  • Reduce back and forth required to clarify design intent
  • Source: CoLab Proprietary Customer Benchmarks

2x again with AI

  • Automate repetitive tasks
  • Consolidate and leverage design knowledge from multiple suppliers
  • Source: A review of 3 published studies comparing empirical data from hands-on use showed that AI increases worker productivity by 66-126%

If you’re planning to start “soon”, you’re already too late

The best engineering teams have a plan for AI today. It might not be perfect and it might not be fully operational, but they’re starting. And the best way to start anything is with a small pilot, then iterate and improve. 

This is the most challenging thing for large companies. Building a plan without the bureaucracy and red tape that surrounds enterprise decision-making. But, to stay ahead, this is what must happen. A few individuals need to take the reins and declare, “The time is now.” Sounds silly, but AI application is an exponential differentiator. Starting now and starting fast means reaping massive benefits over time. The only thing engineers lack today is the plan and the buy-in.

Here’s the plan. Now, go get the buy-in.

‍

Better products start with better design conversations.

If you’re ready to see how CoLab can transform the way you review designs, here’s how the process works.

Check
Every conversation starts with a discovery call. During this call, a CoLab rep will understand your team’s challenges with design review today.
Check
A tailored demo addresses your challenges and questions. This demo includes you and any key team members you’d like to include.
Check
Finally, we do a workshop with more members of your team. Your team will have temporary CoLab licenses and participate in a sample design review. You and your team will be guided through key features and use cases right inside the CoLab platform.
Schedule a consult