Beta Launch of Zenode Search Engine!

Hello Alpha Testers!  

We’re excited to announce that we have officially launched our component search engine!  We used the LLM you helped evaluate in our alpha to process ~10M datasheets, so now you can query across all of them in seconds!

While there are still areas we’re refining, the tool is already delivering some seriously cool results. Here’s how we envision the search process (but obviously use it however feels best!):

Query across all parts with an LLM search 

Just ask for exactly what you’re looking for, and the LLM will try to match things up.  Right now, it’s only searching specs and descriptions, but we’re working on a way where you can search directly into the datasheet.

Review the recommended parts

There comes a point in every search where you’ve run out of known specs to limit down the options.  At this point, we all just start randomly opening a bunch of parts and reading their datasheets to figure out what other options are available that you might care about.  

This is a super inefficient way to search, so we taught the AI to identify different groups within the results, and it pulls an example part from each to display at the top of the screen.  The theory is that reading these parts means that you are looking at *different* possibilities, rather than different versions of the same part, and can be more efficient as you read. And just to get ahead of it, we don’t actually know *why* the AI thinks these parts are different, as Neural Nets don’t explain themselves very well…

Obviously this is a very new concept, so we’d love your feedback on how it feels for you, and what we might be able to do with it to improve your flow towards finding the right parts faster!  

Use Machine Learning to ‘rank’ the results

The sole goal of all a part search is to find 1 component that will work, but usually there are dozens or even hundreds that could work (and hundreds or even thousands that could not 😳).

Because reading datasheets is so time consuming, we used to search by reading datasheets until we found 1 that would (hopefully) work, and then moving on. But now, with AI reading thousands of datasheets on our behalf in seconds, it's possible to actually OPTIMIZE and find the best possible part!

We're going to be improving our rankings for years to come, but this first version is pretty basic. The best way to improve this is to give the AI a few examples of parts that will and will not work for your needs, and let it use Machine Learning to pattern match amongst the remaining parts.  

So as you review parts, mark them as relevant (or not) to provide these examples.  Doing this for a handful of parts is the best way to ensure that the top options in the list will be the best for your needs! 

Thank You!

Seriously, thank you so much for all your help over the last year.  We could not have gotten to this point without your assistance, and now that it’s live, everytime I use it I get goosebumps because it feels like it’s the future of how we’re going to make hardware!  

So please, give it a whirl and let me know what you think, every single piece of feedback helps us to iterate towards something that will make all our lives easier! 

Cheers,

Brandon and Collin