Competition organizers should really consider extending the deadline

I’m using Genkit to build an object-oriented and developer-first library to remove some of the pain points for the average developer around using the Gemini API. I know that Gemini and Genkit are distinct, but Genkit’s realistically the only practical option for the average developer to work with the Gemini API at this time. Even then, Genkit’s still in beta, and I and other developers continue to find bugs and limitations of its integration with the Gemini API: see this Github issue.

My plan is to open source the library I’m building in order to make Gemini accessible to even more developers, but I’d also love to release it in time to submit to the competition. My concern is that only large teams with existing infrastructure will be able to compete effectively, given that Genkit’s the most-developer friendly way to leverage the Gemini API but still has limitations blocking development. I’d love to see diverse and competitive entries, so I really hope the competition organizers will extend the deadline for a few weeks in order to compensate for the current barriers developers face when working with the Gemini API and its integrations.

Thank you kindly for your attention, and I wish everyone good luck.

5 Likes

Welcome to the forum!

I can’t speak to extending the deadline (I wouldn’t mind more time, to be sure!), but I did want to address one comment you made:

I’d urge you to reconsider.

There are plenty of good, open source, well supported, libraries that are friendly to developers.

  • Genkit, itself, is open source - so you’re certainly encouraged to contribute fixes you see. (The issue you outlined appears to be either a bug in the underlying SDK from Google, or the API itself.)
  • Langchain.js supports both the AI Studio and Vertex AI Gemini (disclaimer - I’m a big contributor to these packages) with several pre-built tools and other integrations and a large developer community.
  • I’m much less familiar with LlamaIndex, but I do see that it has Gemini support.

I understand the desire to build your own library. (Oh, trust me… how I understand.) But there are only so many resources, and I’d rather see folks contribute to making existing good projects even better.

Good luck with your project!

2 Likes

I found out about this competition in late July. I’ve never coded in flutter and I’m a 1 man team that learned and developed everything as well as shot and edited the video, in time. You really just have to go ham!

2 Likes

just submitted now haha i saw the competition early this month and am also working on it solo, you don’t have to make it perfect just submit it anyways.

4 Likes

Firms at Google’s scale face a hiring trade-off: they sacrifice idea-generation for “fungibility”.

At that scale, HR departments become more prone to confirmation bias, begin over-fitting, and fall prey to Goodhart’s Law: “When a measure becomes a target, it ceases to be a good measure.”

Ultimately, firms end up in situations like today, where there are lots of working engineers who can bang out identical Leetcode solutions all day long, but too often struggle to effectively synergize their work with other developers, or generate novel ideas.

There isn’t a great solution around this problem for firms at scale though, because they’re right to operate this way from a short-term profit-seeking and survival perspective, but they’re still missing out on ideas and other opportunities that will ultimately cost them their survival in the long-run.

So, what’s a firm to do in order to cast a wider net and diversify its idea-generation process?

Run a competition.

Genkit is the most developer-friendly AI toolkit available, was released contemporaneously with the Gemini competition, and entrants were encouraged to use it, which was a smart move on Google’s part in order to cast a wider net and bring in more diverse ideas.

That you, I, or anyone else were able to participate and submit our projects on time is beside the point, because we probably aren’t the real maverick idea-generators who would benefit the most from higher-level tools like Genkit, and then pass those ideas on to Google for the competition.

Good for us, but not great for users, the world, or even Google.

There are smarter people than anyone here who are struggling to work and compete for a limited number of hours a day from a library computer or a homeless shelter, and those people will benefit the most from tools like Genkit, and the novelty of their ideas will benefit Google the most in a competition like this.

Any engineer who doesn’t realize there are smarter people than themselves going to waste in society needs to check both their Ego and the statistics.

Ultimately, whether it’s this year or next decade, the presently mild stagnation of ideas in corporate tech firms will escalate into a full-blown crisis, which we know will happen because we’ve seen it happen time and time again as the natural result of creative destruction.

Despite the ubiquity of its user-facing apps, Google knows its real future lay in providing developer-facing solutions like databases and AI/ML platforms, which makes tools like Genkit more critical to Google’s competitiveness, and even its long-run survival, than Gmail, Maps, or YouTube.

Google’s long-run competitiveness likely depends on how many developers it can get to use tools like Genkit, and how fresh it can keep its well of ideas, whether those ideas spring internally, through acquisitions, or through competitions like this.

Google knows it needs diverse ideas for its own evolutionary survival, not to meet DEI metrics, which is largely why it hosts these kinds of competitions, and it knows that it needs diverse developers to adopt its solutions as the foundation of the future.

It’s in Google’s own best interest in the long-run to schedule a competition like this at a pace that optimizes the novelty of ideas generated, not merely at a pace to compete with the release schedule of a competitor’s products, for example.

I’m sure that Google has received lots of competitive entries, but waiting until Genkit was better documented and tested, and I think giving the contest a sixth month schedule, would have generated even more competitive and novel entries, which would have been in everyone’s best interest.

The Genkit team is phenomenally generous with their time, and I’m very grateful to the organizers of the competition because it was really fun and well-orchestrated, but I can’t help but wonder if its schedule wasn’t chosen to maintain mindshare more than optimize for novelty.

Personally, I’m competing for everyone who hasn’t otherwise been given the opportunity.

Good luck to everyone.

2 Likes

There is not a single usable ODM for Firebase as one only one example, and, as well-engineered of a foundation as Genkit is, it still requires a lot of unnecessary boilerplate that will discourage developers from many backgrounds and of all skill levels.

I very much agree with the Firebase team that a modular, procedural style is preferable to OO in order to provide SDKs and tooling at the “service” level for Firestore or Genkit; however, there’s no reason for product developers to have to reinvent the wheel every time they want to simply add a document to Firestore instead of using a sensible ODM.

Present tooling is still painfully lacking for many development styles, and it’s ludicrously inefficient just to tell developers “think this way because that’s how everyone else does it.”

We need sensible object-oriented tooling, agreed that it should be built on easily testable and procedural foundations, because otherwise people will continue to hack together ad-hoc object-oriented solutions on a per-project basis, which I still see happen regularly.

The lack of diverse tooling is also part of why Firestore, despite being wildly powerful, still maintains an undeserved reputation as being limited to prototyping and testing, which is ridiculous.

There is simply never a good reason to discourage anyone from open-sourcing a solution, especially when it makes development safer and more accessible…what happened to developer/hacker culture?

1 Like

Hmmm, interesting take.

Compelling and very factual. However if you want to contribute to the tooling, google hosts many meetups locally and nationally like google I/o where you can provide direct feedback to the direct teams. I did propose my own suggestions to the firebase team at google io Miami last year, it was a great time and am very thankful to google for hosting these events.

First of all, I only learned about the competition very late and had just 5 days to prepare. Today, I completed a major project involving both the app and backend a few hours ago. I intended to create a video and began entering the submission details. However, after filling out just three fields, guess what happened? The electricity went out! After a couple of hours, when the power was restored, I discovered that the submission period had ended. All my hard work, late nights, and health sacrifices were in vain. Unfortunately, they didn’t offer any extension, not even an extra day or two. To top it off, I had a headache from the stress and anger :confounded:.

Hey, you can email lloydhightower+geminiapi@google.com with your form details as an extension. Make sure to submit it by 12:00 pm EST.