Too risky to build things in Generative AI right now?
Yash
34 replies
- New startups & products launched daily
- Old startups in the same space pivoting
- Giants moving like startups
What's your strategy in this chaos?
Replies
Benard Ogutu@benard_ogutu
I'll still take my chances 😅
Share
I'd still go ahead build it. The experience will elevate you to next level!
If you just connect to GPT and that's it, then yes, because there's no moat and it's a soon-to-be saturated market.
Still in this space, I don't think there's a winner yet. The key here is to understand the value your startup is giving, clear moat (apply strong prompt engineering to connect with the LLM), and execute it very well.
Focus on next 5 years not in daily or monthly basis
embrace the chaos
The important part is to 1) stay focused on improving your customer 2) keep innovating and stay different. Generative AI is evolving so fast that it is still possible to do both.
While there are risks associated with generative AI, it is possible to manage these risks through careful planning and implementation, allowing organizations to take advantage of the technology's potential benefits.
@random_video_md Good thoughts, Mukesh! How would you recommend these risks are managed?
@yashthakker To avoid problems when using generative AI, companies should test it well, be fair and transparent in how they use it, consider how it might affect different people, and keep an eye on how well it's working.
I would focus on creating AI generative products that focus on synthesis and decision making enhancement rather than content generation based on instructions
Try and have fun I guess :)
I would say right now is the best time and least "risky" for generative AI
You don't want to wait until the market is saturated
But the same logic applies: build something that solves a need. If you're just building a GPT wrapper with no or little prompt tuning, then that would not yield strong returns.
But if you're applying lots of prompting or having your own AI model like in image generation, then it should be much better
This is what we're doing with evoke-app.com
There's going to be a massive massive new opportunity with GPT Plugins. You could train your AI model in a very specific problem and sell it as a GPT plugin.
We're in a pioneering phase right now in AI, lots of opportunity to be made if you know where to dig.
@justin_govy awesome! what are you guys working on?
Our strategy is to use the best new tools to meet our organizational goals. But that has not changed. We were already deploying generative AI into our platform. That said the better the tools become the more useful they are, the more we deploy them. Key thing here is that being an AI company - which we have been for 5 years - was never an aim in itself. We deploy AI to enhance human communication with our platform Conversa - 2.0 launching on PH 20th April
You have a largely of model risk (i.e., OpenAI, Google, et al change er ugprade the model without telling you)
Then you have vendor risk. It's a variety from the above, but your API connection could be removed at any point in time.
Hence, I have become a proponent of Open Source, Local LLMs, and proper evaluation frameworks for prompt quality