PLEASE READ THE DISCLAIMER AT THE END
I wrote it in my analysis on MapMyIndia. (I was very Bullish. I sold MMI immediately after the obvious ethical concerns.) In this analysis I have written on Ola Krutrim AI, I am copying it here.
- Concerns About Ola: I have reservations about Ola, particularly regarding the CEO. My biggest concern is He might not be telling the truth. Let’s see everything -
- Their food delivery service has struggled.
- The Ola Cab business remains unprofitable. Although all such businesses are unprofitable. (We can ignore this one)
- While Ola Electric has the highest market share in India, quality issues could hurt its popularity over time, and their market share is primarily due to the reach of their OLA App.
- I found a rumor claiming “Krutrim uses ChatGPT,”. I did some digging and found - it actually relies on Google, raising credibility issues. To be precise - it’s google Dialogflow
- Ola Maps likely lacks legitimacy as a competitor. They seem to have used MapMyIndia’s data without acknowledgment, relying on the OpenStreetMap API while “failing” to inform the public.
Overall, I don’t view Ola as a significant competitor to MapMyIndia in the navigation sector. Or anyone else for that matter.
Above I asked it the same question in Two different ways.
Below, you can see that Google Dialogue flow was released in 2020.
I have seen a minor glitch on their website, where its reply accepts that it uses OpenAI’s chatGPT and quickly after the glitch, that text is removed automatically and new reply comes. It means that the original rumor is right -“They do use OpenAI”. They might be using both of them. I have seen this glitch, but I couldn’t screenshot it. It is not wrong to use ChatGPT, and build on top of that. However that isn’t what they are claiming. On the contrary, they are trying to hide it, albeit unsuccessfully.
[Copied till this point from here]
Optional
I work in Tech and have been exploring Generative AI for the past couple of months (only). Krutrim AI is not what Ola claims it to be.
This is my speculation regarding Ola’s Krutrim’s working:
-
While Fine-tuning a large language model (like ChatGPT, which is an LLM) can customize its outputs, Ola did a terrible job at it. Fine-tuning is a resource-intensive process, and since Krutrim isn’t making money, it’s not surprising they didn’t invest properly in it. This feels like a classic ‘fake it till you make it’ situation—but honestly, I doubt Ola Krutrim will ever make it. For Ola in particular, you can say things like “Sab kuch karna hein Isko”, “Bahar sab chakaachak hein, andar sab polampol hein”
-
By the way, the pre-training and fine-tuning processes for Generative AI models are key reasons why NVIDIA has reached its lifetime high. Essentially, NVIDIA is leading the charge in the AI boom. Indian techies should onboard it. Unlike Ola who used it the worst way possible
Disclaimer - Clearly my tone here is full of conviction. I am just honestly expressing what I am thinking. However, it is your responsibility - to check things for yourself. Keep in mind that, technically, Krutrim is different from Ola electric. That said, I do respect Bhavish’s courage in pursuing his vision.
The reason why I asked you to read the disclaimer first is that, there is something called Conviction Bias - When a person is full of conviction, it becomes easier for him to persuade you. Thus, they say that “Conviction bias is contagious”. My goal is not to persuade you, yet I might end up doing that. So, please be critical of this text as well.
Let me know if you disagree anywhere, except for the obvious decisive tone, I am aware of that one.