4 ms·
> unless you have "AI" in your name of course Being an AI startup that was founded before about 1 year ago is actually a liability. The way we thank about AI
by CSMastermind 2y ago
> unless you have "AI" in your name of course
Being an AI startup that was founded before about 1 year ago is actually a liability.
The way we thank about AI has radically changed in the last year, maybe last two years if you were really forward thinking.
Any AI company before then will have a mountain of business logic and technical architecture that they need to throw away and redo.
- t-writescode 2y agoWhy? Were the statistics, etc, not useful? I would assume that if the underlying models were already good, about the only thing an LLM might be good for would be filtering new info and/or presenting information in a useful way, both of which can be added to an existing flow?
- Dr4kn 2y agoDeepL for example already used machine learning for their translations. Over the years they accumulated a massive amount of high quality data thanks to language experts they employ. Their current model is much better, because their training data is much better than feeding everything in.
- danielmarkbruce 2y ago100%. This is misunderstood by investors.