Shah, the 31-year-old CEO and founder of Bengaluru-based Duukan, which helps merchants to set up online stores and sell products digitally, posted that "We had to layoff...
… there’s no way their tech is good enough to avoid the billion and one ways to manipulate and confuse current-gen LLMs…
Unless he went with older tech, just feeding specific pre-programmed responses to specific keywords. That’s a glorified phone tree though, nothing really new.
There’s an in between. Have the LLM map into predefined responses, thus leveraging the LLM’s insane parsing abilities with the linearity of tree-like structures.
At that point it’s probably cheaper to purchase support staff in SEA or Eastern Europe… staffing a software product team with ML and prompt engineers for a bespoke support solution is costly.
Granted they probably went with a vendor, but that model has its own costly drawbacks.
… there’s no way their tech is good enough to avoid the billion and one ways to manipulate and confuse current-gen LLMs…
Unless he went with older tech, just feeding specific pre-programmed responses to specific keywords. That’s a glorified phone tree though, nothing really new.
There’s an in between. Have the LLM map into predefined responses, thus leveraging the LLM’s insane parsing abilities with the linearity of tree-like structures.
It can be done, and can be done intelligently.
Which I somewhat doubt was the case here.
At that point it’s probably cheaper to purchase support staff in SEA or Eastern Europe… staffing a software product team with ML and prompt engineers for a bespoke support solution is costly.
Granted they probably went with a vendor, but that model has its own costly drawbacks.