Jev AI - a strategic view
Jev is an AI model with natural language context and questions similar to other AI Large Language Models.
Jev however has taken a differentiated approach to natural language AI, it does what it does well and has made tradeoffs to get there. Unlike other AI models, Jev does not generate text so you cannot ask it to write your emails for you. Instead you give it context and give it a list of possible answers you want it to chose from.
For example giving it a conversation with a user and asking for sentiment analysis such as “Is this conversation positive or negative in tone”. The answer is either yes or no - with a confidence score. Or asking to chose from the selection “happy, angry, neither” - with a confidence score.
You can achieve the above using traditional LLMs so why have a special AI? Jev has made the tradeoff to limit its functionality, in return for low cost. Jev does not charge for the output like traditional LLMs (as the options are not created, they are chosen from a set list) and the input is 444x cheaper than traditional LLMs (and 193x faster). This system has given Jev a competitive advantage - a way to win the market segment it is going after.
Jev has chosen to make a differentiated AI product, made key decisions and choices that mean it is not being commoditised. It cannot serve every use case of a traditional LLM, but the uses cases it does have it does faster and cheaper.
Is it a low end disruptor?
In Clayton Christensen’s disruption theory a low end disruption happens when users are overserved, and a simpler cheaper product comes along. This product then increases its capability over time to eventually overthrow the competitors.
Jev is not a low end disruptor, it meets some but not all the requirements. It is cheap and works well for scenarios where full power LLMs overserve users, but it does not directly compete with LLMs as it cannot do everything they can do (nor does it have a path to do so). Jev will not disrupt traditional LLMs as it is not designed to write your emails for you.