flâneur

The Perfect AI Assistant - Caelin Sutch

substack.com · 1,176 words · saved by 1 readers

Since shutting down Lookbk, I’ve been thinking a lot about the concept of an AI personal assistant. To an extent we started to build a narrow niche of that with Lookbk - a mixture of proactive alerting (for sales, new releases, updates on influencers you follow) with raw utility (search, price watching) - but specifically for fashion. In my mind, what makes an AI assistant better than straight Chat GPT revolves around what I call “Always On” computing. Traditionally, we interact with LLMs in a synchronous one-off way, we ask a question, it responds back after a period of time, rinse repeat throughout your convo. “Always on” takes this and adds an event-driven dimension that reacts either based on external input other than a person texting it With this, the LLM is acting proactively either on your data, a set cron, or some sort of task system that operates in the background. This means that you texting an LLM is simply the entry point to a world of AI agents that work behind the scenes

Since shutting down Lookbk, I’ve been thinking a lot about the concept of an AI personal assistant. To an extent we started to build a narrow niche of that with Lookbk - a mixture of proactive alerting (for sales, new releases, updates on influencers you follow) with raw utility (search, price watching) - but specifically for fashion. In my mind, what makes an AI assistant better than straight Chat GPT revolves around what I call “Always On” computing. Traditionally, we interact with LLMs in a synchronous one-off way, we ask a question, it responds back after a period of time, rinse repeat…

saved by

related reading