Magical LLMs and working with new tech
This has been the year of large language models. Everyone wants to build an AI startup. Everyone wants to use ChatGPT for everything. LLMs open up a lot of possibilities; they are powerful. But an LLM is not a magical black box to throw stuff into. When you think of them as small reasoning engines, they are more reliable and helpful.
It’s crucial to find the right application for LLMs because the infrastructure needed for them is still quite expensive. You should identify specific places where a tiny reasoning engine will improve experiences and ensure that this is high-ROI for your business. A great example is an LLM powered chatbot that significantly reduces the cost of your customer support team.
LLMs can be used in a few different ways in your product:
A feature-level solution could take the form of inline context menus like Linear or Notion.
A platform-level solution could take the form of a copilot/chat as a capability across your tool like Copilot or Bard.
And a chat solution is a good idea if your product benefits from being personalised and conversational, like Intercom and Duolingo.
We’ve put together a guide on How to Build with LLMs and How to Craft LLM Experiences in Obvious University if you’d like to read more.
Until next time!