Build an LLM app

According to a report from Bloomberg’s Mark Gurman, Apple is racing to catch up with OpenAI and Google. Internally, their new chatbot is apparently being called ‘Apple GPT.' As Danny Lev pointed out, “Apple's focus on privacy could be a big advantage in the AI race.” What do you think?

LangChain’s new developer platform is a milestone on the road to democratizing access to the tools needed to build the new era of LLM-powered applications.

Let's take a step back: LLMs—aka large language models—refer to the type of software that powers products like ChatGPT, Bard, and Bing.

An LLM is ‘large’ in the sense that it is trained on massive datasets; ‘language’ in that it can ‘understand’ and ‘respond’ in the same ways humans communicate; and a ‘model’ in that it is trained to make predictions based on the patterns it finds in its training data.

Wrangling complexity: LLMs are good at a number of very useful things—including text generation, translation, and summarization—and tech companies make their LLMs available to independent makers via APIs like Google’s PaLM or Meta’s Llama, but that doesn’t mean developing new products with these LLMs isn’t still extremely complex.

Thanks to new developer platforms like LangSmith, makers can now more easily build, deploy, debug, test, evaluate, and monitor their LLM apps, without having to build that tooling in-house.

For the love of tinkering: “We started with an open-source Python package when the main blocker for building LLM-powered applications was getting a simple prototype working. We remember seeing Nat Friedman tweet in late 2022 that there was “not enough tinkering happening,” says LangChain’s launch announcement.

Less than a year later, “people are building everything from chatbots over internal company documents to an AI dungeon master for a Dungeons and Dragons game […] We see lots of tinkering happening now (Nat agrees).”

• Fabric for iOS and Android is a home for notes, links, screenshots and files.
• Swiftbrief combines keyword clustering, data and AI for SEO writers.
• ElevateHQ automates sales commission plans for sales teams.
• Second Nature’s AI Simulation provides training through simulated conversations.

• Everfund integrates flexible donation experiences in any website.
• Superthread (0.1) is a task tracker with roadmaps, issues, and documents.
• ComeUp connects freelancers and service sellers to companies and buyers.