Superluminal @superluminalapi

Copilot for your data dashboard. superluminal.dev Earth Joined April 2021

Tweets

18
Followers

340
Following

50
Likes

57

Superluminal @superluminalapi

3 years ago

Integrating Superluminal into your dashboard is now easier than ever with our new React component. Check out our 5 minute tutorial here! superluminal.dev/docs/#react-tu… #ai #data

0 1 6 312 0

View Details

We often use a technique called prompt routing to better optimize prompts for particular use cases, while increasing cost-effectiveness and improving performance. Read more about what it is and how it works here: getluminal.com/articles/promp… #AI #data #ChatGPT

2 1 4 296 1

View Details

Superluminal @superluminalapi

3 years ago

Introducing... Superluminal AI 2.0 Our goal is to build the most powerful data AI in the world, and today we took a huge step in that direction. Instead of using a fixed order of operations to complete a task, the Superluminal AI now iteratively picks what action to perform to complete a task until it's finished. This enables it to solve tasks it couldn't before, and paves the way for even more upgrades in the future! Try it now at getluminal.com #ai #data

0 2 6 498 2

View Details

Superluminal @superluminalapi

3 years ago

Minimizing GPT-4 Cost When you send GPT-4 a prompt, it doesn't see the individual characters—it only sees tokens. While perhaps unintuive at first, understanding tokens well is key to optimizing prompt speed & cost. We've built a free online tool to do just that. Read more at: getluminal.com/articles/token… #ai #data

0 1 3 213 0

View Details

Superluminal @superluminalapi

3 years ago

Optimizing GPT-4 Response Time A key aspect of building production-level AI data interaction is performance. Specifically, how long it takes to get a response from the model you're using. At Superluminal, we use GPT-4 for parts of our stack. And though it's extremely powerful, it can still be slow depending on the task. OpenAI has been working on improving the performance of GPT-4, but we didn't want to passively wait for OpenAI to make GPT-4 faster. So, we decided to look into how to optimize GPT-4's response time. Read more at: getluminal.com/articles/optim… #ai #data

0 1 2 247 0

View Details

Superluminal @superluminalapi

3 years ago

Introducing the Superluminal API! In building Superluminal Lab, our natural language data interaction platform, we came to realize that although impressive looking demos are everywhere, building production-quality AI data interaction is hard. And we discovered that many other businesses are facing the same challenge. So, we decided to make our API public! Now you can ship production-quality, natural language data interaction to your users with just a few API calls. Read more at: getluminal.com/articles/api.h… #ai #data

0 2 3 304 0

View Details

Superluminal @superluminalapi

3 years ago

@blcartwright @Luke_WA Hey Brian, that's a lot of personal data! Luminal doesn't work with SQL yet, but will in the future — we'll let you know when it does so you can give it a spin!

0 0 0 3 0

View Details

Superluminal @superluminalapi

4 years ago

Thanks @kateinoigakukun for including us in your talk at #iosdc!

Yoshimasa Niwa @niw

4 years ago

While attending #iosdc in Tokyo, watched @kateinoigakukun’s Swift Wasm presentation and suddenly coming across @ppijnappel’s @getluminal. Small world for sure, also wondering how it works.

0 1 9 0 0

0 0 6 0 0

View Details

Superluminal @superluminalapi

4 years ago

Beat our online Swift coding challenge and get a $75 reward! Know a great software engineer using Swift? Have them beat the challenge and earn $25 for you + $75 for them. getluminal.com/challenge

1 6 60 0 2

View Details

Superluminal @superluminalapi

5 years ago

@9ninevolt Only kind of thanks to auto-complete!

1 0 0 0 0

View Details

Superluminal @superluminalapi

5 years ago

@0x606 We're indeed using shared memory–and definitely agreed switching to a binary format close to the in-memory representations (+ minimizing JS boxing) will be the way to go once encode/decode becomes significant.

0 0 0 0 0

View Details

Superluminal @superluminalapi

5 years ago

At Luminal, we wanted to use a modern, strongly typed, performant language for the application core. We also wanted cross-platform UI and the ability to run in the browser. So we built (the world's first?) Swift ↔ Electron IPC. @SwiftLang 🤝 @electronjs