A simple prompt that takes in an image prompt and outputs a better SD 1.5 one.

This bot is designed to ask to help students fill out scholarship applications from their school records and identity documents. It asks for their ID documents, schoolmarks and then visually understands the data inside in them and then builds to build a profile of the user. It's a great example of using GPT4-Turbo in conjunction with advanced OCR to recognize text (which GPTV doesn't do well), combined with audio speech recognition in a WhatsApp bot using 11labs

Takes code on Gooey functions pages and attempts to re-write the code so that it's better.

RAG + Azure ASR + TTS + AfroLLaMA

An AI Bot built for Goethe Bangalore

We are hosting SEA-LION for a limited time and would love feedback on it. Also available in Copilot. Based on the model here: https://github.com/aisingapore/sealion

Transcribe mp3, WhatsApp audio + wavs with OpenAI's Whisper or AI4Bharat / Bhashini ASR models. Optionally translate to any language too.

This example is the default analysis script to categorize Copilot conversations and to determine whether user's questions we successfully answered.

This recipe takes in a Copilot's Description and Instruction Prompt and then creates a shorter description and 4 sample questions.

A private bot for the participants of the AIPalace. It has all the public files here as its knowledge base: All Docs and Public AI Whitepaper

Yes, this is a silly (but useful!) AI recipe. Notice the differing emoji styles of Claude3.5 vs GPT-4o. :-)

This workflow creates a vector database (vectorDB) index on a collection of documents, links, pdfs, docx, txt files and then, given a query, searches them, returns page-level citation chunks and summarizes them with a GPT4 prompt. In particular, this workflow:

1. Takes in links to PDFs
2. Runs a synthetic data creation recipe on each scanned page of the PDF, to create a sample FAQ for each
3. Queries the vector DB with the search query
4. Gets back results
5. Summarizes the resulting chunks with the task instructions into an answer
6. Provides page-level URL citations for each referenced chuck returned from the vector DB search.
   All of this is available as an API as well, such that you can modify the source documents, prompt or synthetic data creation processing while maintaining the same API endpoint

Here we provide a reference image of the Sheikh Zayed Grand mosque & a matching image generation prompt (plus a URL of course), to create a QR code that closely mirrors the reference image. This technique of having a reference image + matching prompt works well to create scannable and beautiful QR Code.

To know more, head to our AI QR CODE HELP GUIDE!

Create interactive and engaging QR codes with stunning visuals that are amazing for marketing, branding, and more. Combining AI Art and QR Codes has never been easier!
Enter your URL and image prompt, and in just 30 seconds, we'll generate an artistic QR code tailored to your style.

Create interactive and engaging QR codes with stunning visuals that are amazing for marketing, branding, and more. Combining AI Art and QR Codes has never been easier!
Enter your URL and image prompt, and in just 30 seconds, we'll generate an artistic QR code tailored to your style.

Which language model works best for your prompt? What are the biases inherent in each? Compare LLaMA2, Gemini, Mistral, OpenAI GPT-4 engines with more LLMs being added each month.

This recipe is looks up the email's public social profile (from LinkedIn, Facebook, the web, etc) and then use the profile's name, employment history, city, etc to create a DallE3 image prompt.

Here we provide a reference image of the cute cat & a matching image generation prompt (plus a URL of course), to create a QR code that closely mirrors the reference image. This technique of having a reference image + matching prompt works well to create scannable and beautiful QR Code.

To know more, head to our AI QR CODE HELP GUIDE!

This workflow takes a Google doc of sample questions and golden expert created answers as an input. It then runs competing versions of the Ulangizi copilot (here with different LLMs - Gemini Pro 1 vs GPTV) and scores the answers. This helps us us determine if changes to our workflows actually increase performance (vs speed and cost as wel)

Transcribe mp3, WhatsApp audio + wavs with OpenAI's Whisper or AI4Bharat / Bhashini ASR models. Optionally translate to any language too.