FlowiseAI Profile picture
Jun 22 5 tweets 2 min read Twitter logo Read on Twitter
Function usually takes in structured data. But it's hard to tell LLM to return structured data, until @OpenAI Function Calling.

In Flowise v1.2.13, we are introducing - Custom Tool ✨

Together with @LangChainAI agent, it will choose the right tool to call the function

🧵
🛠️ Custom Tool (1/2)

Create your own custom tool by specifying:
✅ Name
✅ Description
✅ Output Schema
✅ Function
🛠️ Custom Tool (2/2)

ChatGPT will decides:
🤔When to use the tool
🧱Return a structured data
⚡Call the function using the data

Result is way more reliable and way faster!

Here's an example of getting stock price movement and add it to @airtable inspired by @jasonzhou1993👇
🤖 OpenAI Agent

The new OpenAI Agent is armed with:
💿Memory
💬System prompt

Making it much more usable and reliable!
📚 More Document Loaders

✅Puppeteer
✅Playwright
✅Subtitles
✅Figma
✅JSONL docs loader

✂️Code Text Splitter
Split text by 17 programming languages👀

Thanks to all contributors! @aidantilgner, @govindkumar73, @pravesh295, and many others!

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with FlowiseAI

FlowiseAI Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @FlowiseAI

May 23
🔍Metadata Filter is such an underrated topic.

When you have more vectors stored in vector database, it becomes harder to retrieve top relative vectors, resulting in irrelevant answer from LLM.

Watch how you can specify metadata to your documents and later filter by it 👇 Image
With Flowise, you can specify a set of key-value pair as metadata to your document.

Most commonly used key value pairs are:
📚 document_id
🧑 user_id

The idea is to have a unique value for each document to better differentiate between them
When retrieving, you can then specify the metadata key-value pair.

This greatly increases the query flexibility, enabling much more advanced and targeted searches on your vector data ⚡
Read 4 tweets
May 17
🥷 We've gone stealth for the past 2 weeks. Now back with Flowise v1.2.6:

💬 Embedded Chat Widget
🔍 Metadata Filter
🌿 Support for AzureOpenAI, 100k-Claude @AnthropicAI models
⚡ Zapier NLA, WebBrowser tools
🔐 Apps authentication

🧵
💬 Embedded Chat Widget

Simply embed a chat widget to your websites that connects to your flow.

Watch how you can insert a link, and have a chat bot dedicated to answer question of the website 👇
🔍 Metadata Filter

Q: How to retrieve documents related to specific keyword such as userid🤔

First, we add metadata key-value pairs to documents:
Read 9 tweets
Apr 12
Want to quickly experiment and build LLM apps but lazy to write the repetitive code?

Introducing Flowise - an open source UI visual tool for @LangChainAI JS, written in Node Typescript.

Repo:

Here's 7 templates ready to be used 🧵 https://t.co/jN9XuwxYk2github.com/FlowiseAI/Flow…
Basic example of LLM Chain with a Prompt Template and LLM Model:

Source: https://t.co/JpwxGBjnzUjs.langchain.com/docs/getting-s…
MRKL Agent: an agent that uses the ReAct Framework to decide what action to take:

Source: https://t.co/VkcpSt5GKKjs.langchain.com/docs/modules/a…
Read 9 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(