Dust

Dust is an AI-powered tool that is helping individuals and businesses solve unique problems with AI. It is built on Chained LLM apps, which enables users to input multiple options when choosing the model, making it highly versatile. Dust features caching to speed up access to frequently used models, making solving complex issues easier. The tool is easy to deploy and use, making it an excellent asset for those who want to quickly address complex issues. Its semantic search feature allows users to ask questions about the IPCC AR6 report and get high-quality answers with references to independent sources. The tool also solves the blank page problem for wedding thank-you notes, making the task of showing gratitude to guests much easier. Dust is an exceptional AI tool that provides several integration blocks for users to easily integrate with external systems. With its pricing model based on the number of chained LLM apps it provides, Dust offers a range of pricing options that cater to the needs of both individuals and businesses.

TLDR

Dust is a highly versatile AI-powered tool built on Chained LLM apps that allow users to input multiple options when choosing the model, making it easy to find the most suitable solution for unique problems. Its version history is impeccable, and it features caching for faster access to frequently used models. Dust's semantic search feature provides high-quality answers with references to independent sources. The tool also solves the blank page problem for wedding thank-you notes and allows users to answer questions with high factual accuracy by searching online and compiling responses based on content downloaded from websites (with references). Dust offers several integration blocks for users to easily integrate with external systems, making it an excellent tool for those who want accurate, reliable, and fast services. Its flexible pricing model is based on the number of chained LLM apps, catering to the needs of both individuals and businesses. Dust is an exceptional tool with a user-friendly interface, making it easy to deploy and use, and showcases the full potential of what AI can offer.

Company Overview

Dust is an AI-powered tool that provides a range of applications to solve various problems. The tool is built on Chained LLM apps, which allows users to input multiple options when choosing the model. Dust's version history is impeccable, making it easy for users to keep track of updates and changes. It also features caching, which is useful for speeding up access to frequently used models. The tool is easy to deploy and use, with clear data sources and connections.

Dust's app community is a highlight for those who want to get started quickly. It provides great examples that showcase the full potential of the tool. The semantic search feature of Dust allows users to ask questions about the IPCC AR6 report, and get high-quality answers with references to independent sources. It also solves the blank page problem for wedding thank you notes, making the task of showing gratitude to guests much easier.

With Dust, you can also use the web search assistant to answer questions with high factual accuracy by searching online and compiling responses using content downloaded from websites. The maths generate code function is excellent for answering maths questions by providing complete code generated by Dust's algorithms. The toolbot repro function also teaches LLM to teach itself new tasks by generating few shot examples from high-level tasks descriptions and then prompts itself with them.

Dust is a highly versatile tool that is helping businesses and individuals solve unique problems with AI. Its user-friendly interface coupled with its advanced features makes it a great asset for those who want to quickly address complex issues.

Features

Chained LLM Apps

Input Multiple Options

Dust is an AI-powered tool built on Chained LLM apps that allows users to input multiple options when choosing the model. This feature lets you choose from a range of options that can help you find the most suitable solution for your unique problem.

Impeccable Version History

Dust's version history is impeccable, making it easy for users to keep track of updates and changes. With every update, the tool becomes more robust and equipped with the latest AI features.

Speeding Up Access with Caching

Dust features caching, which allows for faster access to frequently used models. This is highly useful when solving complex issues that require repeated use of certain models.

Semantic Search Feature

Questions about IPCC AR6 Report

The semantic search feature of Dust allows users to ask questions about the IPCC AR6 report and get high-quality answers with references to independent sources. This feature is highly useful for those conducting research on climate change and its impact on the planet.

Blank Page Problem Solver

Dust also solves the blank page problem for wedding thank-you notes. This feature makes the task of showing gratitude to guests much easier by generating personalized messages in just a few clicks.

Factual Accuracy with Web Search Assistant

Dust's web search assistant feature allows you to search online and compile responses using content downloaded from websites. This feature ensures high factual accuracy, making it highly beneficial for researchers, students, and professionals.

Maths Generate Code Function

Complete Code Generation

Dust's maths generate code function is excellent for answering math questions by providing complete code generated by its advanced algorithms. This feature is highly useful for engineers, scientists, and mathematicians who work with complex equations.

Teaching LLM to Teach Itself

The toolbot repro function of Dust teaches LLM to teach itself new tasks by generating few shot examples from high-level task descriptions and then prompts itself with them. This function is highly beneficial for those seeking to automate or improve their workflow processes, saving time and increasing efficiency.

User-Friendly Interface

Dust's user-friendly interface, coupled with its advanced features, makes it a great asset for those who want to quickly address complex issues. This feature ensures that even those without prior experience in AI can quickly deploy and use the tool with clear data sources and connections.

App Community

Great Examples

The app community of Dust is a highlight for those who want to get started quickly. It provides great examples that showcase the full potential of Dust, allowing users to easily understand how the tool can solve their unique problems.

Highest Versatility

Dust is a highly versatile tool that is helping businesses and individuals solve unique problems with AI. The tool's highly adaptable nature makes it an excellent asset for those looking to explore the benefits of integrating AI into their work processes.

Full Potential

The Dust app community also provides users with the opportunity to showcase the full potential of the tool. With a wide range of use cases, Dust ensures that users get the most out of the tool, regardless of their industry or domain.

Integrations

Dust's tool provides several integration blocks for users to easily integrate with external systems, mainly for retrieving data. These blocks can be used to search the web, retrieve content, or make API calls. Each block comes with its own set of specification parameters and configuration parameters. Below are the details on the integration blocks currently offered:

Curl Block

The curl block is primarily used to make HTTP(s) requests to external systems. Users can use this block, for example, to call into a Replit endpoint to perform advanced computations or retrieve data from an internal or private API. Users can specify the HTTP method to use (e.g. GET, POST, PUT, PATCH) and the URL to use for the request. The URL parameter can utilize templating to dynamically generate the URL based on previously executed block outputs. Additionally, users need to provide two Javascript functions. One is the _fun function, taking a single argument env (which should be defined in Code block documentation), which returns null (no headers) or an object representing the headers to use for the request. The other is also the _fun function, which takes a single argument env and returns null (no body) or a string representing the body to use for the request. Users can choose to enable the automated caching mechanism of the block, which will return the cached response if a previous request with the same specification parameters was made.

Search Block

Dust's Search block is currently a wrapper around external search engines, such as SerpApi and Serper. Users can set up the SerpApi and/or Serper provider in their account's provider settings. Users can specify the search engine to use, the query to use for the search (which can rely on templating to dynamically generate the query based on previously executed block outputs), the number of results to return (defaults to 10), and the search provider to use (either serpapi or serper). Users can choose to enable the automated caching mechanism of the block, which will return the cached response if a previous search with the same specification parameters was made.

Browser Block

The Browser block is used to scrape content from web pages and is a wrapper around Browserless.io. Users can set up the Browserless provider in their account's provider settings. Users can specify the URL to scrape (which can rely on templating to dynamically generate the URL based on previously executed block outputs), the CSS selector to use to select the content to scrape (use body to scrape the entire page), the timeout in ms to use for the scrape request (defaults to 16000), and one of load, domcontentloaded, networkidle0, networkidle2 to specify the wait event to happen to return content. If set to true, users can enable the automated caching mechanism of the block, which will return the cached response if a previous request with the same specification parameters was made. By default, the block will error if the scrape request times out, interrupting the flow of execution of the app. If set to true, the block will return the error as an object and execution will continue.

Pricing

Dust offers a range of pricing options that cater to the needs of both individuals and businesses. Their unique pricing model is based on the number of chained LLM apps, which are designed to provide multiple inputs and allow users to choose from different models. Dust offers a free trial for users who want to test their tools and services before committing to a paid plan.

The paid plans start with a basic plan priced at $29 per month, which includes one chained LLM app with limited access to data sources and connections. The standard plan, priced at $99 per month, offers three chained LLM apps, access to unlimited data sources and connections, and the ability to choose from different models. The professional plan, priced at $399 per month, offers five chained LLM apps, version history, caching, easy deployment and use, access to all data sources and connections, and the ability to choose from different models.

Dust also offers a community plan which is coming soon. This plan will allow users to discover apps created by the Dust community, providing great examples for users to get started with Dust.

No matter which plan you choose, Dust provides great tools and services that are designed to be easy to use and provide high factual accuracy. With their semantic search, users can easily find answers to questions about IPCC AR6 report, generate code to solve complex maths questions and teach LLM to teach itself new tasks. Additionally, Dust's web search assistant allows users to answer questions with high factual accuracy by searching online and compiling responses based on content downloaded from websites (with references).

Overall, Dust is a great choice for individuals and businesses who need accurate, reliable tools and services to help them solve complex problems. Their pricing model is flexible and affordable, and their tools and services are designed to be easy to use and deploy. Sign up for a free trial today to see for yourself what Dust can do for you.

FAQ

What is Dust?

Dust is an AI-powered platform that enables users to create intelligent applications with pre-built components, data sources, and integrations. The platform allows businesses, developers, and data scientists to easily build AI-powered applications with just a few clicks.

What is the Document Q&A app?

The Document Q&A app is a tool created using Dust's platform that uses Data Sources to build and index a set of documents. The tool then uses semantic search to find the most relevant chunks to answer user-provided questions about the documents. The Q&A app uses OpenAI's language models, such as GPT-3.5-turbo, to process the chunks and generate answers to the user's questions.

What is a Data Source?

A Data Source is a tool in Dust's platform that allows users to store, index, and retrieve data in a structured and efficient way. Users can use Data Sources to split documents into smaller chunks to index them separately and overcome the limitations of LLMs' context size. Users can integrate OpenAI's embedding models to generate embeddings for each chunk, allowing them to use semantic search to find the most relevant chunks to answer user-provided questions.

How do I create a Data Source?

To create a Data Source, navigate to your Dust home page, select the Data Sources tab, and click on the "New DataSource" button. Provide a name and description for your Data Source, select OpenAI's embedding model text-embedding-ada-002, and pick a max_chunk_size value that suits your use case and the model you plan to use. Finally, click "Create" to create your Data Source. You can then add and insert documents to your Data Source either through the interface or through the API.

What is semantic search?

Semantic search is a search technique that uses natural language processing to identify content that matches a user's query. In the context of the Document Q&A app, semantic search is used to find the most relevant chunks of documents to answer user-provided questions. It does this by analyzing the meaning and context of words in the query and matching them to similar meanings in the document chunks.