Welcome to the AgentHub community 👋
If you're interested in building powerful automations without writing a single line of code, you've come to the right place. AgentHub offers an intuitive and collaborative platform to create AI-powered workflow pipelines called automations using modular components known as nodes.
Drag, drop, and link nodes onto a canvas to create powerful automations.
Agent Building Create custom chatbots with access to your automations. These agents are like personal assistants that are infinitely customizeable to your exact workflows!
Share and build automations together with your team members under a project.
Integrations: Connect to popular services like Twitter, AWS, GitHub, Outlook, Google, and more.
Nodes: The modular building blocks that you drag onto the canvas. Individual no-code Lego pieces.
Automations: A series of nodes connected to achieve some goal. No-code lego creations
Agents: An intelligent chatbot with access to your automations as tools. Like an assistant who is able to use your lego tools dynamically.
Nodes are the modular components make up an automation. Each one is a step in a series of steps towards a goal.
You can drag-and-drop them on to the canvas from the node menu that can be shown/hidden through the >>/<< icons.
Nodes generally take in inputs, have some outputs, and can also have parameters. Many of these attributes may also be optional. Let's take a look at an example:
The Website Scraper node has a
url as a parameter, and the File Reader node has the name of an
uploaded_file as a parameter. They are both filled out respectively.
If you look closely, you can also see that there is an option to have the
uploaded_file_name as an input instead of a parameter, in which case the parameter box would become greyed out, and the input value would take precedence.
The CombineStrings node is receiving two inputs, and the format to combine these two inputs is a parameter here. It then outputs that
combined_string, which can be used as an input to another node.
Automations are the heart of AgentHub, allowing you to connect nodes to execute specific tasks.
When a automation is run, data passes from one node to the next, with each node manipulating or generating content along the way.
Here is an example of a automation that I use to summarize my unread emails every morning:
You can simply run the automation by hitting the run button in the top right corner:
After hitting run, a log feed will pop up describing the current stage of the automation in real-time as actions are happening.
Each node's logs available in these collapsable components.
Automations are only as good as they are useful. Agents make automations easier to use by providing a simple interface between a human and complex automations.
Agents are persistant entities with access to a list of automations, like a human with tools in their toolkit.
An agent can chat normally, but when prompted in a way that suggests it should use a tool, it picks from the available automations. If the automation requires inputs, it will infer what to fill the input field with.
The automation building page is a large canvas for you to design your personalized automation. Each node can be found in the library on the left side.
Nodes are dragged and dropped onto the canvas one at a time. Once they're on the canvas they're ready to be connected together.
The inputs of one node are derived from the outputs of another.
Most automations we've built or user build follow a general pattern -> Ingest data, prepare it for the AI, process it with AI and output a result.
- Data Loaders : Pipelines normally start off with the ingestion of some sort of data. The data being ingested can be thousands of links in a csv, an uploaded file, a scraped website etc. This data is the content
- Data Modifiers: These nodes are the backbone of our automations. They are the helper functions that make everything possible. They allow us to perform any number of data manipulation tasks like reformatting strings, reading values from jsons, extracting columns from csv etc.
- Using AI: Nodes that use AI generally follow. These nodes are the source of reasoning in the automation and can ingest the loaded data and summarize, answer questions, categorize etc.
- Data Writers: These are almost always the very last step of our automations. They touch the outside world and turn the AI processed output into a tangible, valuable result. Nodes that generate files, raise GitHub PRs, post Tweets and more live in this category.
The #1 way to get started is by using a prebuilt template!
We've built 50+ functional automations with annotations for you to browse. You can use these to understand how things work or to run out of the box, each is built to be immediately useful
Get started with templates here: https://www.agenthub.dev/templates