⭐️ Join us on Slack or visit the Forum to connect with the team

Supercog is a no-code platform for building and running AI agents to make you and your team more efficient at work.

Just a few of the things you can do with the platform:

• Send your AI bot to take notes at a meeting, storing and remembering a summary afterwards
• Have your AI assistant prepare call notes before you meet with someone
• Load up docs and other knowledge into long term memory - your AI assistant never forgets
• Connect your AI assistants to Salesforce, Github, JIRA and more systems to get data in real-time
• Access any agent right from Slack

(Read What can I do with Supercog? for background info on what Supercog can help with.)

An agent is composed of these basic parts:

1. The LLM model powering the agent. You can choose amongst many different LLMs, including GPT3/4/4o/4o-mini, Claude, and LLama3.x. See our models page for descriptions of available models.

(You can use GPT-4o-mini with the free tier. To avoid bankrupting our startup you will need to provide our own API Key to use other models. Visit the Settings page to set your API key using the appropriate Env Var.

2. The Agent Instructions describe the work that you would like the agent to perform. If you just want to experiment it is not necessary to write any instructions. But when you want to devote your agent to a specific task, then you will want to explain that task in the instructions.

3. The enabled tools. You enable agents to perform work by giving them tools. Without any tools an agent is limited to the capabilities of the underlying LLM.

The built-in My Supercog agent is available any time as your personal assistant (both in Slack and on the web). It can connect to your calendar, email, and docs, and help you find information quickly and stay on track. Send your assistant to meetings to record the minutes and then keep those notes in its long term memory. I use mine to:

• Keep my meeting notes
• Answer questions from all the docs I have spread around
• Answer questions from any Slack channels that I've joined
• Pull information from my Email
• Scan or create new issues in Github

My Supercog can enable the use of any tool available in the system, just ask it do so:

• Read text from images, or generate new images
• Convert text to audio or vice versa
• Search for information on the web (including LinkedIn)
• Read and write all sorts of file types

1. From the Home page, click Create Your Own Agent, then Create Agent.
2. Name your agent something like "Scott's Weather Agent", and click Save.
3. Now in the chat box (at lower right), ask a question: "What is the weather like today in Berkeley, CA?" and press <enter>.

4. Your agent will answer like this: I don't have real-time access to current weather data.
5. Now click the button. Open the Live Info section, find the Weather Tool and click Add->Add Tool.

6. In the chat box, press the up-arrow key to retrieve your question, and press <enter> to ask it again.
7. Your agent should respond with an accurate weather report.

There are lots of tools which any agent can "just use". But agents get really powerful when they have authenticated access to your real work systems.

To use a tool like the Salesforce tool to connect to Salesforce, go to the Connections page:

On this page you can create a secure connection to a variety of systems. Select the system type in the drop-down menu and click Add. In the popup box name your connection, and then add your system key, or click Connect if the system uses Oauth for authentication.

Click the Save button to save your new connection. By default all Connections are private. But if you toggle Shared Connection then anyone else on your team can re-use that Connection in their agent.

Once you have created a connection, then it will appear in the drop-down next to the tool in the Add Tool popup window.

Now that you understand the Supercog basics, you can learn more if you wish:

• Learn more about Tools & Connections
• See tips on how to build agents effectively
• Learn about Triggers for activating your agents automatically.
• Support for different LLM models
• Examples of more complex agents
• Understanding token usage