The walkthrough on this page covers 5 steps: 1. Connect an AI provider 2. Start a chat 3. Create a project 4. Build your own agent 5. Pin the agents you use most.

What are the limitations of Berd?

Pricing and plan details are not stated on the homepage, so cost is unclear before installing. Supported operating systems are not listed, which complicates platform planning. No integrations, API or extension options are described beyond connecting an AI provider.

Berd

Name: Berd
Author: Berd

A playful desktop app for building with a crew of AI agents

Intermediate

Visit Berd Back to AI tools

Screenshot of Berd, A playful desktop app for building with a crew of AI agents

What is Berd?

Berd is a playful desktop app from Block for building with AI agents. It combines chats, projects and model providers in one workspace, with built-in agents like Pushback and Tinker plus custom ones.

Berd is a desktop app for building with AI agents, published by Block and presented with a deliberately weird, playful personality. Its pitch is less chatting and more building: instead of one generic assistant in a single window, the app brings chats, agents and model providers into one workspace that the user controls. Projects can be linked to the folders and repositories already used for day-to-day work, and the canvas can be arranged and pinned to taste. The center of the product is its agent system. An agent is treated as a specialist with a defined role, a set of skills and a distinct personality, and it carries that behavior into every chat it joins. Several come built in. Berdy helps with working inside the app and handles delegated tasks, Pushback acts as a blunt critic for drafts, Choosey helps untangle decisions without making them, Copycat assists with writing while learning the user's style, Wildcard offers unconventional angles, and Tinker works out whether a job needs an agent, a skill or something else, then builds it. Users can also create their own agents and pin or unpin them as needed. Getting started follows a short sequence: connect an AI provider, start a chat, create a project, then build an agent. Because providers are connected by the user, the model behind a task can be switched at any time, which suits workflows where only some jobs justify a large model. Example prompts on the homepage show the range of output, including an animated logo, an interactive presentation and a 404 page that visitors can draw on. Berd sits between general chat clients and heavier agent frameworks. It is aimed at people who want reusable, character-driven helpers tied to real project folders without assembling a toolchain by hand. Pricing, supported operating systems and integrations are not detailed on the homepage, so those points need checking before adoption.

How do you use Berd?

1Connect an AI provider
Open the desktop app and link the AI model provider you want Berd to use. This choice can be changed later when a task needs a different model.
2Start a chat
Begin a conversation to get a feel for the app. Call in an agent such as Berdy when you want help working inside Berd.
3Create a project
Set up a project and link it to a folder or repository you already work in, so your agents operate with the right context.
4Build your own agent
Define a role, skills and personality for a custom agent, or ask Tinker to build one for you.
5Pin the agents you use most
Pin favorite agents to your canvas for quick access and unpin them when a job is finished.

Pros and cons

Pros

Agents carry defined roles, skills and personalities into every chat instead of being a name on a chat windowAI
Chats, agents and providers live in one workspace the user owns, reducing tool switchingAI
Model choice is flexible, so lighter tasks need not use a large, costly modelAI
Projects can be linked to existing local folders and repositoriesAI
Ships with a useful starter cast, including a critic, a decision helper and a style-matching writerAI

Cons

Pricing and plan details are not stated on the homepage, so cost is unclear before installingAI
Supported operating systems are not listed, which complicates platform planningAI
No integrations, API or extension options are described beyond connecting an AI providerAI
The quirky, character-driven tone may feel gimmicky for teams wanting a conventional enterprise toolAI
Requires the user to bring and manage their own AI provider connectionAI

How much does Berd cost?

Integrations

The homepage only mentions connecting an AI provider and linking projects to local folders and repositories. No other integrations are listed.

Features

Customizable canvas with pinning, projects linked to local folders and repos, built-in agents (Berdy, Pushback, Choosey, Copycat, Tinker, Wildcard), custom agent creation with roles and personalities, and switching between AI models and providers.

Frequently asked questions about Berd

How do you use Berd?
The walkthrough on this page covers 5 steps: 1. Connect an AI provider 2. Start a chat 3. Create a project 4. Build your own agent 5. Pin the agents you use most.
What platforms does Berd support?
Berd is available on MacOS.
What does Berd integrate with?
The homepage only mentions connecting an AI provider and linking projects to local folders and repositories. No other integrations are listed.
What are the limitations of Berd?
Pricing and plan details are not stated on the homepage, so cost is unclear before installing. Supported operating systems are not listed, which complicates platform planning. No integrations, API or extension options are described beyond connecting an AI provider.