AI Agents and Skill Editor

Bridge enables you to create intelligent agents that automate routine tasks, handle data operations, and interact with users — directly within your workspace and chats. Your AI agents can access and work with any entity in your workspace – from chat conversations and databases to tasks, documents, and even other AI agents. Chain these capabilities together to create powerful workflows that automate your processes end-to-end. If our skill library isn't enough, we'll help create any custom skill you need.
 

Agent Structure

Each agent is built as a modular system, consisting of:
 

• Prompt — a description of the agent’s behavior and task scope
• Variables — input fields used during execution (e.g., date, amount, user)
• Knowledge — internal documents that the agent can access
• Skills — reusable actions and logic that define what the agent can do
   

Variable types supported:

Text, date, number, money, list, file attachments, arrays.

The more detailed your variable descriptions are, the more effectively the agent can process and apply them.
 

Knowledge empowers the agent to reason and respond based on live documentation.


For example, if you upload internal policies or procedures to an agent, it can:
 

• Answer employee questions based on those documents
• Reference relevant sections
• Guide actions using the uploaded rules or instructions

This turns documentation into a dynamic part of your operational intelligence.
 

Skills (Reusable Agent Actions) 

Skills are the core building blocks of agent functionality. They define what an agent can do — either inside the platform or with external systems.

• Skills can be reused across multiple agents
• Connected in a single click
• No coding required

Visual Skill Editor 

Bridge includes a visual skill editor — a flow-based automation builder.

Each skill is composed of logic blocks connected by data flows.

Examples:

• Database → Create Entry – adds a record to a database
• Database → List Entries – fetches filtered records
• Chats → Send Message – sends a message to a chat or channel
• Agents → Generate Message – creates a text output using AI
• Logic → Iterator / Template Renderer – conditional flows, loops, and templating
   

The skill library is constantly evolving, and soon it will support integration with any external system via API.
 

Combining Scenarios

Skills can be chained to create workflows like:

Fetch data → Save to database → Notify a user in chat

⚠️ Description Matters

Each skill must have a clear name and description — this is how agents understand what the skill does.

Good descriptions help agents choose the right skill during execution.
 

Examples:

✅ “Save IT expenses to database”
✅ “Get all active deals this week”
❌ “Skill 1” or “Test”
 

Agents may use multiple skills dynamically based on their goals — descriptions determine relevance.
 

Use Case Example: “Expense Logger” Skill

• Input: merchant, date, sum
• Action: Save to "IT Services" database
• Notify user with confirmation in chat

All of this can be configured visually — no code required.

What You Can Automate

With skills and agents, you can automate:

• Request and ticket submission
• Database queries and filtering
• Auto-replies in chat
• Document/report/post generation
• Integration with external tools and systems via API
   

Version Management

Each agent supports versioning:

• Work in draft mode
• Publish stable versions
• Switch between versions instantly

Integrations

Agents can be connected to:
 

• Bridge Workspace — for internal use
• Telegram — via bot token (run agents in familiar messaging apps)

AI Model Options

Choose between models based on performance needs:
 

• GPT-4o — high-quality, multi-step reasoning
• GPT-4o Mini — lightweight, fast-response model
   

Model choice affects token consumption and performance.