Optimization

​

Why Optimization Matters

AI agents face key challenges when using tools:

1. Token Limits: Large API responses consume context windows fast
2. Latency: Slow API calls frustrate users
3. Tool Selection: Too many tools overwhelm agents

Tadata’s connectors come with built-in optimizations that solve these problems automatically.

​

Optimization Features

​

1. Response Compression

Achieves 60%+ token savings API responses typically include hundreds of fields that AI agents don’t need. Tadata compresses responses to only essential data. Result: More context window space for your agent’s conversation and reasoning.

​

2. Response Projection

Returns only relevant fields for each operation Instead of returning entire objects with nested arrays and unnecessary metadata, Tadata projects responses to include only what matters for the task. Benefits:

• Reduces token count significantly
• Keeps agents focused on relevant information
• Cleaner data that’s easier to understand and process

Result: Agents work with exactly the data they need, nothing more.

​

3. Response Filtering

Optimizes tools that return complex query results Especially relevant for database queries and data-intensive operations, response filtering removes unnecessary data to improve performance. Benefits:

• Faster API calls
• Better user experience
• Lower token consumption
• Critical for complex database queries

Result: Agents respond faster, users wait less.

​

4. Tool Batching

Reduces runtime drastically by parallelizing operations AI agents can’t parallelize tool calls on their own. Tool batching executes related operations in parallel, dramatically improving performance. Benefits:

• Parallel execution of related operations
• Drastically reduced runtime
• Fewer sequential bottlenecks

Result: Complex workflows complete much faster.

​

5. Tool Workflows

Pre-defined workflows reduce runtime by up to 80% While agents can figure out what to do at runtime, many workflows are common and predictable. Defining these workflows upfront eliminates the planning overhead and executes them optimally. Example Workflows:

• Triage: Fetch error → Search code → Create issue → Notify team
• Deploy: Merge PR → Deploy → Monitor → Rollback if errors
• Support: Ticket received → Search docs → Auto-respond → Escalate if needed

Benefits:

• Up to 80% runtime reduction
• No planning overhead - workflows execute immediately
• Consistent execution every time
• Graceful error handling

Result: Common workflows complete in seconds instead of requiring agent reasoning at each step.

​

6. Suggested Tool Names & Descriptions

AI-optimized tool naming for better selection Generic API endpoint names like POST /issues don’t help AI agents understand what a tool does. Tadata optimizes tool names and descriptions to be intuitive. Before:

• Tool: POST /api/v1/issues
• Description: “Creates a new issue resource”

After:

• Tool: create_linear_issue
• Description: “Create a new issue in Linear with title, description, priority, and assignee”

Benefits:

• Agents choose the right tool more often
• Reduced confusion and errors
• Better alignment with natural language requests

Result: AI agents understand and use your tools correctly.

​

7. Automated Tool Selection

Focus agents on relevant operations only When you have toolsets with many connectors, automated tool selection suggests only the tools relevant for specific use cases. How It Works:

• You enable all tools in a connector (maximum flexibility)
• Configure use-case-based tool selection in server settings
• Agents only see tools relevant to their task

Example: For a “production triage” agent:

• ✅ Show: Error monitoring, code search, issue creation
• ❌ Hide: Marketing tools, CRM operations, analytics

Benefits:

• Agents aren’t overwhelmed by irrelevant options
• Better tool selection accuracy
• Faster decision-making

Result: Your agents focus on what matters for their specific role.

​

How Tadata Handles Optimizations

​

Automatic (All Pre-Built Connectors)

These optimizations are always enabled for Tadata’s pre-built connectors: ✅ Response Compression ✅ Response Projection ✅ Response Filtering No configuration needed. Just connect and go.

​

Optional (Configure for Custom Toolsets)

These features are available when bringing your own APIs or creating custom toolsets: ⚙️ Tool Batching ⚙️ Tool Workflows ⚙️ Suggested Tool Names & Descriptions ⚙️ Automated Tool Selection Tadata lets you add these optimizations to your custom connectors.

​

Best Practices

​

Next Steps