👤 For pipe admins
🔐 Available on all plans
🎯 For those who want to harness the power of artificial intelligence to enhance their processes with automations.
What is Max Effort?
Max effort is a behavior-level capability opt-in that elevates the assertiveness and reasoning precision of AI Agents when dealing with complex business workflows.
Instead of operating in a standard single-shot execution mode (processing everything in one pass), an agent with Max effort enabled follows an advanced cognitive loop: it maps out a plan, searches for specific pages and/or pieces of context, reviews its own understanding, and executes subsequent targeted searches as many times as needed before performing a final action.
Ideal Use Cases
This feature is highly recommended for Business Analysts and Compliance Analysts in Finance and Insurance business units whose workflows involve:
Long Document Auditing: Reviewing complex files like financial invoices, contracts, and comprehensive reports exceeding 30 pages.
Cross-Document Reconciliation: Comparing and cross-referencing values or clauses across multiple document attachments.
Multi-Factor Conditional Rules: Executing logic dictated by interconnected criteria, such as complex calculations, custom discounts, or strict compliance policies.
Before you start
Permissions: You must be a Pipe Admin to configure AI Agent behaviors.
Dependencies: Enabling Max effort automatically triggers semantic indexing and web search capabilities.
Model Compatibility: Ensure the behavior is configured to use Pipefy's native LLM structure.
How to activate Max effort
Max effort is applied directly to specific behaviors within an AI Agent. Behaviors without this flag remain completely unaffected, allowing you to isolate advanced reasoning only where it is truly needed.
Go to the specific pipe containing your automation and click AI Agents.
Select or create a New AI Agent on the AI Agents tab.
Inside the AI Agent configuration screen, navigate to the Behaviors tab on the left menu.
Locate the specific behavior card you want to enhance (e.g., Expense Report Evaluator).
Toggle the Effort: Max switch to the On position, or select Max from the Effort dropdown menu at the bottom of the instruction box.
Caption: Activating Max effort within an AI Agent behavior.
⚠️ Attention: Enabling Max effort changes the execution mechanics of the sub-agent. Review the limits and billing logic below before deploying to production environments.
What is included in this Beta version
Iterative Document Reasoning: An intelligent loop that performs semantic searches and targeted page inspections, capable of interpreting structural visual contexts like charts, signatures, and embedded scanned tables.
Granular Behavior Isolation: Ability to turn the capability on or off per individual behavior without altering other tasks in the same agent.
Proven Real-World Validation: Core logic rigorously tested and validated using high-complexity Finance and Insurance scenarios.
Known Limitations Beta Pricing Structure
Runs utilizing Max effort are billed using a multi-step execution model:
Because the agent dynamically iterates—searching, reviewing, and searching again—each cycle counts as a single reasoning step. Complex document processing will naturally accumulate more reasoning steps than a standard single-shot automation run.
ℹ️ Tip: The OCR extraction component is completely excluded from the pricing structure during this Beta timeframe.
Validation & Safe Rollback Checklist
[ ] Verify Inputs: Ensure your input document attachments contain high-quality digital text layers (avoid heavy reliance on unmapped scans due to the temporary OCR limitation).
[ ] Test Output Fields: Check that mapped output fields successfully receive the structured answers after a complex multi-step analysis run.
[ ] Safe Rollback: If a workflow requires immediate BYOM integration or encounters timeout issues, open the behavior settings and switch the Effort toggle back to Standard. This immediately downgrades the behavior to a standard single-shot execution with zero impact on the rest of your automation setup.
Troubleshooting
The AI Agent fails to extract text from a specific document attached to the card
Symptom: The agent returns blank results or notes that it cannot find the text inside an attached PDF.
Fix: Check if the document is a flattened, scanned image. The Beta version does not include native OCR processing. Re-upload the document with an active digital text layer or extract text beforehand.
Behavior run halts or drops completely during complex document reviews
Symptom: The execution stops unexpectedly without outputting errors, particularly on extremely long workflows.
Fix: The run may have exceeded the 10-minute threshold and hit the current lack of checkpoint resilience. Try splitting the target document into smaller, modular files or streamlining the instruction parameters.
Max effort option is grayed out or cannot be activated on the behavior
Symptom: You cannot toggle the "Effort: Max" option within the behavior settings window.
Fix: Check your LLM settings. If the behavior is bound to a custom model key (BYOM), Max effort will remain unavailable. Revert the provider back to Pipefy's native LLM structure to unlock the option.
What comes next (Public Roadmap)
Complete Run Resilience: Checkpoint and recovery mechanics designed for long-running executions .
Advanced Text Extraction: Alternative solutions and replacements for OCR dependencies (Under investigation).
Enhanced Audit Logs: Full end-user visibility into internal sub-agent execution paths and reasoning steps.