UiPath Maestro Flow: A No-Code AI Orchestration Tool

About this video

Presenter Garrick, in partnership with UiPath, builds a complete no-code orchestration in UiPath Maestro Flow from scratch. Flow is a canvas inside UiPath where steps called nodes are connected in order, and it is also available in Studio Desktop, VS Code, and coding-agent-friendly forks like Cursor, Windsurf, and Kiro. The example flow has four steps for a hotel invoice: a manual trigger accepts a document input argument set to file type, an autonomous AI agent reads it, a human checks the result, and the clean data is saved. The agent is given the Analyze Files tool, pointed at the trigger’s document variable, and told to extract the vendor, invoice date, and total amount due. Its system prompt is generated from a one-sentence description using the sparkle button, its user prompt is replaced with ‘Process the attached invoice and return the requested fields,’ and outputs are set to vendor (string), date (string), and total (number). Human review uses a Quick Form Escalation with Approve (continue) and Reject (end) outcomes, assigned by email; crucially, the agent’s system prompt must be edited to explicitly require sending results for review before finishing. The final node is a Google Sheets Write Row action writing to a pre-existing sheet called ‘invoice reader test’ in append mode. The flow is tested twice in debug mode with two different invoices, with approvals handled via My Actions / Action Center (assign to self, then approve), and both runs add a correct row to the sheet. Deploying publishes it as a live process, but a manual trigger must be started each time; swapping in another trigger could let it pick up invoices from an email inbox or shared folder. For quicker, minutes-long processes, he notes Maestro Automate uses the same canvas tuned for short, fast agent and API workflows.

Key takeaways

• Maestro Flow builds orchestrations by connecting nodes on a canvas — trigger, agent, human check, save — with no code required.
• Give the autonomous agent the Analyze Files tool and structured output variables (vendor and date as strings, total as a number) so extracted data can flow downstream.
• The AI can generate a detailed system prompt from a single-sentence description, which you should then read and edit.
• An escalation is optional unless instructed: add an explicit, unconditional line to the system prompt requiring human review before the agent finishes.
• Write Row only writes into a sheet that already exists and should be set to append; a manual trigger does not run automatically, so swap the trigger for automatic pickup from an inbox or folder.

Episode details: published October 15, 2026 · 18:48 long. Summary written by Self Growth Videos from the episode itself. Business results and any income figures mentioned are Kevin Stratvert’s own claims, shared for education — they are not typical or guaranteed.

← Previous: How to Transcribe Audio to Text with ChatGPT · Next: Surface Laptop Ultra: Who Is This Actually For? →

All Kevin Stratvert videos · Kevin Stratvert profile · Kevin Stratvert on YouTube