Descriptions:
Nick Saraev walks through a practical system for taking local agentic workflows and deploying them to the cloud so they can run autonomously — triggered by webhooks, scheduled on a cron, or called by other automation tools — without a persistent server and at near-zero cost. The primary hosting platform he covers is Modal, a serverless infrastructure service that minimizes cold-start latency (critical for agent workloads that time out easily), provides webhook URLs out of the box, and bills only for actual compute used. Saraev reports running several hundred requests over a few days for roughly one cent in total cost.
The tutorial is built around his “directive orchestration execution framework,” a pattern that separates high-level agent instructions — stored as Markdown files in a directives folder — from the underlying Python execution logic. This structure makes agents both consistent and flexible: the model always follows the same high-level workflow while the implementation details remain easy to swap. He demonstrates connecting this framework to N8N and Make.com, showing both directions: agents calling N8N webhooks as tools, and N8N flows triggering Claude Code agents via HTTP requests.
Live demos include an hourly lead scraper that autonomously populates a Google Sheet, a proposal generator triggered on demand, and a Google SERP scraper whose entire N8N flow JSON is fed directly to Claude Code to auto-generate a directive file. Saraev also provides a comprehensive setup document viewers can feed directly to an agent to handle the Modal deployment. For builders who want AI agents running around the clock without managing infrastructure, this video offers a concrete, low-cost path from local prototype to cloud-hosted production.
📺 Source: Nick Saraev · Published December 05, 2025
🏷️ Format: Tutorial Demo
