ID Bot Automation System — AI Pipeline for Instructional Design

Independent Project | Claude AI · n8n · Google Docs API · Tally · Netlify | March 2026

What This Project Is

The ID Bot Automation System is a six-stage AI pipeline that takes a course brief from a client and produces a complete, design-ready instructional package — automatically. Each bot handles one stage of the instructional design process. The output of every stage feeds the next.

The vision: a subscription SaaS product for instructional designers and L&D teams — where a client submits a single intake form and receives a fully structured, production-ready course package in minutes instead of weeks

The ID Thinking Behind the System

This is not a content generator. It is an instructional design system — built on the principle that AI should do the silent analytical work that takes an ID hours, while the human designer retains judgment and approval at every stage.

Every design decision in the pipeline is derived from instructional theory:

  • Bloom's Taxonomy — silently applied to determine cognitive level per objective based on course category

  • Backward Design — assessments designed before activities; every activity prepares learners for the specific assessment already approved

  • ADDIE alignment — Discovery → Outline → Assessments → Activities → Script → Storyboard maps directly to Analysis → Design → Development

  • Kirkpatrick Levels — silently mapped per category — Level 1+2 for compliance, Level 3+4 for performance gap courses

  • ABCD Method — applied for compliance and knowledge transfer courses

How It Was Built

Phase 1 — System Prompt Design in Claude Projects

All six bot system prompts were fully designed, written, and tested inside Claude Projects — Anthropic's AI environment for building persistent, instruction-following AI agents. Each system prompt was refined across multiple real-world case studies — including ConnectNow Telecom, Nova Payments, SafeWork Industries, Meridian Consulting, Apex Financial, and BrightPath HR — before any automation was built.

This phase required encoding complete instructional design theory into AI behaviour — every rule, every edge case, every output format specified precisely so the bot produces consistent, pedagogically sound output without human intervention at each step.

Phase 2 — n8n Automation Build

Once system prompts were validated, automation was built in n8n — connecting the AI outputs to real-world delivery infrastructure.

Bot 1 — Discovery (Fully Operational in n8n)

Bot 1 is the only fully built, tested, and operational automation in the pipeline. It receives structured data from a Tally intake form — no questions asked in the chat interface. It silently applies instructional theory and generates a three-section Discovery Summary:

  • Section 1 — Needs Analysis: course topic, category, problem statement, goals and gaps, instructional theory, learner profile, course parameters

  • Section 2 — Learning Objectives: Bloom's or ABCD method, two versions per objective — designer reference and learner-facing

  • Section 3 — Course Blueprint: complete summary table of all design decisions

The output is automatically formatted as a Google Doc via batchUpdate API, shared with the client, and delivered via HTML email — with a link to a Netlify-hosted review form where the client approves or requests revisions. A revision handler workflow reads the original prompt from Google Sheets and regenerates the document if needed.

Bots 2–6 — Prompt Complete, n8n Builds In Progress

All six system prompts are complete and tested. The n8n automation builds for Bots 2–6 are in progress — estimated 9 hours of build time remaining.

Each bot has been designed with specific instructional rules:

  • Bot 2 (Outline) — asks one question only. All other decisions made silently from Discovery output

  • Bot 3 (Assessments) — follows Backward Design; maximum two questions; 9 fixes applied

  • Bot 4 (Activities) — zero questions; purely generative; one activity per lesson

  • Bot 5 (Script) — complete four-column format; Context Compression Logic prevents tone drift across long courses

  • Bot 6 (Storyboard) — zero questions; purely transformative; every slide includes layout, narration, visuals, interactions, and accessibility notes

Current Status

✅ Bot 1 — Discovery — Fully operational in n8n — end-to-end tested
✅ All 6 system prompts — Complete and tested in Claude Projects across 6 case studies
⏳ Bot 2 — Outline — n8n build in progress
⏳ Bot 3 — Assessments — n8n build in progress
⏳ Bot 4 — Activities — n8n build in progress
⏳ Bot 5 — Script — n8n build in progress
⏳ Bot 6 — Storyboard — n8n build in progress

What This Project Demonstrates

Most instructional designers use AI as a writing assistant. This project treats AI as an instructional system — with theory, logic, rules, and quality checks built in at every stage.

Building this required simultaneously applying deep instructional design knowledge to encode ID theory into AI behaviour, technical architecture thinking to design a six-bot pipeline with shared state management, and systems thinking to anticipate edge cases, build revision handlers, and ensure output quality across diverse course types.

The pipeline is in active development. Bot 1 is live and operational. The full system will be available as a subscription product for instructional designers and L&D teams.