Training OS — Notion AI Post-Session Workflow

A scenario-based eLearning project that turns post-session chaos into a repeatable, audit-ready system — built entirely in Notion.

View Live Project →

Project Overview

Role	Instructional Designer (solo — analysis through evaluation)
Type	Scenario-based eLearning (35 slides, ~30–45 min)
Audience	Freelance and in-house trainers running recurring 1:1 or group sessions
Tools	Notion (databases, AI Meeting Notes, AI chat, button automations), DALL·E, Mermaid, HTML/CSS/JS
Deliverables	Interactive HTML presentation, downloadable Notion template (3 databases + 2 prompts), visual assets
Live Project	josephgendron.com/id/traineros/project.html

The Problem

Trainers finish sessions mentally fatigued — and that's exactly when they're expected to write follow-up notes, assess performance, assign exercises, and send everything to the learner, themselves, and the training center.

The result? Inconsistent documentation, missed follow-ups, and audit gaps.

I wanted to solve a problem I face daily: how do you produce consistent, high-quality post-session output when you're cognitively spent?

The Solution

I designed a scenario-based training that teaches trainers to run a complete post-session workflow inside Notion — from recording to sending — in under 5 minutes per session.

The learner follows Sophie, a freelance English trainer juggling 15 sessions a week, as she moves from scattered notes and audit anxiety to a structured, repeatable system. The training uses Dan Harmon's Story Circle to create emotional stakes and a realistic adoption arc — including the hardest moment: trusting a new system on Monday morning with a full calendar.

The workflow produces two artifacts per session:

Learner Profile (created once, after the first session) — structured from intake interview data
Post-Session Message (Lesson, Assessment, Exercises) — generated every session, sent to three recipients with one click

The Process

Action Map

I started with Cathy Moore's action mapping to ensure every slide ties to a measurable business goal — not just content coverage.

Business Goal: Improve output quality by 30% while reducing documentation time by 50%

Set up 3 databases (Prompts, Learners, Sessions)

Record sessions with AI Meeting Notes

Run Learner Profile prompt (first session)

Run Post-Session Notes prompt (every session)

Review AI output and edit if needed

Send to 3 recipients with one click

Storyboard — Narrative Structure

The storyline follows the Story Circle (Dan Harmon), mapping each beat to a teaching moment:

1YouSophie's "good enough" routine

2NeedAudit coming — notes are scattered

3GoBuild 3 databases + open template

4SearchFirst session with Marc — record, prompt, generate

5FindWorkflow works — both artifacts produced

6TakeMonday morning — 15 sessions, trust the system

7ReturnClick Send — 3 recipients, done

8ChangeRepeatable, audit-ready, consistent

Each beat maps to specific slides with varied interactions: reflection prompts, knowledge checks, a matching exercise, and an error-detection practice activity — keeping engagement high without breaking the narrative.

Visual Design

I chose a warm, Rockwell-style illustration direction — human, narrative-first, updated for contemporary remote-teaching settings. The goal: reduce cognitive load while supporting the scenario tone.

Style decisions:

Warm off-white/cream background with charcoal text (not pure black)
Inter font stack — clean and highly legible at all sizes
Burnt Orange accent used sparingly for key moments and interactive elements
Card-based layout with generous whitespace — mobile-first, responsive to widescreen
Accessibility: WCAG AA contrast on all text, 44×44px touch targets, reduced-motion support

Custom Assets

This project isn't built on a template — it's an original system I designed from my own trainer experience:

Notion workflow architecture: 3-database end state (Prompts, Learners, Sessions) with relations, AI Meeting Notes integration, prompt-driven generation, and button automation
AI prompt engineering: two production prompts (Learner Profile Generator + Post-Session Notes) designed to read page context and produce structured output
Prompted illustrations: AI-generated visuals (DALL·E) created to match the defined visual direction
Mermaid diagrams: custom action map and story circle visuals, built inside Notion
Downloadable Notion template: a ready-made workspace with all 3 databases, properties, relations, and both prompts pre-configured

Full Development

The training was built as a single-file HTML deliverable — responsive, self-contained, no external dependencies. I generated a first draft with AI, then manually reviewed and edited the HTML for accuracy, structure, and polish.

The 35-slide storyline covers the full ADDIE cycle: analysis (real pain point, action-mapped objectives), design (Story Circle narrative, wireframes), development (databases, prompts, visual assets, HTML), implementation (live deployment, template), and evaluation (Kirkpatrick L1–L4).

Results & Takeaways

Evaluation Plan

I designed the evaluation using Kirkpatrick's four levels. Levels 1–2 draw on real-world usage evidence; Levels 3–4 are proposed with conditional framing.

Level	Measures	Evidence / Approach
L1 — Reaction	Learner satisfaction	I built and used this workflow myself. Learners consistently appreciate receiving the structured post-session emails — the format is clear, actionable, and sets expectations for the next session.
L2 — Learning	Knowledge & skill acquisition	The training includes 5 self-check questions, 3 knowledge checks, a matching exercise, and an error-detection activity — all embedded in the storyline. Real usage confirms the workflow is learnable in a single walkthrough.
L3 — Behavior	On-the-job application	I would track adoption over 4 weeks: are trainers running the full workflow for every session? A drop-off after week 1 would signal a trust or habit barrier.
L4 — Results	Business impact	I would measure two KPIs: (1) output consistency rated by training center reviewers; (2) documentation time — target 50% reduction per session.

Key Takeaways

Design Under Cognitive Load

The core constraint wasn't content complexity — it was designing for users who are mentally fatigued. Every UX and instructional decision prioritized low friction.

Full ADDIE Cycle, Solo

I owned every phase — from needs analysis and action mapping through narrative design, asset creation, development, and evaluation planning.

AI as a Production Tool

I used AI for prompt engineering, illustration generation, and HTML first-draft generation — then reviewed and edited every output manually.

Systems Thinking

The real deliverable isn't a slide deck — it's a replicable workflow architecture (3 databases, 2 prompts, 1 button) that any trainer can duplicate and use immediately.

Impact & Value

This project demonstrates the ability to identify a real workflow problem, design a complete instructional solution, and build production-ready assets — all as a solo designer. The training is live, the template is downloadable, and the workflow is in active use.

For a team, this translates to: I can take a messy operational problem, structure it into a learnable system, and ship something people actually use — from analysis to deployment, with clear evaluation criteria.

Future Considerations

🌍

Localization — The workflow is currently in English; a French version would serve the primary learner population directly.
📦

LMS Integration — The HTML deliverable could be wrapped in SCORM/xAPI for tracking inside an LMS.
⚡

Advanced Automation — Notion's API could replace the manual Send button with a fully automated trigger.
📋

Expanded Prompt Library — Additional prompts for progress reports, level assessments, and end-of-program summaries.

Joseph Gendron, thinking about the future