back to labs
lab experiment LAB-003 workflow automation · 2025 claude code · systems thinking

content ops
dashboard.

What happens when you apply instructional systems thinking to your own workflow, and automate the entire stack with Claude Code. Daily briefs, Canva generation, research pipelines, analytics. None of it required a developer.

see it in action

the walkthrough

A full tour of the automated content operations dashboard, from the morning brief that synthesizes email and calendar, to the Sunday night analytics report that generates itself.

the thesis

this isn't about social media.

I spent several weeks automating my entire content operations stack with Claude Code. Every piece of it (the daily brief, the content generation, the research pipeline, the analytics) maps directly to problems learning designers face every day. None of this required a developer. It required systems thinking, which is exactly what we already do.

the automations

four workflows. zero manual triggers.

Each automation has a direct L&D parallel: the same logic applies whether you're managing content or managing learning operations.

01

the daily brief

Every morning: top emails, calendar events, and Notion projects, synthesized into one prioritized brief. No context switching before coffee.

for L&D

A project status digest. Client deliverables, SME review threads, LMS feedback: one view, every morning.

02

canva automation

A Notion queue feeds Claude Code every Sunday. It opens designated Canva templates, builds every carousel marked ready, and marks them done. Monday morning, the content exists.

for L&D

Job aids. One-pagers. Slide decks. If you have a Canva template and a content queue, this builds your materials while you sleep.

03

research pipeline

Three Apify scrapers run every Wednesday (Reddit, TikTok, YouTube) pulling what the audience is struggling with right now. Results go to Notion with a Perplexity report.

for L&D

Needs analysis. Replace manual community lurking with structured scraping of forums, LinkedIn groups, and industry channels. Real pain points, updated weekly.

04

analytics report

Every Sunday night, a performance report generates itself and surfaces inside the dashboard. No logging into apps. No manual data pulling.

for L&D

Evaluation data. LMS exports, survey results, completion rates, piped into one place, automatically. Kirkpatrick Level 1 and 2 data that actually gets reviewed.

the secret ingredient

One single markdown file contains the entire brand guide: colors, fonts, tone of voice, writing rules. Every Claude Code session reads it automatically. Every output follows it. I wrote it once. It runs everything.

tools used

the stack

claude code cowork notion canva apify gmail api google calendar api netlify react perplexity

see it live

The actual dashboard, running the daily brief, research pipeline, and analytics reports on a real schedule.

open the dashboard

limitations & what's next

This is a working system, not a finished product.

deeper analytics

Current reports surface top-level metrics. Next: trend analysis across weeks, content type performance, and audience growth patterns.

smarter content briefs

Research pipeline delivers raw findings. Next: AI-generated content briefs that connect audience pain points to specific content formats and angles.

end-to-end automation

Some steps still need manual triggers. Goal: fully autonomous pipeline from research insight to published content with human review gates.

template expansion

Currently handles carousel templates. Expanding to video thumbnails, story templates, and long-form article layouts.

next in the lab

programmatic video
& automation.

Code-driven video templates with Remotion and automated Canva carousel generation, from template to published content with no manual design work.

explore the experiment