Context & Chaos

The Cats of Context & Chaos

One calm cat, many enterprise-AI disasters. Every cartoon from the newsletter, in one place — free to download and drop into your own posts.

Use them anywhere. A link back to Context & Chaos is appreciated, never required.

Godfather-themed cartoon: a distinguished 'Don of Context' cat in a leather chair turns away an AI-agent cat who presents a data column with no lineage. A nameplate reads 'The Don of Context — Meaning is earned.'

You come to me with a column… and bring me no lineage.

From The Godfather Problem
DownloadLinkedInPost
Two-panel cartoon. Day 1, Deployment Day: a party of cats around a crib labelled 'AI Model' cheer 'It's a model!' with cake, balloons and 'Ship it!' cards. Day 90, Reality: a lone bespectacled cat returns to the same crib carrying stacked books (food, monitoring, retraining, governance, performance reviews) and asks 'Where did everyone go?' Caption: deployment isn't delivery, it's parenting.

Deployment isn't delivery. It's parenting.

From Production AI & The False Finish Line
DownloadLinkedInPost
Cartoon of cats hauling an 'AI Agents' mine-cart up a hill toward 'Satya's Vision' along 'The Context Layer' track. Captions read 'The hill-climbing machine needs fuel. Stop the leaks. Compound the learning.' and 'Human capital + Token capital = Compounding learning.'

The hill-climbing machine needs fuel. Stop the leaks. Compound the learning.

From How to make Satya Nadella's vision a reality
DownloadLinkedInPost
Trade-show cartoon: a crowd of cats at 'Snowflake Summit 2026 — Everyone owns context now' hold booth signs (Semantic Layer, Metadata & Lineage, Agent Governance, Open Standards, Knowledge Graphs, Vector Stores, 'Context solved!'). Two cats point to a wall asking 'Who owns revenue? Which definition is correct? Who updates it? — The question that didn't have a booth.'

Everyone owns context now. The question that didn't have a booth: who owns revenue?

From Snowflake Summit 2026: Everyone Owns Context Now.
DownloadLinkedInPost
Cartoon contrasting Agent #1 'Day One' ('I think I can answer one question') with Agent #10 'Day Many' ('I've seen this before') around a central 'Context Layer — the company brain' machine of five capabilities. Banner reads 'The model didn't get smarter. The context did.'

The model didn't get smarter. The context did.

From What an Enterprise Context Layer Actually Is
DownloadLinkedInPost
Cave-library cartoon: an orange 'Building Agents' cat with a headlamp says 'We need a new theory for agent memory!' to a scholarly cat at a desk stacked with RDF and SHACL notes, who replies 'Welcome. We've been working on it.' Wall carvings list open-world assumption, reification, context graphs, and named graphs.

"We need a new theory for agent memory!" — "Welcome. We've been working on it."

From Four Arguments Ontologists Never Finished (And Why AI Teams Will Have Them Again)
DownloadLinkedInPost
Archaeological-dig cartoon: cats and an AI robot excavate an 'AI Context Layer powered by archaeological findings'. One cat notes 'Sarah left nine months ago' beside a stale record (Active customer = 3+ logins, owner: Sarah, last updated 2024); an orange cat cheers 'Perfect. Deploy the agents.'

"Sarah left nine months ago." "Perfect. Deploy the agents."

From Your AI Context Layer Is Being Built on Stale Metadata
DownloadLinkedInPost
Two-panel cartoon. Left, 'We upgraded the model': a frazzled cat at a control panel where the newest model still returns conflicting revenue numbers — a 'definition conflict', outcomes all wrong. Right, 'We fixed the context': a calm cat whose verified numbers reconcile across finance and sales. Caption: same model, different moat.

"We upgraded the model." "We fixed the context." Same model, different moat.

From If Intelligence Is Abundant, What Is the Moat?
DownloadLinkedInPost
Two-panel cartoon 'Onboarded / Prompted.' A new-hire cat unpacks a 'Welcome to the team' box full of business definitions, org charts and tribal knowledge ('ask Sarah', 'never use the Q3 dashboard'); an AI-agent cat gets a single line — system_prompt 'You are a helpful data analyst.' Caption: same job, different first day.

Onboarded vs. prompted: same job, different first day.

From The Onboarding Gap is Killing your AI Agents
DownloadLinkedInPost
Two-panel cartoon titled 'The bar is blank. It has always been blank.' An anxious orange cat asks 'What if the AI got something wrong?'; a deadpan grey cat replies 'Compared to what?' — beside a data asset whose human-written description was empty all along.

"What if the AI got it wrong?" "Compared to what?"

From Your Team Isn't Afraid of AI. They're Afraid of Being Wrong.
DownloadLinkedInPost
Cartoon: a frazzled grey 'AI Agent' cat at a keyboard runs the query 'What is our customer lifetime value?' while a thought bubble cycles conflicting definitions of 'customer' — 'Nobody wrote down what any of this means, so I'm going with vibes.'

Nobody wrote down what any of this means, so I'm going with vibes.

From Conceptual Modeling Is the Context Engineering Nobody Is Doing
DownloadLinkedInPost
Two-panel cartoon: 'Ontologist. Data Steward. Knowledge Specialist. 2021' — a weary orange cat buried in papers and old servers; 'Context engineer. 2026' — the same job, now a crowned grey cat on a throne labelled 'Context Engineering.'

Ontologist, 2021. Context engineer, 2026. Same cat, new crown.

From Metadata Weekly Is Now Context and Chaos (& Cats!)
DownloadLinkedInPost

A new cat,

Context & Chaos

Every Week

Cat in a boat
[Website env: production]