Back to archivecertified generic

Case file

kilo.ai

ShitScore 63 / 100AI AgentsCaptured 2026-05-17Submitted by communityVisit crime scene ↗

An AI coding tool whose hero says "Great workflows don't [change]" and whose footer contains eleven named SEO comparison pages, plus a link to "KiloClaw Alternatives" — a page helping you find alternatives to one of Kilo's own products.

Kilo is an AI coding agent built on a strong premise: models change, workflows don't. The footer lists eleven named comparison pages against competitors. The homepage features a blog post titled "The Subsidy Was Never About You," explaining Kilo's own price increase by arguing the original low price was always Anthropic's subsidy, never the user's benefit. The agent's secondary product is described as "where your agent lives after hours" — after hours being when the office is closed and no one is accountable.

Share this roast

PostShare

Exhibit A — Evidence

Captured 2026-05-17

Hero viewport of kilo.ai on a near-black background. Top nav: Kilo logo, Product, Models, Pricing, Support, Docs, Blog, a GitHub star count (59.4k), Sign In, and a yellow "Sign up" button. Large headline: "{ models } change. Great workflows don't." with "models" in yellow curly braces. Subheadline: "Build, ship, and iterate with any model, everywhere you work." Four IDE/platform tabs: VS Code, JetBrains, CLI, Cloud. A terminal-style input bar reads "vscode-extension/kilocode kilo-code." Two yellow CTAs: "Code for Free" and "Host an Agent." Small link: "Using a VS Code-like editor? Open VSX." Below: a large dark IDE screenshot showing Kilo Code in action with a chat panel and code editor side by side.
Screenshot — kilo.ai (1940×1080)

Score breakdown

Prompt residue6/10
Feature grid density9/10
Meaningless value prop5/10
Trust signal suspicion8/10
Founder face AI probability3/10
Product proof absence6/10
ShipFast resemblance7/10
Hero claim
"{ models } change. Great workflows don't." Subheadline: "Build, ship, and iterate with any model, everywhere you work." CTAs: "Code for Free" / "Host an Agent."
Proof problem
"Trusted by developers at the world's most innovative companies" — Meta, Amazon, Airbnb, PayPal, Square, Red Hat (individual installs, not enterprise contracts). Stats: "#1 Open Source Product of the Month" (month unspecified), "3M+ Kilo Coders" (unverified, no source), "30T+ tokens processed" (unverified). Blog post "The Subsidy Was Never About You" featured on homepage — explains own price increase as Anthropic subsidy revelation. "KiloClaw Alternatives" link in footer (self-directed competitor research for own product). 11 named competitor comparison pages in footer COMPARE column.
Visual pattern
Near-black background, yellow accent (#FFD700-range). Standard dark SaaS nav with GitHub star count. Hero: centered headline with code-syntax curly braces + subheadline + 4 IDE tabs + terminal-style input + two CTAs. IDE product screenshot. "Trusted by developers at" logo strip (Meta, Amazon, Airbnb, PayPal, Square, Red Hat). Three stat tiles (#1, 3M+, 30T+). "Use Kilo Everywhere" icon row (12 integrations). "Multiple modes, one agent" alternating feature section. "Meet KiloClaw" hosted agent section. "Code is our main act. KiloClaw is where your agent lives after hours." Two product cards (Kilo Code, KiloClaw). "Recent posts" 2×3 blog grid. FAQ accordion (14 questions). Yellow "Ready to get started?" CTA. Massive footer with 6 column groups including two COMPARE columns.
Why it still might convert
Kilo Code is genuinely open source (Apache-2.0) with 59k+ GitHub stars — a real product with real usage. The "any model" positioning is accurate and useful for developers who want to swap between Claude, GPT-4, DeepSeek, or local models without changing their workflow. The multiple agent modes (Architect, Code, Debug) are a real differentiator over single-mode tools. KiloClaw as a hosted OpenClaw alternative with one-click deploy addresses a real pain point for developers who want persistent agent operation without self-hosting YAML. The 30T+ tokens processed, while unverified, suggests scale.

Editorial roast

By Editorial Desk · Filed against kilo.ai

¶ 01

The hero says: "{ models } change. Great workflows don't." The footer's COMPARE KILO CODE column lists: vs Roo Code, vs Cursor, vs Windsurf, vs GitHub Copilot, vs Claude Code, vs Cline, vs Tabnine, vs Replit, vs Lovable, vs CodeRabbit, vs Augment Code. Eleven named competitors, eleven SEO comparison pages. The workflow that does not change is publishing a comparison page against every entrant in the market by name. The footer also contains a "KiloClaw Alternatives" link — a page on the Kilo website dedicated to helping you find alternatives to one of Kilo's own products.

¶ 02

"The Subsidy Was Never About You." That is the title of a blog post featured in the "Recent posts" section on the homepage. The subtitle: "Kilo Code's pricing change [redacted] reveals what the Claude subscription was always for." The product is explaining its own price increase as an exposé — the original low price was a subsidy, the subsidy was Anthropic's, and it was never for you. This is the customer communication the company chose to place on its landing page, between the stats row and the FAQ.

An AI coding tool whose hero says "Great workflows don't [change]" and whose footer contains eleven named SEO comparison pages, plus a link to "KiloClaw Alternatives" — a page helping you find alternatives to one of Kilo's own products.

¶ 03

"Trusted by developers at the world's most innovative companies" — Meta, Amazon, Airbnb, PayPal, Square, Red Hat. The word "at" carries the full weight of this sentence. It means one developer at each of those organisations has installed the VS Code extension. Any tool available in the VS Code marketplace is one search away from being used by someone at Meta. "Trusted by developers at" is the weasel construction that converts individual installs into enterprise logos.

¶ 04

"Code is our main act. KiloClaw is where your agent lives after hours." After hours is when the office closes, the lights go off, and no one is present or accountable. The product is selling autonomous AI agent operation in the exact language used to describe what a building does when everyone goes home. The selling point is unsupervised operation. The phrasing chose the metaphor of abandonment.

— 30 —

More from the archive

Similar offenders on file.

fluenta.spaceAI AgentsSS 74

A startup idea validator that labels its own output "Not AI slop" in step one and reveals a six-LLM council as the engine in step two — whose hero example card shows an idea with LUKEWARM demand, CHRONIC burn rate, and CROWDED OUT competition.

Open file →
elevenlabs.ioAI AgentsSS 62

The company that clones human voices from 30 seconds of audio leads with "Bringing technology to life" — a headline any router manufacturer could use — then describes its research as redefining "human technology interaction," a phrase already old when Siri launched.

Open file →
customgpt.aiAI AgentsSS 78

The AI platform trusted by 10,000+ organizations worldwide displays one logo — MIT — then quotes a Product Lead from MIT whose testimonial ends with "AI enabled the future of AI," and touts being "Awarded Top 7" by Inc Inspire.

Open file →

Correction channel

Wrong, unfair, or outdated?

Ask for a correction or update. Satire is more effective when the facts are not lazy.