Services New Software Skill — AI skill for Claude Code
A Claude Code skill that applies Sequoia's "Services: The New Software" framework to evaluate AI product ideas.
How to install Services New Software Skill
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open takeshiyu/services-new-software-skill and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Services New Software Skill does
A Claude Code skill that applies Sequoia's "Services: The New Software" framework to evaluate AI product ideas. Runs 5 checks (Tool vs Outcome, Intelligence vs Judgement, Outsourcing wedge, Model Upgrade Test, Flywheel) to prevent building commoditizable AI tools.
Alternatives in AI
- Startup Canvas — Generate Startup Canvas combining Product Strategy and Business Model 7.8k ★
- Bot On Anything — A large model-based chatbot builder that can quickly integrate AI models (including ChatGPT, Claude, Gemini) i 4.2k ★
- Fast Agent — Code, Build and Evaluate agents - excellent Model and Skills/MCP/ACP Support 3.7k ★
README
Services: The New Software — Claude Code Skill
A Claude Code skill that applies Sequoia's ["Services: The New Software"](https://sequoiacap.com/article/services-the-new-software/) framework to evaluate AI product ideas and business models.
The Problem
Every founder building an AI product faces the same risk: the next model upgrade turns your product into a built-in feature. Most AI startups are building **tools** (copilots) when they should be selling **outcomes** (autopilots).
This skill acts as a guardrail — it automatically triggers when you discuss side project ideas or business models with Claude Code, and runs the idea through Sequoia's framework to check whether you're building something defensible or something that will be commoditized.
What It Does
When triggered, the skill walks through 5 checkpoints:
| # | Check | What It Catches |
|---|---|---|
| 1 | Tool or Outcome? | Are you selling software or selling completed work? |
| 2 | Intelligence or Judgement? | Is the work rule-based (automatable now) or experience-based? |
| 3 | Already Outsourced? | Is there an existing budget you can substitute, or do you need a reorg? |
| 4 | Model Upgrade Test | Does the next Claude/GPT version help you or kill you? |
| 5 | Flywheel Potential | Does doing more work generate proprietary data as a moat? |
It also flags 5 common traps: "AI-powered X tool", "Dashboard for Y", "Copilot for Z profession", "Automating internal workflow", and the "Feature vs Service" test.
Install
**Option A: Claude Code Skills directory**
mkdir -p ~/.claude/skills/services-new-software
curl -o ~/.claude/skills/services-new-software/SKILL.md \
https://raw.githubusercontent.com/takeshiyu/services-new-software-skill/main/SKILL.md
**Option B: CLAUDE.md (append)**
echo "" >> CLAUDE.md
curl https://raw.githubusercontent.com/takeshiyu/services-new-software-skill/main/SKILL.md >> CLAUDE.md
Customization
The skill is designed to
Related Skills
Allocate
Stage 04. Invoke the solution-architect in allocate mode to place the intelligence — assign every operating-ma
Pde Loop
Brainstorm product ideas and have a different model grade them, so the model that proposed an idea never decid
Business Idea Validator AI Skill
Brutally honest business idea validator — a Claude skill using HBS frameworks to evaluate startup ideas
Product Mode
The most expensive bug in AI-assisted building is shipping the wrong thing, well. A CLAUDE.md for PM + eng tea
Monozukuri
An AI coding skill that applies Japanese-inspired craftsmanship to build simple, correct, verifiable, and main
Domain Name Brainstormer
Generates creative domain name ideas for your project and checks availability across multiple TLDs (.com, .io,
Related Agents
Magellan
Magellan — Research Analyst. Runs ONE open-web research lane of a pre-project evaluation: competition, demand,
Competitive Strategist
Maps everyone a new product would have to beat, from direct competitors to substitutes to doing nothing. Recor
Po Wedge
Wedge seat on the Atlas PO Council. Reduces requests to first principles, checks the live competitive landscap