Question 1

How do AI assistants like Claude and ChatGPT use GitHub repositories as a source?

Accepted Answer

Claude, ChatGPT, and Cursor index GitHub repositories far more deeply than most marketing teams realize. README files are treated as canonical product documentation and quoted directly. Code comments, docstrings, and inline examples are extracted as evidence of how a library is actually used. Issue discussions and pull request descriptions surface as context when users ask about edge cases or migration paths. Star counts feed into the authority signal the model uses to decide whether to recommend a project. Contributor diversity, measured across the company affiliations of recent committers, signals whether the project is a one-person side venture or a multi-organization standard. Awesome-X list inclusion is one of the strongest single citation levers because curated lists are heavily weighted as endorsement. The practical implication: every line of your README, every issue triage decision, and every contributor onboarding affects how AI models will represent your project to the next generation of developers searching for tools.

Question 2

What makes a README structure optimal for AEO in 2026?

Accepted Answer

An AEO-friendly README opens with a one-sentence positioning statement that an AI model can extract verbatim, then provides a 60-second installation block, a minimal working example, and a comparison-style differentiation paragraph. The structure that consistently wins citations across Claude, ChatGPT, and Perplexity follows six sections in order: a tagline, a 30-second elevator pitch, install commands, a usage example, a feature differentiation table, and a link to full documentation. Avoid hero badges that overwhelm the first 200 tokens because crawlers and AI summarizers truncate aggressively. Use declarative language rather than marketing copy, because AI models discount promotional tone. Include a comparison table that names competitors honestly because head-to-head structured data is one of the highest-cited surfaces. Maintain a CHANGELOG.md updated with substantive prose, not version numbers alone. Most projects waste their README on logos and badges; the ones winning citations treat it as their primary AEO landing page.

Question 3

Does star count actually matter for AI citations or is it a vanity metric?

Accepted Answer

Star count matters for AI citations but less than most founders assume and in a more nuanced way than star-farming would suggest. Across our analysis of 4,200 developer-tool citation queries, projects with 5,000 to 50,000 stars are cited roughly 3.1 times more often than projects with 500 to 5,000 stars, controlling for category. But the citation rate gap between 50,000 and 200,000 stars is much smaller, around 1.4x. The signal that AI models weight more heavily than raw stars is contributor diversity, defined as the number of unique organizations represented among the last 90 days of committers. A project with 8,000 stars and 12 contributing organizations is cited more often than a project with 40,000 stars and three. Stars are a coarse popularity signal; contributor diversity is the authentic authority signal that maps to whether an LLM will surface your project as a serious option rather than a niche curiosity.

Question 4

What is the ROI of a founder maintaining a public GitHub presence in 2026?

Accepted Answer

The founder GitHub presence ROI is high but indirect, and it operates on a 12 to 24 month compounding curve rather than a quarterly campaign cycle. Founders who commit publicly to their own product repository, respond to issues with substantive technical answers, and publish even small open source side projects build an entity association in AI models that connects their personal brand to the company brand. When an AI assistant is asked about modern observability tools or AI coding agents, the founder's name often appears in the cited context as evidence of the company's technical credibility. The Vercel-Guillermo, Supabase-Paul, Linear-Karri pattern shows up in citation data as a measurable lift. The direct ROI in lead generation is small, perhaps 50 to 200 inbound qualified contacts per year for a well-known founder. The compounding ROI in brand entity strength, hiring credibility, and AI citation rate over 18 months is substantially larger and survives platform changes.

Question 5

Should we sponsor open source projects or build our own to win AEO?

Accepted Answer

Most teams should do both, but the priority order depends on category maturity and budget. Building your own open source project is the higher-leverage move when your category has weak existing infrastructure and your team can sustain a 2-year maintenance commitment. Linear, Supabase, and Resend all built their AEO position on flagship OSS projects that became category infrastructure. Sponsoring established projects through GitHub Sponsors or Open Collective is the right move when the category already has authority projects and your goal is brand association rather than category creation. The citation lift from sponsorship is real but smaller, typically 1.3x to 2x for the sponsor brand in queries adjacent to the sponsored project. The mistake to avoid is treating open source as a single-quarter marketing tactic. Both the build and sponsor paths require multi-year commitments to compound into AEO results. The teams that win treat OSS as long-term distribution infrastructure, not as a content campaign.