From Tutoring Product to Skill Ecosystem: aitutors.me's Multi-Surface Plugin Strategy

Most subscription products treat their core app as the only real surface: everything else is marketing. aitutors.me has been building differently — the tutoring subscription is still the thing people pay for, but around it sits a small, growing set of surfaces that don't require a subscription at all: a public plugin repo, a standalone flashcards skill, install paths into three different AI platforms. Looking at that whole shape together says more about the strategy than any single piece does alone.

The core stays where it is

Nothing about this multi-surface push changes where the actual tutoring lives. Seven subject professors, the Socratic hint ladder, curriculum mapping, the safeguarding mechanism — all of it runs through the same aitutors.me/mcp connector, gated behind a subscription, whether a family reaches it through Claude, ChatGPT, or Codex. This site's own Best AI Tutors series and AI Tutoring series both cover that core product in depth — this piece is about what's been built around it, not a repeat of that ground.

Distribution: meeting families where they already are

The plugin repo's most direct function is distribution. A family already living in ChatGPT doesn't have to adopt Claude to reach the tutors — codex plugin marketplace add and ChatGPT's "Add plugin marketplace" flow both point at the same connector Claude users reach natively. This is a genuinely different move from building a standalone app for each platform, which is the more common (and more expensive) way products chase multi-platform reach. Shipping one MCP connector and three thin install paths gets most of the same distribution benefit for a fraction of the maintenance burden — the tutors don't need three separate codebases, just three separate doors into the same room.

The flashcards skill: giving away the technique, keeping the product

The flashcards skill is the more interesting piece strategically, not technically. It's MIT licensed, works without an aitutors.me account, and its JSON format is documented well enough that another tool entirely could adopt it. That's not accidental generosity — it's a bet that a well-designed, freely available card format becomes a small standard other builders route through, which does more for aitutors.me's visibility and credibility than keeping the format closed would. Giving away a technique that doesn't cannibalize the subscription is close to free marketing, if the technique is actually good enough that people cite and adopt it — which is the bet this specific skill is making.

Heddy IP: the asset that stays closed

The contrasting piece — a proprietary, "do not redistribute" brand-illustration skill sitting in the same public repo — shows the boundary of that generosity. Heddy is the character asset, not a technique, and the strategy doesn't extend to giving away the brand itself, even though the skill describing how to generate it is visible for anyone to read. That's a coherent line: open the parts that build goodwill without costing anything specific to the brand, keep closed the parts that are the brand.

Why this matters beyond one product

A tutoring company shipping a plugin ecosystem is a specific instance of a broader shift worth naming: products built around a connector protocol (MCP, in this case) get distribution options that products built around a single app never had. The same connector that serves a subscriber inside Claude serves them inside ChatGPT and Codex with almost no additional product work — the marginal cost of a new "surface" drops to roughly the cost of a manifest entry and an install guide, rather than a new app build. That changes the calculus for smaller teams specifically: a two-person product can plausibly reach three platforms' worth of users without three platforms' worth of engineering, as long as the core capability is genuinely connector-native rather than app-native from the start.

Whether other education products — or products in any category with a similarly connector-friendly core — follow the same shape is worth watching. The pattern here (core product gated behind subscription + connector, one genuinely useful technique given away MIT, brand assets kept proprietary but documented) is a reasonably complete playbook for any team weighing how much of its own skill catalogue to open up.

Part of the "AI Tutoring" series on aiskill.market.