3 AI wrappers with fresh traction: basketball coaching, thesis writing, and prank photos (2026)

[data] Three AI wrapper SaaS products cleared this week's evidence bar: Swish AI, Medra, and GoMytho. Swish AI is a basketball shot-analysis app with $2,347 MRR verified through RevenueCat. Medra is an academic-writing assistant with $137 MRR verified through Stripe. GoMytho is a prank-photo editor with $167 MRR verified through Stripe. All three were founded within the past 12 months, and each public record names one founder without listing a larger team. 1 2 3 4 [editor's view] The code is manageable. The harder question is whether you can reproduce the acquisition motion.

Three takeaways before the teardowns:

[data] Swish AI has the best public consumer signal: 30,000 downloads, about 800 active subscribers including annual plans, 327 App Store ratings, and $2,347 verified MRR. 1 2
[editor's view] Medra is easy to prototype but hard to market cleanly. Its "quickly and undetectably" thesis promise creates an immediate trust problem.
[editor's view] GoMytho is the clearest distribution lesson. The founder claims €562 MRR in 10 days from organic TikTok; the current verified snapshot is $167 MRR, with different currency and time windows. 4

Replication scoreboard

[editor's view] More stars mean harder to copy. For legal risk, more stars mean more exposure. These are editorial estimates from the public evidence.

Product	Technical lift	Information edge	Capital needed	Legal risk	Clone read
Swish AI	★★★☆☆	★☆☆☆☆	★★☆☆☆	★★☆☆☆	Strongest consumer proof, but video feedback and App Store retention are the work
Medra	★★☆☆☆	★☆☆☆☆	★☆☆☆☆	★★★★☆	Fast to prototype, exposed on trust and academic-use boundaries
GoMytho	★★☆☆☆	★★☆☆☆	★☆☆☆☆	★★★☆☆	Cheap web build with a useful TikTok asset, but the product can be copied quickly

Swish AI: a basketball coach in the camera roll

Field	Teardown
Positioning	[data] Swish AI is an iPhone basketball coach that records shooting practice, analyzes form frame by frame, and returns feedback on elbow alignment, foot placement, follow-through, release timing, balance, and posture. <MarkdownCitation index="2" title="Swish AI: Basketball Coach
Public traction	[data] TrustMRR reports $2,347 MRR, $3,390 in revenue over the last 30 days, 849 active subscriptions, and RevenueCat verification. The same listing reports about 30,000 downloads and about 800 active subscribers including annual plans. <MarkdownCitation index="1" title="Swish AI - $3,390 last 30 days
Launch and team signal	[data] Swish AI was founded in August 2025. Josh Karathra is the named founder, developer, and seller; no larger team is disclosed. <MarkdownCitation index="1" title="Swish AI - $3,390 last 30 days
Job-to-be-done	[editor's view] The job is "show me what is wrong with my shot, then give me a drill I can repeat." The input, feedback, and next action are visible.
Primary acquisition channel	[data] TrustMRR tags influencers as the marketing channel. The founder says scaling likely requires strong Meta Ads and TikTok Ads skills. <MarkdownCitation index="1" title="Swish AI - $3,390 last 30 days
Pricing and monetization	[data] The App Store lists free download with in-app subscriptions, including monthly tiers of $5.99 and $9.99, weekly tiers of $3.99 and $7.99, and annual tiers from $24.99 to $39.99. The listing says shot recording and analysis require a subscription. <MarkdownCitation index="2" title="Swish AI: Basketball Coach
Technical lift	[editor's view] ★★★☆☆. React Native and Expo reduce the mobile setup burden, but reliable frame-by-frame video analysis, feedback quality, camera permissions, and subscription handling are a real product loop. RevenueCat is listed as the backend billing component; the AI provider is not disclosed. <MarkdownCitation index="1" title="Swish AI - $3,390 last 30 days
Information edge	[editor's view] ★☆☆☆☆. No proprietary dataset or exclusive coaching content is disclosed. The possible edge is accumulated examples of bad form and better drills, but the public record does not show that this data is already a moat.
Capital needed	[editor's view] ★★☆☆☆. A narrow prototype can use one phone workflow and one shot type, but video inference and repeated user sessions create ongoing costs. The source record does not disclose operating costs.
Legal risk	[data] The App Store says Swish AI does not provide medical advice and lists a privacy policy. <MarkdownCitation index="2" title="Swish AI: Basketball Coach

[editor's view] The retention risk is more revealing than the download count. One App Store reviewer called the feedback generic, and the developer replied that the feedback engine was being overhauled. 2

[editor's view] If you wanted to copy this, start with one situation: high-school guards practicing catch-and-shoot threes alone. Ask 20 players to upload one clip, return a human-reviewed diagnosis plus one drill, and charge for the next five analyses. The first likely failure mode is generic feedback. If a player cannot tell what changed between two sessions, the app is a novelty camera filter with a subscription screen.

Medra: the thesis factory with a trust problem

Field	Teardown
Positioning	[data] Medra describes itself as an AI platform for studying, revising, solving assignments, and writing papers. Its sharper promise is generating theses and final-year projects in minutes. <MarkdownCitation index="3" title="Medra - $279 last 30 days
Public traction	[data] TrustMRR reports $137 MRR, $279 in revenue over the last 30 days, $4,907 in all-time revenue, eight active subscriptions, and Stripe verification. <MarkdownCitation index="3" title="Medra - $279 last 30 days
Launch and team signal	[data] Medra was founded in January 2026. Founder Mourad EL BAGHDADI says one operator can run it all. <MarkdownCitation index="3" title="Medra - $279 last 30 days
Job-to-be-done	[editor's view] The job is to move a student from a blank thesis outline to a structured draft or revision pass. The product's "undetectably" language makes the boundary less clear.
Primary acquisition channel	[data] The reviewed TrustMRR record does not identify a marketing channel. It reports a domain rating of 0 and does not provide a public traffic or conversion figure. <MarkdownCitation index="3" title="Medra - $279 last 30 days
Pricing and monetization	[data] No public plan prices or usage limits were available in the reviewed sources. The verified subscriptions show that payment exists, but the public record does not establish whether monetization is seat-based, credit-based, or subscription-only. <MarkdownCitation index="3" title="Medra - $279 last 30 days
Technical lift	[editor's view] ★★☆☆☆. A first version could combine document upload, outline generation, drafting, citations, and revision into a web workflow. The source record identifies Stripe but does not name the model provider or any proprietary retrieval system. <MarkdownCitation index="3" title="Medra - $279 last 30 days
Information edge	[editor's view] ★☆☆☆☆. No exclusive academic corpus, institutional partnership, or proprietary dataset is disclosed. The likely advantage is workflow packaging and language coverage, both of which are available to a focused competitor.
Capital needed	[data] The founder claims an 80% profit margin, very low operating costs, and one-operator operation. <MarkdownCitation index="3" title="Medra - $279 last 30 days
Legal risk	[editor's view] ★★★★☆. The product's academic-writing category is already sensitive, and the explicit promise of completing work "undetectably" raises a higher integrity and reputational risk than a normal writing assistant. The score reflects exposure, not a legal conclusion.

[editor's view] Medra shows why a low technical score does not equal a good business idea. A safer direction is a transparent tutor that explains sources, quizzes the student, and tracks revisions.

[editor's view] If you wanted to copy this, start with one compliant job: help nursing students turn approved lecture notes into an oral-exam study plan. Build source-grounded summaries, flashcards, and a citation checklist before adding long-form drafting. The first likely failure mode is trust. If the tool cannot show where an answer came from or what the student still needs to understand, the buyer has little reason to keep using it.

GoMytho: a prank-photo loop built for TikTok

Field	Teardown
Positioning	[data] GoMytho is an AI-powered web app for creating and editing photos to prank friends, family, and social circles. The listing positions it around instant results, low friction, and virality. <MarkdownCitation index="4" title="GoMytho - $233 last 30 days
Public traction	[data] TrustMRR reports $167 MRR, $233 in revenue over the last 30 days, $885 in all-time revenue, 12 active subscriptions, and Stripe verification. <MarkdownCitation index="4" title="GoMytho - $233 last 30 days
Launch and team signal	[data] GoMytho was founded in April 2026; Milo De Oliveira is the named founder, with no additional team listed. <MarkdownCitation index="4" title="GoMytho - $233 last 30 days
Job-to-be-done	[editor's view] The job is to turn a joke idea into a believable image before the group chat moves on. Speed and shareability are part of the product.
Primary acquisition channel	[data] TrustMRR tags TikTok. The founder claims €562 MRR in 10 days with zero ad spend from organic TikTok, says the TikTok account is included in the sale, and says the format is proven. <MarkdownCitation index="4" title="GoMytho - $233 last 30 days
Pricing and monetization	[data] The reviewed record does not disclose public subscription tiers or per-image prices. It confirms 12 active subscriptions but not the pricing structure behind them. <MarkdownCitation index="4" title="GoMytho - $233 last 30 days
Technical lift	[editor's view] ★★☆☆☆. Next.js, Supabase, Stripe, and Vercel are a familiar small-SaaS stack. The hard part is choosing edits that produce a reliably funny result quickly, not assembling the account and payment layer. <MarkdownCitation index="4" title="GoMytho - $233 last 30 days
Information edge	[editor's view] ★★☆☆☆. The TikTok account and repeatable content format are distribution assets, not proprietary information. They earn two stars because a proven creative loop is more useful than an empty launch account, but another creator can still reproduce the same formats.
Capital needed	[editor's view] ★☆☆☆☆. A clone can start with a small set of image transformations and pay per generation. The source record does not disclose costs, so the score assumes a deliberately narrow first version rather than unlimited free generation.
Legal risk	[editor's view] ★★★☆☆. Prank images can create consent, impersonation, privacy, and reputational problems. The score is an editorial risk estimate; the public listing does not provide a detailed rights or moderation policy.

[editor's view] GoMytho has the clearest distribution lesson, but little room for a passive owner. The founder's own sales pitch is "keep posting." The asset is partly a content operation.

[editor's view] If you wanted to copy this, start with one safe transformation that works on consenting users, such as turning a friend's portrait into a clearly fictional movie poster. Publish ten short demos, measure shares and paid conversion, and add moderation before expanding the menu. The first likely failure mode is a viral clip with no repeat use. A joke that works once is not a retention loop.

What this week says about copyability

[editor's view] The three cases point to different bottlenecks: Swish AI needs trusted feedback, Medra needs a transparent reason to exist, and GoMytho needs a stream of shareable content. None is solved by swapping in GPT, Claude, or an image model.

[editor's view] Swish AI is the best clone test if you can reach basketball players and review early outputs by hand. GoMytho is the fastest experiment if you understand short-form video. Medra is the cheapest technical build, but its public promise needs the sharpest narrowing.

Sources table

Source	Link
TrustMRR: Swish AI	Open listing
Apple App Store: Swish AI	Open listing
TrustMRR: Medra	Open listing
TrustMRR: GoMytho	Open listing

What to do tomorrow morning

[editor's view] Pick Swish AI or GoMytho and run a 20-person copy test. Record each prospect's input, paid result, reachable channel, and platform risk. If vague, do customer discovery before coding.

Cover image: generated editorial artwork for this week's three product categories.

3 AI wrappers with fresh traction: basketball coaching, thesis writing, and prank photos