Summary

Rough Cut Pete (a filmmaker's channel; this video reportedly has nearly 1.5 million views against 8.3K subscribers) looks at Brandon Thomas, an Atlanta developer who says he was hit by Adobe's early-cancellation fee and responded by building free, open-source rough equivalents of seven Adobe apps, largely with Claude, in about a week. The video runs in three parts: the project and how far it really gets, Adobe's long record of customer-hostile moves alongside its record revenue and falling share price, and whether Adobe can stop it legally. Pete is careful to puncture his own headline: the apps are self-described 'super early alpha', not a solo effort, and 'free' is funded by AI credits. He discloses that he uses Claude for research and edits in DaVinci Resolve, not Premiere. All figures are his, reported not verified. Caption errors: the suite is called 'Artboard' in later sections, and one free Photoshop alternative from 1996 (almost certainly GIMP) drops out of the transcript.

• Brandon Thomas (Atlanta; describes himself as engineer and filmmaker; about 15 years in software, about 10 in Rust; founder of FakeYou and the AI film startup Storyteller, now Artcraft) says he was charged Adobe's 'dark pattern' cancellation fee more than once.
• Claimed timeline: first line of code 30 Sep 2026, first release of PhotoCraft 2 Oct, posted to Reddit and Hacker News 7 Oct. Seven apps: PhotoCraft (Photoshop), VectorCraft (Illustrator), FilmCraft (Premiere), LightCraft (Lightroom), EffectCraft (After Effects), DesignCraft (InDesign) and PrintCraft, renamed PDFCraft (Acrobat). They run on Mac, Windows, Linux and in a browser, under MIT/Apache open-source licences.
• Built with Claude: Pete says he went through the public code history and found around 7,500 changes in about 10 days, roughly 60% listing Claude as co-author. PhotoCraft reportedly drew nearly 40,000 GitHub stars and nearly 6,000 forks within days.
• Thomas is quoted as saying 'Software is over' and promising to replace Google Search, build smartphones and 'de-enshittify' tech. He has reportedly started on PowerPoint, Excel, AutoCAD and Pro Tools.
• Quality, per one reviewer on a weak laptop: fast, with layers, masks, smart objects and brushes, and it opens PSD files; but blend modes and filters don't match Photoshop, it used 5 GB of memory on one file, and the verdict was 'not for professionals'. Thomas calls it a 'super early alpha' and has walked back his promise of 100% Adobe parity within a month to 'months, not years'.
• The caveats Pete adds: it is not a lone genius — a funded startup with a small team and dozens of volunteers — and free isn't quite free: the apps cost nothing, but the company sells AI image/video credits and cloud storage.
• Adobe's position: record quarterly revenue reported at $6.7 billion and more than a billion monthly users claimed, but the stock reportedly fell from about $680 (Nov 2021) to around $240, roughly 65% down. Goldman Sachs reportedly issued a rare sell rating, new subscription business reportedly fell by about a third, and the CEO of 19 years steps down in December.
• Adobe's record as Pete lists it: the 2013 move to subscription-only (Master Collection was a one-off $2,500; plans have risen from $50 to $70 a month, or $105 month-to-month — over $8,000 across 10 years); the 2013 breach of 38 million users' details plus Photoshop source code; the 2019 withdrawal of old versions with infringement warnings; the 2019 Venezuela account shutdown under US sanctions; Pantone colours turning black in 2022 without an extra subscription; and the 2024 terms-of-use row over access to user content.
• The cancellation fee: early exit from the annual plan paid monthly can cost half the remaining year. The US government sued in 2024 alleging the fee was buried and cancelling made deliberately hard, naming two executives. Adobe reportedly settled in March 2026 for $150 million ($75M penalties, $75M in free services) while denying wrongdoing. The UK CMA has reportedly opened its own investigation.
• Firefly, sold as the 'ethical' AI, was reported by Bloomberg to include Midjourney-generated training images. Stock contributors reportedly couldn't opt out, and the Ansel Adams estate publicly told Adobe it was 'on our last nerve'. The $20 billion Figma deal was blocked, costing Adobe a reported $1 billion break fee.
• Competition: Canva claimed at over 260 million users and has made the Affinity apps free; DaVinci Resolve has long had a free tier. Adobe's CEO reportedly said on the last earnings call he was glad they had 'deferred pricing actions in favor of user growth'. Adobe's free users reportedly rose 70% last year.
• Legal outlook, by someone who stresses he is not a lawyer: copyright precedent favours rebuilders (Lotus v Borland 1995 on menu structures; Apple's lost look-and-feel case; Google v Oracle 2021 on APIs; reverse engineering for compatibility as fair use). Adobe publishes the PSD spec, Thomas says it is a clean-room rebuild with no Adobe code, and Adobe has apparently never sued GIMP or Photopea. Exposure: trade dress (some apps look near-identical), Adobe names in project tags, and Adobe's thousands of patents.
• Precedent for crushing open source: Nintendo's Yuzu settlement ($2.4M, 8,500+ copies removed), but that involved breaking encryption for piracy. youtube-dl was restored within a month, and the DVD-cracking code ended up on t-shirts. MIT/Apache licences can't be withdrawn from copies already out there.
• Pete's four scenarios: ignore it (most likely), a lawyer's letter about icons and layout, a 'nuclear' patent suit (bad PR given the fee settlement), or compete with cheaper plans, free tiers and easier cancelling.
• His conclusion: the story is the recipe, not this project — if a small team plus AI can rough out seven apps in 10 days, others will follow. Adobe probably survives as a company (pipelines, plugins, training and decades of files don't switch overnight), but not as a business built on rent, price rises and lock-in. He advises downloading only from the official page, not trusting the alpha with client deadlines, keeping backups, and checking your plan type before cancelling.

Why it matters

This is a concrete, checkable data point for the 'AI collapses the cost of building software' thesis: claimed git history showing about 60% Claude co-authorship across seven apps in roughly 10 days. The same source also notes that the output is alpha quality, the team is funded and not one person, and the 100%-parity promise has already slipped. So the durable lesson is about lock-in more than about Adobe being replaced: when a rough clone can appear in a week, subscription businesses that keep customers by trapping them lose that hold, and the creative tools Ashley's world depends on face real price pressure. Treat the star counts, commit figures and Adobe financials as one creator's reporting until a second source confirms them.

pTAZtrsU-84-transcript.txt