Anthropic's Boris Cherny: Why Coding Is Solved, and What Comes Next

Kernpunkte

1. Datum / Quelle. Sequoia-Capital-Videointerview (YouTube), URL https://www.youtube.com/watch?v=SlGRN8jh2RI. Genaues Datum im Transkript nicht erkennbar; interne Hinweise (Opus 4.7 als aktuelles Modell, "we also just launched routines") verorten es auf Ende April / Anfang Mai 2026. Datum der Aufnahme ins Veille-Dossier: 2026-05-05.
2. Sprecher. Boris Cherny, Schöpfer von Claude Code, Anthropic. Interviewerin: Lauren Reader (Sequoia). Anmoderation: Asia (Sequoia).
3. Enthüllte Hintergrundinformationen zu Cherny.
4. YC-Alumnus, erster Mitarbeiter bei einem YC-Unternehmen, mehrere frühere Startups.
5. Autor des Buchs "Programming TypeScript".
6. Hat 2026 keine einzige Zeile Code geschrieben. (Aussage bei Sequoia vor dem Interview).
7. Anekdote aus der Mittelstufe: schrieb einen BASIC-Leitfaden für den TI-83 Plus, "still on the internet, extremely embarrassing".
8. Komischer Einstieg.
9. "Show of hands, who here uses Claude Code?" → Mehrheit.
10. "Show of hands, who here has Claude code psychosis?" → Gelächter.
11. "My team lovingly says I have Claude code psychosis, which may or may not be true." ### Geschichte von Claude Code (neuer Bericht)
12. Ursprung. Cherny tritt Ende 2024 Anthropic Labs bei — Anthropics internem Inkubator, einem Team aus wenigen Personen.
13. Ergebnisse von Anthropic Labs. Claude Code, MCP, Desktop-App. Team nach Erfüllung des Auftrags aufgelöst.
14. Aktuelle Runde 2. Team neu formiert, geleitet von Mike Krieger (Anthropic Chief Product Officer, ehemaliger Instagram-Mitgründer).
15. Warum Cherny mit dem Coding begann. "product overhang" — "the model can do all the stuff that no product has yet captured."
16. Stand der Technik im Coding, Ende 2024. "type ahead" (IDE öffnen, Tab drücken, eine Zeile vervollständigen). Das hatte Sonnet 3.5 ermöglicht.
17. Die Wette. weiter gehen — "the model was almost ready for the next big step. So we don't have to do type ahead anymore, we can just have the agent write all of the code."
18. Die ersten 6 Monate. "it just really didn't work. It was like not very good. It was barely usable. I used it for maybe 10% of my code."
19. Erstes Claude-Code-Release. "It was not a hit. There's a lot of people that used it, but it did not have this exponential growth that it has today."
20. Wendepunkt. Opus 4 im Mai 2025. "That's like when the exponential growth started, and then it kind of inflected with every model release. It started with Opus 4, then 4.5, then 4.6, now 4.7. It just kind of keeps inflecting."
21. Eine bewusste Pre-PMF-Wette. "essentially we were trying to build this thing that was pre-PMF, and we knew that it wouldn't have PMF for 6 months because we were building for the next model." ### "Coding is solved"
22. Handzeichen-Umfrage:
23. "Who writes 100% of their code by hand?" → wenige
24. "Who writes 100% using an agent like Claude Code?" → wenige
25. "Who's somewhere in between?" → Mehrheit ("like 50% solved")
26. Cherny selbst: 100%. .
27. Codebasis von Claude Code.
28. "It leaked, so people know."
29. TypeScript + React. — gewählt, weil "very on distribution for the model" (das damalige Modell war in diesen Technologien stark).
30. "Pretty simple. There's nothing really complicated."
31. Das Modell hat seit Oktober/November 2025 100% des Codes geschrieben. .
32. "For me today, the model writes 100% of my code. I write somewhere usually a few dozen PRs every day."
33. "There was a day last week I did like 150 PRs in a day. That was a record."
34. Eingeräumte Grenzen. "this is not the case everywhere. There's very big complicated codebases. There's weird languages the model isn't good at yet. Usually the answer is just wait for the next model." ### Aktuelles persönliches Setup (Updates)
35. iOS zuerst. "most of my work I do from my phone."
36. Claude-App → Code-Tab. "a little code tab on the left-hand side, I just have a bunch of sessions going."
37. 5-10 Sessions. geöffnet, "a few hundred agents" tagsüber aktiv, "a few thousand" nachts für tiefergehende Arbeit.
38. Sub-Agenten. + Slash-Befehl /loop = die neue zentrale Primitive.
39. *"Loops are the future at this point. If you haven't experimented with it, highly highly recommend it.". * ### Loop: die neue Primitive
40. Mechanismus. "You have Claude use cron to schedule a job for some point in the future, and it's a repeat job."
41. Taktung. jede Minute / alle 5 Minuten / täglich — konfigurierbar.
42. Chernys Beispiele.
43. PR-Babysitter. "fixing CI, auto-rebasing"
44. CI-Healthkeeper. "if there's a flaky test or whatever, it'll go and fix it"
45. Twitter-Feedback-Clusterer. "grabs feedback from Twitter and clusters it for me every 30 minutes"
46. Routines. = das serverseitige Äquivalent des Loops: "we just launched routines. Even if you close your laptop, it keeps going."
47. 4.7 startet Loops von sich aus. "I'll tell it 'pull this data query.' And it's like 'Hey, I noticed the data is changing over time. I'll start a loop and I'll give you a report every 30 minutes.' And I'm like 'Great. Can you send it to me over Slack?' And then it uses the Slack MCP to do that." ### Ausblick für Teams
48. "There's going to be a lot more generalists than there are today."
49. Nicht nur "product engineers cross-platform" (iOS+Web+Server), sondern fachübergreifende Generalisten: "engineers that are really good at product engineering, but also really great at design. Or really great at product and data science and engineering."
50. Im Claude-Code-Team.
51. "Every single person on our team writes code."
52. Liste: Engineering Manager, PM, Designer, Data Scientist, Finance-Mitarbeiter, User Researcher.
53. "They're specialist in something, but now also everyone's just coding." ### Ausblick SaaS / Disruption
54. Keine SaaS-Apokalypse. , aber eine Neuordnung über Hamilton Helmers 7-Powers-Framework.
55. Powers, die an Bedeutung verlieren.
56. Switching costs. — "because you can just use the model and you can port from one thing to a different thing."
57. Process power. — "Claude is getting really good at figuring out process. And especially with 4.7, it can just hill climb anything. So if you give it a target and you tell it to iterate until it's done, it will just do it. I think this is the first model like that."
58. Powers, die weiterhin wichtig bleiben.
59. Network effects.
60. Scale economies.
61. Cornered resources.
62. Startup-Prognose. "the number of startups in the next 10 years that are just going to disrupt everything is going to increase like 10x."
63. Warum gerade jetzt. "a tiny startup can build a thing that's as valuable as a large company and actually compete head-to-head, because the large company has to evolve their business process, retrain everyone, face internal resistance. But you, starting fresh, can build with AI natively from the ground up."
64. "It's the best time to build. It's the best time to be a startup. There's so much disruption coming." ### Pivot-Analogie — die Gutenberg-Presse
65. Europa im 15. Jahrhundert vor der Druckerpresse. "essentially 10% of the European population was literate. They knew how to read and write. They were often employed by kings and lords that were not literate."
66. 50 Jahre nach der Erfindung. "more literature published in Europe than in the thousand years before."
67. Kosten eines Buchs. ↓ 100-mal im gleichen Zeitraum.
68. Einige Jahrhunderte später. "literacy globally went up to like 70%. Now we can all read and write. You don't need a degree in reading and writing to know how to read and write. Although still there are professional writers."
69. Auf 2026+ angewandte Analogie. "software will be a thing that is fully democratized, that anyone can do. And it's going to be much faster than 50 years."
70. Denkwürdiger Pivot-Satz. "The best person to write accounting software, I think maybe even today, is not an engineer, it's a really good accountant. Because they know the domain really well, and coding is the easy part. It's knowing the domain that's the hard part." ### Modell vs. Harness (Q&A)
71. "If you asked me a year ago, the ratio was maybe 50/50. If you asked me 6 months ago, would be 50/50. In 2 years? I don't know dude, we plan in like 1 week out."
72. Der Harness wird weniger wichtig. , je besser das Modell wird.
73. Entwicklung des Harness.
74. Loops als First-Class-Feature
75. Einfacheres Ausführen vieler Agenten
76. Sub-Agenten (eine Idee unter anderen)
77. "There's a bunch more stuff that we're cooking."
78. In 1 Jahr. "the model will be much better aligned. And so all the safety mechanisms that we have today around prompt injection and static verification of commands and permission modes, human in the loop, all this kind of stuff is just going to be less important because the model will just do the right thing." ### Demokratisierung: Jeder codet so selbstverständlich, wie jeder eine SMS schreiben kann
79. "I think it's going to be a skill like — yeah, like I know how to send a text message."
80. Starker Bezug zum oben genannten Gutenberg-Presse-Modell. ### Interne Praktiken bei Anthropic
81. Kein Modellvorteil. "internally, we use the same models everyone else does." Mythos zum Testen, viel Opus 4.7 fürs Dogfooding.
82. Radikale Praxis.
83. "We use Claude for literally everything."
84. "Our Claudes are talking all day. As I'm coding, as my Claudes are coding in a loop, they will communicate over Slack to talk to other people's Claudes that are also running in a loop, to figure out unknowns."
85. "We have no more manually written code anywhere at the company."
86. "All of the SQL is written by models. Everything is just built by the models."
87. Anthropics Vorsprung liegt nicht in der Technologie, sondern im Prozess. "the place that we're ahead is not the technology — same technology available to us is available to everyone here, because fundamentally we are building a platform. There's actually a far bigger lead in organizational structure and organizational process." ### Multi-Agent / Parallelisierung
88. "On the product side, it really just comes down to prompting. As the model gets better, it just naturally does this."
89. "It's not on users to figure out how to hold the tools better. If that's the case, it's actually a product design problem and I'm not doing a good job." ### Lokal vs. Cloud, Computer Use, MCP
90. Lokal vs. Cloud: "doesn't matter. By a couple years from now, the model is just going to be doing all the code. It's going to be starting the agents. It's going to be building the environments. If it decides 'actually I'll use local models to do this,' that's what it'll do. These I don't think these will be decisions that we are making as engineers anymore."
91. Wissensarbeit: "largely, it's there already with Salesforce and Docs. For us, it's always just the simplest answer. It's MCP."
92. Computer Use: "I think currently, Anthropic is pretty far ahead on computer use. It's quite good. It's very slow, but it does it quite well now, especially with 4.7."
93. "All this stuff just doesn't matter that much. It could be MCPs, APIs, just some sort of programmatic access — to the model, it's just tokens." ### Kommende Produkte
94. Claude Design. "It's pretty good today. It's going to get a lot better."
95. Claude Code. "There's also a few things that we're cooking up for Claude Code that are going to be landing over the coming weeks."
96. Loop, Batch, massive Agentenparallelisierung. "that's going to get better."
97. Computer Use. . ### Bezug zum Veille-Dossier
98. Vierte Cherny-Fiche. im Dossier, nach:
99. cherny-yc-lightcone-claude-code-origin-story-2026-02. (YC-Lightcone-Entstehungsgeschichte)
100. cherny-claude-code-10-tips-team-x-2026-02-01. (10 Tipps auf X)
101. nunez-cherny-claude-code-workflow-venturebeat-2026-01-05. (VentureBeat-Workflow)
102. wu-cherny-use-claude-code-builders-every-2025-10-29. (Wu/Cherny bei Every)
103. Sowie die Fiche anthropic-claude-code-quality-postmortem-2026-04-23, deren impliziter Mitautor Cherny ist.
104. Spezifischer Beitrag dieser Fiche. im Vergleich zu früheren:
105. Ausschließliches iOS-Setup. = neu (im Vergleich zum CLI-/Desktop-Setup in früheren Fiches).
106. /loop + Routines. = eine neue, öffentlich vorgestellte agentische Architektur-Primitive.
107. Bericht "Anthropic-Labs-Inkubator + Mike-Krieger-Runde-2". = neues organisatorisches Detail.
108. Opus-4-Wendepunkt, Mai 2025. = präzise Datierung des Durchbruchs.
109. "100% Code generiert seit Oktober/November 2025". = persönliche Datierung.
110. Gutenberg-Analogie. + "der beste Buchhalter schreibt die beste Buchhaltungssoftware" = denkwürdige Pivot-Sätze.
111. Neuordnung der "7 Powers" nach Helmer. = eine strukturierte SaaS-Analyse.
112. Sequoia / Lauren Reader. = neuer Rahmen, Investorenpublikum.
113. Wesentliche Konvergenz.
114. Cron-Agenten via /loop. = eine radikale praktische Anwendung des Ralph Loop (Trivedy 2026-03-10), von Everys Compound Engineering (Shipper/Klaassen 2025-12-11) und Compounding Teams (Schillace 2025-09-28).
115. "Claudes talking to Claudes over Slack". = eine praktische Umsetzung von Levies Vorhersagen in Building for Trillions of Agents (2026-03-07) und der von Karpathy propagierten agentennativen Infrastruktur (2026-04-29).
116. "Coding is solved" + 100% generierter Code + 150 PRs/Tag. deckt sich mit Curran/Intercoms 3-facher F&E-Produktivität (2026-04-16, 93,6% agentengesteuerte PRs), Stripe Minions (2026-02-09 und 2026-02-19), StrongDM Software Factory (2026-02-06).
117. Demokratisierung nach Gutenberg-Vorbild. = dieselbe These wie bei Karpathy (2026-04-29) zu neural computers und bei Andreessen zur Neudefinition des Programmierers (2026-02). Auch übereinstimmend mit dem von Wescale (2026-05-03) und Habert PROJ-AI (2026-05-05) vertretenen Projekt "jeder kann coden".
118. "7 Powers" und SaaS-Neuordnung. = legt ein im 2026er-Korpus seltenes Unternehmensstrategie-Framework dar, komplementär zu Sun NYTs Permanent Underclass (2026-04-30) auf der ökonomischen Seite und Pawlowskis Tension Map (2026-03-30) auf der marktforschungsseitigen Seite.
119. "Der Harness wird weniger wichtig, je besser das Modell wird". = eine direkte Nuancierung des Trends Harness Engineering (Trivedy 2026-03-10, Osmani 2026-04-19, Böckeler 2026-04-02). Cherny antizipiert die zunehmende Aufnahme des Harness in das ausgerichtete Modell.
120. "Der beste Buchhalter schreibt die beste Buchhaltungssoftware". = ein Korollar des Long Tail an Geschäftsbedürfnissen, den die Wescale-Fiche (2026-05-03, "long-standing needs that were too costly can finally be addressed") und Sotos Developer Taste (2026-04) bereits antizipiert hatten.
121. Zu benennende Grenzen.
122. Bewusster technikzentrierter Optimismus. — Cherny ist der Schöpfer von Claude Code, ein offensichtlicher Evangelisten-Bias bei "coding is solved".
123. Keine externen Zahlen. — er stützt sich auf seine eigene Praxis und die von Anthropic. Zu vergleichen mit Curran/Intercoms belastbaren Zahlen (3-fache F&E, 93,6% agentengesteuerte PRs).
124. Sequoia-Format. keine kontroverse Befragung, gesellige Atmosphäre, kein Widerspruch zur jagged intelligence (Karpathy 2026-04-29) oder zur permanent underclass (Sun NYT 2026-04-30).
125. "Coding is solved". eine provokante Aussage, die durch Karpathys "Opus 4.7 will simultaneously refactor a 100,000 line codebase and tell me to walk to a car wash 50 meters away" (2026-04-29) relativiert werden sollte — oberflächliches Coding ist tatsächlich automatisierbar, aber systemisches Urteilsvermögen bleibt ein schwieriges Feld.
126. Nutzbar für. eine 18-Monats-Retrospektive zu Claude Code (Narrativ + Datierung); einen Startup-Pitch "beste Zeit zum Gründen" für die Geschäftsleitung; eine Präsentation "Loops als Zukunft" für Engineering-Teams; einen Kurs/Vortrag zur "Softwaredemokratisierung anhand der Gutenberg-Analogie"; eine Debatte über die Stellung des Harness in der 2026er-Doktrin.