{"id":"9ff09b590acf347204deada6fc5b61dfb16e537adc47c7595f58799f4bfb8aac","title":"Ethan Mollick argues in One Useful Thing that AI tools can sustain real work","argument":"Ethan Mollick’s article, published January 7, 2026, argues that Claude Code’s combination of a capable AI model and an “agentic harness” lets it carry out sustained work with limited input. In his demonstration, it built and deployed a website, then used a browser to test it; Mollick notes that the site contained “very sketchy fake marketing claims.” He describes tools including Skills, which provide instructions and tools for tasks, and subagents, which handle specialized work. He argues that these approaches help address limits such as a context window filling up: Claude Code can compact its conversation into notes, then continue from them. The tool can also access files and a browser, which Mollick says creates risks including deleting files, unintended code execution and exposure of sensitive data. His conclusion is that programmers should explore these tools, while people who work with data, design or ideas for things to build should experiment too. More broadly, he argues that AI with the right harness can do sustained work that matters, and that new harnesses for other knowledge tasks are coming. The article leaves open how these tools will change work beyond programming.","topicIds":["artificial-general-intelligence"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on October 24, 2026; originally published January 7, 2026. Mollick’s account describes demonstrations and his assessment of the tools; it does not establish how reliably they perform across other tasks or users. The article says the website contained fake marketing claims; it does not establish that any customer purchased the product.","speaker":{"name":"One Useful Thing","roleAtSource":"Publisher"},"source":{"title":"Claude Code and What Comes Next","publisher":"One Useful Thing","format":"Article","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","publishedAt":"2026-01-07T23:00:10.000Z","dateBasis":"Date supplied by the publisher’s feed."},"evidence":[{"text":"Ethan Mollick says Claude Code built and deployed a website after a single request, but the site included fake marketing claims.","excerpt":"creating hundreds of code files and prompts. And then it gave me a single file to run that created and deployed a working website (filled with very sketchy fake marketing claims)","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"},{"text":"Mollick attributes the tools’ capabilities to advances in autonomous work and to an “agentic harness” of tools and approaches.","excerpt":"First, the latest AIs are capable of doing far more work autonomously while self-correcting many of their errors, especially in programming tasks. Second, the AIs are being given an “agentic harness”","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"},{"text":"Mollick says Claude Code compacts its conversation into notes when its context window fills, then resumes from those notes.","excerpt":"When it runs out of context, it stops and “compacts” the conversation so far, taking notes about exactly where it was when it stopped.","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"},{"text":"Mollick warns that access to a user's computer and browser creates risks, including file deletion, unintended code execution and access to sensitive data.","excerpt":"The AI might delete files it shouldn't, execute code with unintended consequences, or access sensitive data in your browser.","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"},{"text":"Mollick argues that people beyond programmers should experiment with these tools, and that new harnesses for other knowledge tasks are coming.","excerpt":"If you're programming-adjacent (an academic who works with data, a designer who wants to experiment with code, anyone who wants to try building a thing they are imagining) this is your moment to experiment.","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What kinds of knowledge work can these tools sustain when adapted beyond programming, and what risks will that create for people using them?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Ethan Mollick argues in One Useful Thing that AI tools can sustain real work","summary":"Ethan Mollick’s article, published January 7, 2026, argues that Claude Code’s combination of a capable AI model and an “agentic harness” lets it carry out sustained work with limited input. In his demonstration, it built and deployed a website, then used a browser to test it; Mollick notes that the site contained “very sketchy fake marketing claims.” He describes tools including Skills, which provide instructions and tools for tasks, and subagents, which handle specialized work. He argues that these approaches help address limits such as a context window filling up: Claude Code can compact its conversation into notes, then continue from them. The tool can also access files and a browser, which Mollick says creates risks including deleting files, unintended code execution and exposure of sensitive data. His conclusion is that programmers should explore these tools, while people who work with data, design or ideas for things to build should experiment too. More broadly, he argues that AI with the right harness can do sustained work that matters, and that new harnesses for other knowledge tasks are coming. The article leaves open how these tools will change work beyond programming.","question":"What kinds of knowledge work can these tools sustain when adapted beyond programming, and what risks will that create for people using them?","claims":[{"text":"Ethan Mollick says Claude Code built and deployed a website after a single request, but the site included fake marketing claims.","excerpt":"creating hundreds of code files and prompts. And then it gave me a single file to run that created and deployed a working website (filled with very sketchy fake marketing claims)","support":"source-stated"},{"text":"Mollick attributes the tools’ capabilities to advances in autonomous work and to an “agentic harness” of tools and approaches.","excerpt":"First, the latest AIs are capable of doing far more work autonomously while self-correcting many of their errors, especially in programming tasks. Second, the AIs are being given an “agentic harness”","support":"source-stated"},{"text":"Mollick says Claude Code compacts its conversation into notes when its context window fills, then resumes from those notes.","excerpt":"When it runs out of context, it stops and “compacts” the conversation so far, taking notes about exactly where it was when it stopped.","support":"source-stated"},{"text":"Mollick warns that access to a user's computer and browser creates risks, including file deletion, unintended code execution and access to sensitive data.","excerpt":"The AI might delete files it shouldn't, execute code with unintended consequences, or access sensitive data in your browser.","support":"source-stated"},{"text":"Mollick argues that people beyond programmers should experiment with these tools, and that new harnesses for other knowledge tasks are coming.","excerpt":"If you're programming-adjacent (an academic who works with data, a designer who wants to experiment with code, anyone who wants to try building a thing they are imagining) this is your moment to experiment.","support":"source-stated"}],"limitations":["Text as retrieved on October 24, 2026; originally published January 7, 2026.","Mollick’s account describes demonstrations and his assessment of the tools; it does not establish how reliably they perform across other tasks or users.","The article says the website contained fake marketing claims; it does not establish that any customer purchased the product."],"mediaId":null,"correctionReason":null,"id":"9ff09b590acf347204deada6fc5b61dfb16e537adc47c7595f58799f4bfb8aac","publicationId":"f49da133-ac89-2252-207b-a087a68d72fc","version":1,"sourceUrl":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","sourceTitle":"Claude Code and What Comes Next","publisher":"One Useful Thing","publishedAt":"2026-01-07T23:00:10.000Z","originalPublished":{"value":"2026-01-07T23:00:10.000Z","precision":"instant","basis":"backfill:feed_published"},"firstObservedAt":"2026-09-23T05:41:18.181Z","createdAt":"2026-11-09T09:33:24.488Z","historical":true,"entityIds":["one-useful-thing"],"topicIds":["artificial-general-intelligence"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"e5d0d0e47aa4d7fe531842c94c988c055b91d6431ea28159a4cccff16dde94b0","rubricVersion":"unrecorded","provider":"unrecorded","model":"anthropic/claude-opus-5.5","reasonCodes":["bridge_admitted_by_model","second_opinion_categorical"],"confidence":1,"decidedAt":"2026-10-24T07:51:42.482Z"},"provenance":{"textHash":"f8f3bef4870be94d2b54f9b9d8aff5a814bf5de2c970ded6262a6c418a40e718","observationId":"7ae219803cf3b656a6f92b0fffa26d31527eb3a655cc37788385d0080f9ae84e","retrievedAt":"2026-10-24T07:53:38.044Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"6a808321-377f-f91b-448e-28c9a1e4b277"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"e45ae6c4-5b4a-13ea-5401-0673c8fd1cba"}},"audit":{"status":"passed","reviewedAt":"2026-11-09T09:28:38.014Z","model":"anthropic/claude-opus-5.5"}},"canonicalUrl":"https://escapemeridian.com/readings/ethan-mollick-argues-in-one-useful-thing-that-ai-tools-can-9ff09b59","formats":{"html":"/readings/ethan-mollick-argues-in-one-useful-thing-that-ai-tools-can-9ff09b59","markdown":"/readings/ethan-mollick-argues-in-one-useful-thing-that-ai-tools-can-9ff09b59.md","json":"/readings/ethan-mollick-argues-in-one-useful-thing-that-ai-tools-can-9ff09b59.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-agentic","name":"Agentic AI","explanation":"AI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.","referenceUrl":"https://developers.google.com/machine-learning/glossary#agentic"},{"id":"term-context-window","name":"Context window","explanation":"How much text a model can take in and consider at one time.","referenceUrl":"https://developers.google.com/machine-learning/glossary#context-window"}],"entities":[{"id":"ethan-mollick","name":"Ethan Mollick","kind":"person","context":"Ethan Mollick is a Wharton researcher and author exploring how AI changes work, learning and decision-making.","referenceUrls":["https://mgmt.wharton.upenn.edu/profile/emollick/","https://www.oneusefulthing.org/"],"path":"/entities/ethan-mollick","json":"/entities/ethan-mollick.json","markdown":"/entities/ethan-mollick.md"},{"id":"one-useful-thing","name":"One Useful Thing","kind":"organization","context":"One Useful Thing is Ethan Mollick’s publication about AI in work, education and everyday practice.","referenceUrls":["https://www.oneusefulthing.org/"],"path":"/entities/one-useful-thing","json":"/entities/one-useful-thing.json","markdown":"/entities/one-useful-thing.md"},{"id":"claude","name":"Claude","kind":"model","context":"Claude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.","referenceUrls":["https://claude.com/product/overview","https://www.anthropic.com/news","https://www.anthropic.com/research/team/economics"],"path":"/entities/claude","json":"/entities/claude.json","markdown":"/entities/claude.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"9ff09b590acf347204deada6fc5b61dfb16e537adc47c7595f58799f4bfb8aac","revision":"f49da133-ac89-2252-207b-a087a68d72fc","revisionKind":"publication-id","title":"Ethan Mollick argues in One Useful Thing that AI tools can sustain real work","url":"https://escapemeridian.com/readings/ethan-mollick-argues-in-one-useful-thing-that-ai-tools-can-9ff09b59"},"source":{"type":"source","id":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","revision":"f8f3bef4870be94d2b54f9b9d8aff5a814bf5de2c970ded6262a6c418a40e718","revisionKind":"sha256-source-text","title":"Claude Code and What Comes Next","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next"},"status":"partial","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"9ff09b590acf347204deada6fc5b61dfb16e537adc47c7595f58799f4bfb8aac","title":"Ethan Mollick argues in One Useful Thing that AI tools can sustain real work","argument":"Ethan Mollick’s article, published January 7, 2026, argues that Claude Code’s combination of a capable AI model and an “agentic harness” lets it carry out sustained work with limited input. In his demonstration, it built and deployed a website, then used a browser to test it; Mollick notes that the site contained “very sketchy fake marketing claims.” He describes tools including Skills, which provide instructions and tools for tasks, and subagents, which handle specialized work. He argues that these approaches help address limits such as a context window filling up: Claude Code can compact its conversation into notes, then continue from them. The tool can also access files and a browser, which Mollick says creates risks including deleting files, unintended code execution and exposure of sensitive data. His conclusion is that programmers should explore these tools, while people who work with data, design or ideas for things to build should experiment too. More broadly, he argues that AI with the right harness can do sustained work that matters, and that new harnesses for other knowledge tasks are coming. The article leaves open how these tools will change work beyond programming.","topicIds":["artificial-general-intelligence"],"domain":"other","claimType":"attributed reading","limits":"Text as retrieved on October 24, 2026; originally published January 7, 2026. Mollick’s account describes demonstrations and his assessment of the tools; it does not establish how reliably they perform across other tasks or users. The article says the website contained fake marketing claims; it does not establish that any customer purchased the product.","speaker":{"name":"One Useful Thing","roleAtSource":"Publisher"},"source":{"title":"Claude Code and What Comes Next","publisher":"One Useful Thing","format":"Article","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","publishedAt":"2026-01-07T23:00:10.000Z","dateBasis":"Date supplied by the publisher’s feed."},"evidence":[{"text":"Ethan Mollick says Claude Code built and deployed a website after a single request, but the site included fake marketing claims.","excerpt":"creating hundreds of code files and prompts. And then it gave me a single file to run that created and deployed a working website (filled with very sketchy fake marketing claims)","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"},{"text":"Mollick attributes the tools’ capabilities to advances in autonomous work and to an “agentic harness” of tools and approaches.","excerpt":"First, the latest AIs are capable of doing far more work autonomously while self-correcting many of their errors, especially in programming tasks. Second, the AIs are being given an “agentic harness”","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"},{"text":"Mollick says Claude Code compacts its conversation into notes when its context window fills, then resumes from those notes.","excerpt":"When it runs out of context, it stops and “compacts” the conversation so far, taking notes about exactly where it was when it stopped.","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"},{"text":"Mollick warns that access to a user's computer and browser creates risks, including file deletion, unintended code execution and access to sensitive data.","excerpt":"The AI might delete files it shouldn't, execute code with unintended consequences, or access sensitive data in your browser.","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"},{"text":"Mollick argues that people beyond programmers should experiment with these tools, and that new harnesses for other knowledge tasks are coming.","excerpt":"If you're programming-adjacent (an academic who works with data, a designer who wants to experiment with code, anyone who wants to try building a thing they are imagining) this is your moment to experiment.","url":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","relationship":"Source-stated"}],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","checkedAt":null,"items":[]},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What kinds of knowledge work can these tools sustain when adapted beyond programming, and what risks will that create for people using them?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Ethan Mollick argues in One Useful Thing that AI tools can sustain real work","summary":"Ethan Mollick’s article, published January 7, 2026, argues that Claude Code’s combination of a capable AI model and an “agentic harness” lets it carry out sustained work with limited input. In his demonstration, it built and deployed a website, then used a browser to test it; Mollick notes that the site contained “very sketchy fake marketing claims.” He describes tools including Skills, which provide instructions and tools for tasks, and subagents, which handle specialized work. He argues that these approaches help address limits such as a context window filling up: Claude Code can compact its conversation into notes, then continue from them. The tool can also access files and a browser, which Mollick says creates risks including deleting files, unintended code execution and exposure of sensitive data. His conclusion is that programmers should explore these tools, while people who work with data, design or ideas for things to build should experiment too. More broadly, he argues that AI with the right harness can do sustained work that matters, and that new harnesses for other knowledge tasks are coming. The article leaves open how these tools will change work beyond programming.","question":"What kinds of knowledge work can these tools sustain when adapted beyond programming, and what risks will that create for people using them?","claims":[{"text":"Ethan Mollick says Claude Code built and deployed a website after a single request, but the site included fake marketing claims.","excerpt":"creating hundreds of code files and prompts. And then it gave me a single file to run that created and deployed a working website (filled with very sketchy fake marketing claims)","support":"source-stated"},{"text":"Mollick attributes the tools’ capabilities to advances in autonomous work and to an “agentic harness” of tools and approaches.","excerpt":"First, the latest AIs are capable of doing far more work autonomously while self-correcting many of their errors, especially in programming tasks. Second, the AIs are being given an “agentic harness”","support":"source-stated"},{"text":"Mollick says Claude Code compacts its conversation into notes when its context window fills, then resumes from those notes.","excerpt":"When it runs out of context, it stops and “compacts” the conversation so far, taking notes about exactly where it was when it stopped.","support":"source-stated"},{"text":"Mollick warns that access to a user's computer and browser creates risks, including file deletion, unintended code execution and access to sensitive data.","excerpt":"The AI might delete files it shouldn't, execute code with unintended consequences, or access sensitive data in your browser.","support":"source-stated"},{"text":"Mollick argues that people beyond programmers should experiment with these tools, and that new harnesses for other knowledge tasks are coming.","excerpt":"If you're programming-adjacent (an academic who works with data, a designer who wants to experiment with code, anyone who wants to try building a thing they are imagining) this is your moment to experiment.","support":"source-stated"}],"limitations":["Text as retrieved on October 24, 2026; originally published January 7, 2026.","Mollick’s account describes demonstrations and his assessment of the tools; it does not establish how reliably they perform across other tasks or users.","The article says the website contained fake marketing claims; it does not establish that any customer purchased the product."],"mediaId":null,"correctionReason":null,"id":"9ff09b590acf347204deada6fc5b61dfb16e537adc47c7595f58799f4bfb8aac","publicationId":"f49da133-ac89-2252-207b-a087a68d72fc","version":1,"sourceUrl":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","sourceTitle":"Claude Code and What Comes Next","publisher":"One Useful Thing","publishedAt":"2026-01-07T23:00:10.000Z","originalPublished":{"value":"2026-01-07T23:00:10.000Z","precision":"instant","basis":"backfill:feed_published"},"firstObservedAt":"2026-09-23T05:41:18.181Z","createdAt":"2026-11-09T09:33:24.488Z","historical":true,"entityIds":["one-useful-thing"],"topicIds":["artificial-general-intelligence"],"media":[],"sourceMedia":{"status":"unchecked","sourceUrl":"https://www.oneusefulthing.org/p/claude-code-and-what-comes-next","checkedAt":null,"items":[]},"editorialDecision":{"outcome":"accept","mediaId":null,"inputHash":"e5d0d0e47aa4d7fe531842c94c988c055b91d6431ea28159a4cccff16dde94b0","rubricVersion":"unrecorded","provider":"unrecorded","model":"anthropic/claude-opus-5.5","reasonCodes":["bridge_admitted_by_model","second_opinion_categorical"],"confidence":1,"decidedAt":"2026-10-24T07:51:42.482Z"},"provenance":{"textHash":"f8f3bef4870be94d2b54f9b9d8aff5a814bf5de2c970ded6262a6c418a40e718","observationId":"7ae219803cf3b656a6f92b0fffa26d31527eb3a655cc37788385d0080f9ae84e","retrievedAt":"2026-10-24T07:53:38.044Z","kind":"publisher-page","preparer":{"model":"openai/gpt-6-luna","runId":"6a808321-377f-f91b-448e-28c9a1e4b277"},"reviewer":{"model":"anthropic/claude-opus-5.5","runId":"e45ae6c4-5b4a-13ea-5401-0673c8fd1cba"}},"audit":{"status":"passed","reviewedAt":"2026-11-09T09:28:38.014Z","model":"anthropic/claude-opus-5.5"}}}],"audits":[{"publicationId":"f49da133-ac89-2252-207b-a087a68d72fc","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading correctly attributes the article to Ethan Mollick in One Useful Thing. The evidence is a single authored essay, not a listing. Claims are framed as Mollick's account and arguments.","excerpt":"Claude Code and What Comes Next\nWith the right tools, AI can accomplish impressive things\nEthan Mollick"},{"dimension":"summary","status":"pass","explanation":"The summary leads with the actor and his argument. It explains Skills, subagents, compaction and the context window in plain language, and the definitions cover agentic and context window. It keeps his core demonstration, the fake-claims caveat, the risks and the forward-looking argument. The omitted $1,000 suspicion and Karpathy quote do not distort the meaning. The title and question follow from the evidence.","excerpt":"with the right harness, today's AIs are capable of real, sustained work that actually matters, and that, in turn, is starting to change how we approach tasks."},{"dimension":"date","status":"pass","explanation":"The reading gives January 7, 2026, which matches the source dateline and the UTC day of originalPublished (2026-01-07T23:00:10Z). modifiedAt equals the publication instant. The evidence is a publisher page, not an archive copy. The prediction about new harnesses is reported as Mollick's argument, not as an outcome, and there is no hindsight framing.","excerpt":"Ethan Mollick\nJan 07, 2026"},{"dimension":"support","status":"pass","explanation":"Each claim's excerpt appears verbatim and supports it: the deployed website with fake marketing claims, the two advances, compaction, the access risks, and the invitation to experiment. Mollick says he made a single request and the AI interviewed him, which is consistent with the reading. The limitation about no purchases fits the source, which says he removed the sales link. The question is open, not leading.","excerpt":"You can actually see the site it launched here, though I removed the sales link, which did actually work and would have collected money."},{"dimension":"media","status":"pass","explanation":"No media was selected, and no verified media candidates were supplied in the evidence, so there is no media identity or relevance issue.","excerpt":null}],"createdAt":"2026-11-09T09:28:38.014Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}