{"slug":"ai-research-desks-2027","title":"The AI research desk is becoming a new layer of the internet","description":"A source-backed view of how research, publishing, and AI discovery are converging into one workflow.","summary":"The durable opportunity is not automated prose. It is a verifiable research layer that makes every claim easy for people and AI systems to inspect.","publishedAt":"2026-08-28","updatedAt":"2026-09-01","tags":["AI","Future of work","Publishing"],"claims":[{"id":"c1","text":"Inference cost is becoming a distribution advantage, not only a model advantage.","status":"Supported","sourceIds":["s1","s2"]},{"id":"c2","text":"The next durable moat will be proprietary workflow data and feedback loops.","status":"Needs review","sourceIds":["s1"]},{"id":"c3","text":"Open models may compress margins faster than enterprise adoption expands them.","status":"Open question","sourceIds":["s3","s2"]}],"sources":[{"id":"s1","title":"The State of AI Infrastructure, 2026","publisher":"Atlas Research","date":"2026-08-18","type":"Report","stance":"Primary","quality":94,"url":"https://example.com/atlas-ai-infrastructure","note":"Original survey of 412 technical leaders."},{"id":"s2","title":"Why the AI spending curve is flattening","publisher":"The Ledger","date":"2026-08-21","type":"News","stance":"Independent","quality":82,"url":"https://example.com/ledger-spending","note":"Interview-led analysis with public earnings data."},{"id":"s3","title":"Open models and the cost of intelligence","publisher":"Open Systems Journal","date":"2026-08-12","type":"Interview","stance":"Contrarian","quality":78,"url":"https://example.com/open-models","note":"Counterpoint from open-model maintainers."}],"formats":{"html":"/article/ai-research-desks-2027","markdown":"/article/ai-research-desks-2027.md","json":"/article/ai-research-desks-2027.json"}}