--- title: Chat and citations won't save your vertical AI - Atul... description: Chat interfaces and citations are insufficient for vertical AI products in regulated industries like healthcare, legal, and taxes. The core argument is that AI... canonical: https://daily.dev/posts/chat-and-citations-won-t-save-your-vertical-ai---atul-ramachandran-filed-inc-aocbvvhgb twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Chat and citations won't save your vertical AI - Atul Ramachandran, Filed Inc | daily.dev og:description: Chat interfaces and citations are insufficient for vertical AI products in regulated industries like healthcare, legal, and taxes. The core argument is that AI... og:url: https://daily.dev/posts/chat-and-citations-won-t-save-your-vertical-ai---atul-ramachandran-filed-inc-aocbvvhgb og:image: https://api.daily.dev/og/posts/AOcBVVhgb.png og:image:alt: Chat and citations won't save your vertical AI - Atul Ramachandran, Filed Inc og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Chat and citations won't save your vertical AI - Atul Ramachandran, Filed Inc **[AI Engineer](https://daily.dev/sources/aidotengineer)** · 15 min read · 1 upvotes · 0 comments ## Summary Chat interfaces and citations are insufficient for vertical AI products in regulated industries like healthcare, legal, and taxes. The core argument is that AI products must shift from participation-based design to delegation-based design, treating the product as a conveyor belt where users are supervisors rather than operators. Four key components are needed: identifying delegatable long-running tasks, teaching agents user-specific skills, monitoring agent progress with traces, and giving users control to pause and intervene. A new success metric is proposed — weekly active sessions instead of weekly active users — where agent-completed tasks count even when users are offline. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why do chat and citations fail as the primary interface for vertical AI agent products? Chat is synchronous, forcing users to stay engaged and wait for responses rather than walking away while the agent works, which undermines the promise of agents working while users sleep. Citations shift the verification burden back onto users, who must review outputs one by one, adding extra work especially in high-stakes fields like healthcare, legal, and tax. _Anyone designing agentic UX beyond chat can track emerging patterns for delegation-first products on daily.dev._ ### What metric should replace weekly active users for products built around AI agent delegation? Weekly active sessions is proposed as the better metric, counting a task completed by a human or an agent even when the user isn't on the platform. The goal is for weekly active users to decline while weekly active sessions rise, signaling that customers trust the platform enough to delegate tasks that execute without direct human involvement. _Teams rethinking product metrics for agentic delegation can follow this shift in measurement on daily.dev._ ### What are the four key components needed to build an agentic product according to the conveyor belt model? The four components are: finding tasks worth delegating that take users more than an hour or two to do themselves, teaching agents user-specific skills to capture the last 20% of quirks and preferences, monitoring long-running work through task lists and traces, and giving users control to intervene and pause work when something goes wrong or requires judgment. _Builders designing agent-based workflows can compare frameworks for delegation and control on daily.dev._ ## Similar posts on daily.dev - [Disrupting yourself in the age of AI](https://daily.dev/posts/disrupting-yourself-in-the-age-of-ai-rhh9pc9cu) · Stack Overflow Blog · 0 upvotes · 0 comments - [Vertical AI Agent: What It Is and How to Build One That Works](https://daily.dev/posts/vertical-ai-agent-what-it-is-and-how-to-build-one-that-works-p7iggnxv1) · C\# Corner · 1 upvotes · 0 comments - [Clickatell: Agentic AI turns automation into consequence](https://daily.dev/posts/clickatell-agentic-ai-turns-automation-into-consequence-ooqr2j9wr) · TechCentral · 0 upvotes · 0 comments - [What Would A Serious AI Product Look Like?](https://daily.dev/posts/what-would-a-serious-ai-product-look-like--o9f18yuws) · Planet Python · 0 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#ux](https://daily.dev/tags/ux) [View this post on daily.dev](https://daily.dev/posts/chat-and-citations-won-t-save-your-vertical-ai---atul-ramachandran-filed-inc-aocbvvhgb) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Chat and citations won't save your vertical AI - Atul Ramachandran, Filed Inc","url":"https://daily.dev/posts/chat-and-citations-won-t-save-your-vertical-ai---atul-ramachandran-filed-inc-aocbvvhgb","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/chat-and-citations-won-t-save-your-vertical-ai---atul-ramachandran-filed-inc-aocbvvhgb"},"datePublished":"2026-07-11T17:33:07.454Z","dateModified":"2026-09-14T08:27:51.284Z","description":"Chat interfaces and citations are insufficient for vertical AI products in regulated industries like healthcare, legal, and taxes. The core argument is that AI...","image":"https://i.ytimg.com/vi/RGiXcVxSD3s/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/RGiXcVxSD3s/sddefault.jpg","isAccessibleForFree":true,"articleSection":"AI Engineer","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"AI Engineer","logo":"https://media.daily.dev/image/upload/s--u5PucxNT--/f_auto/v1724338940/logos/aidotengineer","url":"https://daily.dev/sources/aidotengineer"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/chat-and-citations-won-t-save-your-vertical-ai---atul-ramachandran-filed-inc-aocbvvhgb","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,ux","timeRequired":"PT15M","video":{"@type":"VideoObject","name":"Chat and citations won't save your vertical AI - Atul Ramachandran, Filed Inc","description":"Chat interfaces and citations are insufficient for vertical AI products in regulated industries like healthcare, legal, and taxes. The core argument is that AI...","thumbnailUrl":"https://i.ytimg.com/vi/RGiXcVxSD3s/sddefault.jpg","uploadDate":"2026-07-11T17:33:07.454Z","duration":"PT15M","url":"https://api.daily.dev/r/AOcBVVhgb","embedUrl":"https://www.youtube.com/embed/RGiXcVxSD3s"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"AI Engineer","item":"https://daily.dev/sources/aidotengineer"},{"@type":"ListItem","position":3,"name":"Chat and citations won't save your vertical AI - Atul Ramachandran, Filed Inc"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/chat-and-citations-won-t-save-your-vertical-ai---atul-ramachandran-filed-inc-aocbvvhgb#faq","mainEntity":[{"@type":"Question","name":"Why do chat and citations fail as the primary interface for vertical AI agent products?","acceptedAnswer":{"@type":"Answer","text":"Chat is synchronous, forcing users to stay engaged and wait for responses rather than walking away while the agent works, which undermines the promise of agents working while users sleep. Citations shift the verification burden back onto users, who must review outputs one by one, adding extra work especially in high-stakes fields like healthcare, legal, and tax. Anyone designing agentic UX beyond chat can track emerging patterns for delegation-first products on daily.dev."}},{"@type":"Question","name":"What metric should replace weekly active users for products built around AI agent delegation?","acceptedAnswer":{"@type":"Answer","text":"Weekly active sessions is proposed as the better metric, counting a task completed by a human or an agent even when the user isn't on the platform. The goal is for weekly active users to decline while weekly active sessions rise, signaling that customers trust the platform enough to delegate tasks that execute without direct human involvement. Teams rethinking product metrics for agentic delegation can follow this shift in measurement on daily.dev."}},{"@type":"Question","name":"What are the four key components needed to build an agentic product according to the conveyor belt model?","acceptedAnswer":{"@type":"Answer","text":"The four components are: finding tasks worth delegating that take users more than an hour or two to do themselves, teaching agents user-specific skills to capture the last 20% of quirks and preferences, monitoring long-running work through task lists and traces, and giving users control to intervene and pause work when something goes wrong or requires judgment. Builders designing agent-based workflows can compare frameworks for delegation and control on daily.dev."}}]} ```