[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"categories-en":3,"post:\u002Fpost\u002Fconversational-ai-exhausted-how-to-migrate-to-agentic-workflows-and-execute-real-actions":37},[4,11,17,24,30],{"id":5,"slug":6,"name":7,"colorHex":8,"iconName":9,"postCount":10},"0e2c1225-f78a-4c20-8c93-998ab9dc93b9","frontend","Frontend","#40484f",null,25,{"id":12,"slug":13,"name":14,"colorHex":15,"iconName":9,"postCount":16},"bc8df70e-b7fb-4d89-a80c-fd458d36d58c","code-development","Code & Development","#8b5cf6",23,{"id":18,"slug":19,"name":20,"colorHex":21,"iconName":22,"postCount":23},"550e8400-e29b-41d4-a716-446655440001","inteligencia-artificial","Artificial Intelligence","#6366f1","brain",14,{"id":25,"slug":26,"name":27,"colorHex":28,"iconName":9,"postCount":29},"ea63b69c-2eb4-4100-81f3-70ce52d33743","tech-news","Tech News","#e50606",4,{"id":31,"slug":32,"name":33,"colorHex":34,"iconName":35,"postCount":36},"550e8400-e29b-41d4-a716-446655440017","how-to-guides","How-To & Guides","#be123c","book-open",2,{"post":38,"related":557,"translations":622,"resolved":9},{"id":39,"slug":40,"title":41,"excerpt":42,"image":43,"featured":44,"groupId":45,"publishedAt":46,"readingTime":29,"views":47,"author":48,"category":50,"tags":51,"contentJson":61,"metaTitle":554,"metaDescription":555,"updatedAt":556},"99d8b7bf-03aa-4c01-8bc7-ef1aca804b30","conversational-ai-exhausted-how-to-migrate-to-agentic-workflows-and-execute-real-actions","Conversational AI Exhausted? How to Migrate to Agentic Workflows and Execute Real Actions","The chat paradigm is exhausted. Discover agentic workflows: systems that plan, execute, and verify tasks autonomously and in a testable manner.","https:\u002F\u002Fmfgiomkkbihnlpfdxwrt.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fblog-assets\u002Fmedia\u002Ffeatured\u002F1773194326109-featured-g0e36z.jpeg",false,"2438fae0-5c5e-41ce-9543-89007a6888cc","2026-07-08T23:42:19.403576+00:00",0,{"name":49,"avatar":9},"Genildo Souza",{"slug":19,"name":20,"color":21},[52,55,58],{"slug":53,"name":54},"ai","AI",{"slug":56,"name":57},"llm","LLM",{"slug":59,"name":60},"openclaw","OpenClaw",{"type":62,"content":63},"doc",[64,75,83,90,101,111,127,133,140,271,277,354,360,368,381,440,474,482,516,524,548],{"type":65,"attrs":66},"summaryBox",{"id":67,"items":68,"title":74},"sum_d5mr93oi",[69,70,71,72,73],"The chat paradigm has reached its limit and must be replaced by agentic workflows focused on execution and verification.","Complex prompts are fragile and difficult to maintain; it is preferable to structure testable and monitorable workflows.","The transition requires AI systems to adopt Intent, Plan, Action, Verify, and Memory cycles to perform end-to-end tasks.","Developers must prioritize the integration of tools and APIs rather than focusing solely on the quality of textual responses.","Agent observability must focus on execution correctness and loop monitoring, going beyond token metrics.","Migrating from Conversational AI to Agents",{"type":76,"attrs":77,"content":79},"paragraph",{"id":78},"p_n48e9vbt",[80],{"text":81,"type":82},"If you feel your prompts are becoming a fragile, hard-to-maintain glue, you are not alone: the chat paradigm has reached its limit. Jason Cochran, from Strataga\u002FOpenClaw, argues that the next cycle of AI products will be defined by agentic workflows, capable of planning, executing, and verifying actions in real systems, moving beyond simple command-response.","text",{"type":84,"attrs":85,"content":87},"heading",{"id":86,"level":36},"h_ljffwaiz",[88],{"text":89,"type":82},"What has changed",{"type":76,"attrs":91,"content":93},{"id":92},"p_llyhfs58",[94,99],{"text":95,"type":82,"marks":96},"From chat to execution.",[97],{"type":98},"bold",{"text":100,"type":82}," For two years, most AI products followed the same pattern: user types → model returns text → user decides what to do with the output. This model, according to Cochran, has reached its leverage limit.",{"type":76,"attrs":102,"content":104},{"id":103},"p_9hag6u8l",[105,109],{"text":106,"type":82,"marks":107},"Prompts have become \"fragile glue\".",[108],{"type":98},{"text":110,"type":82}," Teams trying to add increasingly complex behavior into prompts discover that the result is fragile code—difficult to test, monitor, or maintain.",{"type":76,"attrs":112,"content":114},{"id":113},"p_0rfj7vpo",[115,119,121,125],{"text":116,"type":82,"marks":117},"The agentic loop replaces one-shot.",[118],{"type":98},{"text":120,"type":82}," The fundamental shift is adding a cycle: ",{"text":122,"type":82,"marks":123},"Intent → Plan → Action → Verify → Memory",[124],{"type":98},{"text":126,"type":82},". The system doesn't just respond; it executes steps, calls tools, verifies results, and learns from the cycle.",{"type":84,"attrs":128,"content":130},{"id":129,"level":36},"h_w5zx6n2u",[131],{"text":132,"type":82},"Technical context and practical impact",{"type":84,"attrs":134,"content":137},{"id":135,"level":136},"h_ff1ifb0x",3,[138],{"text":139,"type":82},"Three emerging patterns",{"type":141,"attrs":142,"content":144},"table",{"id":143},"tbl_36t6qnxx",[145,177,209,240],{"type":146,"content":147},"tableRow",[148,159,168],{"type":149,"attrs":150,"content":152},"tableHeader",{"align":9,"colspan":151,"rowspan":151,"colwidth":9},1,[153],{"type":76,"attrs":154,"content":156},{"id":155},"p_59xigiux",[157],{"text":158,"type":82},"Pattern",{"type":149,"attrs":160,"content":161},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[162],{"type":76,"attrs":163,"content":165},{"id":164},"p_swwhx7cz",[166],{"text":167,"type":82},"Description",{"type":149,"attrs":169,"content":170},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[171],{"type":76,"attrs":172,"content":174},{"id":173},"p_p2mc2603",[175],{"text":176,"type":82},"Implication",{"type":146,"content":178},[179,191,200],{"type":180,"attrs":181,"content":182},"tableCell",{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[183],{"type":76,"attrs":184,"content":186},{"id":185},"p_9c3ae4sf",[187],{"text":188,"type":82,"marks":189},"Parallel execution",[190],{"type":98},{"type":180,"attrs":192,"content":193},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[194],{"type":76,"attrs":195,"content":197},{"id":196},"p_yqqt93w4",[198],{"text":199,"type":82},"Multiple agents working in parallel",{"type":180,"attrs":201,"content":202},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[203],{"type":76,"attrs":204,"content":206},{"id":205},"p_gfpre6pv",[207],{"text":208,"type":82},"One agent is a line; a team of agents is a force multiplier",{"type":146,"content":210},[211,222,231],{"type":180,"attrs":212,"content":213},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[214],{"type":76,"attrs":215,"content":217},{"id":216},"p_90pifomt",[218],{"text":219,"type":82,"marks":220},"Tools over text",[221],{"type":98},{"type":180,"attrs":223,"content":224},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[225],{"type":76,"attrs":226,"content":228},{"id":227},"p_r337htfh",[229],{"text":230,"type":82},"Agents with defined tool schemas",{"type":180,"attrs":232,"content":233},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[234],{"type":76,"attrs":235,"content":237},{"id":236},"p_7o0tdq54",[238],{"text":239,"type":82},"Reliability vs. breakage when the agent needs to \"guess\" the UI",{"type":146,"content":241},[242,253,262],{"type":180,"attrs":243,"content":244},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[245],{"type":76,"attrs":246,"content":248},{"id":247},"p_w3rr3e7r",[249],{"text":250,"type":82,"marks":251},"Always on automation",[252],{"type":98},{"type":180,"attrs":254,"content":255},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[256],{"type":76,"attrs":257,"content":259},{"id":258},"p_nelinvk2",[260],{"text":261,"type":82},"Hosted bots that run 24\u002F7",{"type":180,"attrs":263,"content":264},{"align":9,"colspan":151,"rowspan":151,"colwidth":9},[265],{"type":76,"attrs":266,"content":268},{"id":267},"p_24vn3ssb",[269],{"text":270,"type":82},"Automation that doesn't die when the laptop sleeps",{"type":84,"attrs":272,"content":274},{"id":273,"level":136},"h_owtmxdu7",[275],{"text":276,"type":82},"Why workflows outperform prompts",{"type":278,"attrs":279,"content":282},"bulletList",{"id":280,"tight":281},"ul_14zso443",true,[283,298,312,326,340],{"type":284,"attrs":285,"content":287},"listItem",{"id":286},"li_yy248h79",[288],{"type":76,"attrs":289,"content":291},{"id":290},"p_q097zut0",[292,296],{"text":293,"type":82,"marks":294},"Defined steps",[295],{"type":98},{"text":297,"type":82},": Explicit flow, not ambiguity",{"type":284,"attrs":299,"content":301},{"id":300},"li_tx0n2c3h",[302],{"type":76,"attrs":303,"content":305},{"id":304},"p_9c617tyd",[306,310],{"text":307,"type":82,"marks":308},"Real tool calling",[309],{"type":98},{"text":311,"type":82},": Integration with APIs, file systems, external services",{"type":284,"attrs":313,"content":315},{"id":314},"li_wg56hka9",[316],{"type":76,"attrs":317,"content":319},{"id":318},"p_bhmhqmzz",[320,324],{"text":321,"type":82,"marks":322},"Testable",[323],{"type":98},{"text":325,"type":82},": Can be validated like traditional code",{"type":284,"attrs":327,"content":329},{"id":328},"li_x02kuwzx",[330],{"type":76,"attrs":331,"content":333},{"id":332},"p_ekcv5t0j",[334,338],{"text":335,"type":82,"marks":336},"Monitorable",[337],{"type":98},{"text":339,"type":82},": Execution metrics, not just response quality",{"type":284,"attrs":341,"content":343},{"id":342},"li_fo3lpstw",[344],{"type":76,"attrs":345,"content":347},{"id":346},"p_27gjsif7",[348,352],{"text":349,"type":82,"marks":350},"Repeatable",[351],{"type":98},{"text":353,"type":82},": The same workflow runs consistently",{"type":84,"attrs":355,"content":357},{"id":356,"level":36},"h_kdnjlem8",[358],{"text":359,"type":82},"Implications for developers",{"type":76,"attrs":361,"content":363},{"id":362},"p_rpsqc28w",[364],{"text":365,"type":82,"marks":366},"For those building products with AI:",[367],{"type":98},{"type":76,"attrs":369,"content":371},{"id":370},"p_f8tjx1pv",[372,374,379],{"text":373,"type":82},"The right question is no longer ",{"text":375,"type":82,"marks":376},"\"how do we add AI chat?\"",[377],{"type":378},"italic",{"text":380,"type":82}," but rather:",{"type":382,"attrs":383,"content":385},"orderedList",{"id":384,"type":9,"start":151,"tight":281},"ol_fjnlri6g",[386,400,414,428],{"type":284,"attrs":387,"content":389},{"id":388},"li_mgcq6t34",[390],{"type":76,"attrs":391,"content":393},{"id":392},"p_5n6arhkj",[394,398],{"text":395,"type":82,"marks":396},"What work should the system do end-to-end?",[397],{"type":98},{"text":399,"type":82}," (not \"what response should it give\")",{"type":284,"attrs":401,"content":403},{"id":402},"li_ez9adz8c",[404],{"type":76,"attrs":405,"content":407},{"id":406},"p_tcxwpeia",[408,412],{"text":409,"type":82,"marks":410},"What tools does it need to access?",[411],{"type":98},{"text":413,"type":82}," (not \"what prompt should I write\")",{"type":284,"attrs":415,"content":417},{"id":416},"li_1zo1d6q8",[418],{"type":76,"attrs":419,"content":421},{"id":420},"p_cx2drdxn",[422,426],{"text":423,"type":82,"marks":424},"What constraints keep it safe?",[425],{"type":98},{"text":427,"type":82}," (guardrails, not safety prompts)",{"type":284,"attrs":429,"content":431},{"id":430},"li_rn870xhb",[432],{"type":76,"attrs":433,"content":435},{"id":434},"p_m37wp9py",[436],{"text":437,"type":82,"marks":438},"How do we verify that the work was done?",[439],{"type":98},{"type":441,"attrs":442},"quiz",{"id":443,"title":444,"questions":445},"quiz_aksh4kdo","Quiz: Key Questions in the New AI Paradigm",[446,457,465],{"options":447,"question":456},[448,450,453],{"text":395,"isCorrect":281,"rationale":449},"The focus has shifted from 'what response to give' to the complete work the system must execute.",{"text":451,"isCorrect":44,"rationale":452},"How do we add AI chat?","This question belongs to the old paradigm, focused only on the chat interface.",{"text":454,"isCorrect":44,"rationale":455},"What response should the system provide?","The new paradigm prioritizes the complete work, not just the final response.","What is the correct question to define the system's work in the new paradigm?",{"options":458,"question":464},[459,461],{"text":409,"isCorrect":281,"rationale":460},"It is essential to define which tools the system must use to execute the work.",{"text":462,"isCorrect":44,"rationale":463},"What prompt to write for the system?","The focus is not on isolated prompts, but on access to the necessary tools.","In the new paradigm, what should be considered regarding system tools?",{"options":466,"question":473},[467,470],{"text":468,"isCorrect":281,"rationale":469},"What constraints keep it safe (guardrails)","The new paradigm emphasizes structural constraints, not just safety prompts.",{"text":471,"isCorrect":44,"rationale":472},"Creating safety prompts","Safety prompts are less effective than guardrails for security in the new model.","What is the focus for keeping the system safe?",{"type":76,"attrs":475,"content":477},{"id":476},"p_uzx8l7he",[478],{"text":479,"type":82,"marks":480},"For engineering teams:",[481],{"type":98},{"type":278,"attrs":483,"content":485},{"id":484,"tight":281},"ul_775dqt5c",[486,496,506],{"type":284,"attrs":487,"content":489},{"id":488},"li_dvjvv0cj",[490],{"type":76,"attrs":491,"content":493},{"id":492},"p_8dgjgge8",[494],{"text":495,"type":82},"Prompts as \"glue\" between features will lose ground to structured workflow code",{"type":284,"attrs":497,"content":499},{"id":498},"li_mw4jsctp",[500],{"type":76,"attrs":501,"content":503},{"id":502},"p_hxsftptp",[504],{"text":505,"type":82},"AI testing will need to cover not just response quality, but execution correctness",{"type":284,"attrs":507,"content":509},{"id":508},"li_sx4l4twt",[510],{"type":76,"attrs":511,"content":513},{"id":512},"p_2v83om83",[514],{"text":515,"type":82},"Agent monitoring requires loop observability, not just token observability",{"type":84,"attrs":517,"content":519},{"id":518,"level":36},"h_do0wtb6r",[520],{"text":521,"type":82,"marks":522},"Conclusion",[523],{"type":98},{"type":76,"attrs":525,"content":527},{"id":526},"p_5y1kcrkj",[528,530,534,536,546],{"text":529,"type":82},"Cochran's article reflects a trend already manifesting in other parts of the ecosystem: the acquisition of Promptfoo by OpenAI (February 2026) indicates that ",{"text":531,"type":82,"marks":532},"agent security and testability",[533],{"type":98},{"text":535,"type":82}," are recognized bottlenecks. The ",{"text":537,"type":82,"marks":538},"Anthropic Sonnet 4.6",[539],{"type":540,"attrs":541},"link",{"rel":542,"href":543,"class":544,"title":9,"target":545},"noopener noreferrer nofollow","https:\u002F\u002Fwww.anthropic.com\u002Fclaude\u002Fsonnet","text-cyan-400 hover:underline cursor-pointer","_blank",{"text":547,"type":82},", with advanced \"Computer Use\" capabilities, points in the same direction—models that do not just generate text, but execute actions.",{"type":76,"attrs":549,"content":551},{"id":550},"p_8pn9e4h7",[552],{"text":553,"type":82},"The change is conceptually simple: from \"AI that responds\" to \"AI that does.\" In practice, it requires a deep rethink of architecture, tooling, and engineering mindset. The market that masters agentic execution will be the next major infrastructure ecosystem.","Conversational AI Exhausted? The Future is Agentic Workflows","Conversational AI has reached its limit. Understand why agentic workflows, which plan and execute real actions, are the future of development.","2026-08-24T14:53:53.315534+00:00",[558,581,595,609],{"id":559,"slug":560,"title":561,"excerpt":562,"image":563,"featured":44,"groupId":564,"publishedAt":565,"readingTime":566,"views":47,"author":567,"category":568,"tags":569},"10ee04ce-fa53-44d7-9603-2e375733d313","the-rise-of-shadow-deployment-and-the-dna-behind-the-hidden-ox-alpha-model","The Rise of Shadow Deployment and the DNA Behind the Hidden Ox Alpha Model","The ghost launch of Ox Alpha revealed a new era in software engineering, where strategic anonymity redefines the future of AI models.","https:\u002F\u002Fmfgiomkkbihnlpfdxwrt.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fblog-assets\u002Fmedia\u002Ffeatured\u002F1787604179257-featured-byxw99.webp","ba241d77-9b03-4c87-8e52-ba727a888e59","2026-08-24T22:56:11.386568+00:00",7,{"name":49,"avatar":9},{"slug":19,"name":20,"color":21},[570,573,574,577,578],{"slug":571,"name":572},"china","China",{"slug":56,"name":57},{"slug":575,"name":576},"infraestrutura","Infraestrutura",{"slug":53,"name":54},{"slug":579,"name":580},"performance","Performance",{"id":582,"slug":583,"title":584,"excerpt":585,"image":586,"featured":44,"groupId":587,"publishedAt":588,"readingTime":589,"views":47,"author":590,"category":591,"tags":592},"011ee3e7-61c8-4877-a99b-edc119022f4d","glm-52-vs-kimi-k27-why-glm-wins-the-code-reliability-test","GLM-5.2 vs. Kimi K2.7: Why GLM Wins the Code Reliability Test","Five independent evaluators tested GLM-5.2 and Kimi K2.7 on coding tasks. Discover which model wins the battle for reliability and development speed.","https:\u002F\u002Fmfgiomkkbihnlpfdxwrt.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fblog-assets\u002Fmedia\u002Ffeatured\u002F1782256309298-featured-ubiuao.webp","8707b037-1af8-46ff-972a-99c4ead20adc","2026-07-08T22:58:35.7148+00:00",5,{"name":49,"avatar":9},{"slug":19,"name":20,"color":21},[593,594],{"slug":53,"name":54},{"slug":579,"name":580},{"id":596,"slug":597,"title":598,"excerpt":599,"image":600,"featured":44,"groupId":601,"publishedAt":602,"readingTime":603,"views":47,"author":604,"category":605,"tags":606},"e595e2c6-e464-46d7-ba67-810f8d0cd234","how-to-choose-an-ai-sdk-why-the-fear-of-lock-in-is-a-mistake-and-how-to-decide-based-on-your-apps-format","How to choose an AI SDK: why the fear of lock-in is a mistake and how to decide based on your app's format","The fear of being locked into an AI provider is a common mistake. Learn how to choose the best SDK based on your project's format and productivity.","https:\u002F\u002Fmfgiomkkbihnlpfdxwrt.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fblog-assets\u002Fmedia\u002Ffeatured\u002F1782794440171-featured-aqj363.webp","7839c171-0b65-418a-87dc-7ded7786a84a","2026-07-04T19:26:16.054373+00:00",6,{"name":49,"avatar":9},{"slug":19,"name":20,"color":21},[607,608],{"slug":53,"name":54},{"slug":56,"name":57},{"id":610,"slug":611,"title":612,"excerpt":613,"image":614,"featured":44,"groupId":615,"publishedAt":616,"readingTime":136,"views":47,"author":617,"category":618,"tags":619},"d9638633-92f1-4ee3-88da-001b4e599233","moving-ai-agents-to-production-why-capability-isnt-enough-anymore","Moving AI Agents to Production: Why Capability Isn't Enough Anymore","The challenge for AI agents has shifted from proving capability to ensuring reliability, governance, and cost-efficiency in real-world production.","https:\u002F\u002Fmfgiomkkbihnlpfdxwrt.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fblog-assets\u002Fmedia\u002Ffeatured\u002F1773890753958-featured-dwbmox.jpeg","6edadcfc-9020-4959-86e3-b8ccb17ab5b8","2026-06-23T23:35:06.18764+00:00",{"name":49,"avatar":9},{"slug":19,"name":20,"color":21},[620,621],{"slug":53,"name":54},{"slug":56,"name":57},{"pt":623,"en":40},"ia-conversacional-esgotada-como-migrar-para-workflows-agentic-e-executar-acoes-reais"]