{"id":22770,"date":"2026-05-12T15:50:08","date_gmt":"2026-05-12T15:50:08","guid":{"rendered":"https:\/\/ideainthebox.com\/index.php\/2026\/05\/12\/jbs-dev-on-imperfect-data-and-the-ai-last-mile-from-model-capability-to-cost-sustainability\/"},"modified":"2026-05-12T15:50:08","modified_gmt":"2026-05-12T15:50:08","slug":"jbs-dev-on-imperfect-data-and-the-ai-last-mile-from-model-capability-to-cost-sustainability","status":"publish","type":"post","link":"https:\/\/ideainthebox.com\/index.php\/2026\/05\/12\/jbs-dev-on-imperfect-data-and-the-ai-last-mile-from-model-capability-to-cost-sustainability\/","title":{"rendered":"JBS Dev: On imperfect data and the AI last mile \u2013 from model capability to cost sustainability"},"content":{"rendered":"<div>\n<p>Joe Rose, president at strategic technology provider <a href=\"https:\/\/www.jbs.dev\/\">JBS Dev<\/a>, wants to cut through one of the myths of working with generative and agentic AI systems. \u201cIt\u2019s a common misconception that your data has to be perfect before you do any of these types of workloads,\u201d he explains.<\/p>\n<p>As a <a href=\"https:\/\/aifieldbook.com\/insights\/data-foundation-for-generative-ai\">recent article in AI Fieldbook outlines<\/a>, vendors and consultants \u2013 not surprisingly \u2013 suggest you need huge data lakes and multi-year data transformation programmes respectively. Executives are therefore scratching their heads at it all. The reality is slightly different. \u201cThe tooling has never been better than it is now to deal with poor quality data,\u201d says Rose. \u201cIt\u2019s almost remarkable what an LLM can understand on a half-written prompt.\u201d<\/p>\n<p>It makes sense. If you\u2019ve got such a tool available, then it\u2019s worth utilising that to your advantage \u2013 with the correct guardrails in place. The inherent unpredictability of models means a need to handle bad output, which is where the human in the loop comes in. For textual or category data, there is a resilience in place. \u201cPeople are\u2026 used to \u2018we build it, it works, we forget about it,\u2019\u201d says Rose. \u201cThat\u2019s just not how these systems work.\u201d<\/p>\n<p>Regarding imperfect data, Rose gives an example of a client in the medical sector where the goal was to migrate to another billing reconciliation system. Records were a mix; some were in PDF, others an image; the procedure would sometimes be in the doctor\u2019s name, the doctor\u2019s name would be in the patient\u2019s name, and so on. The gen AI was able to scope the clean data from a simple prompt, from OCR to the images to text extraction for the PDFs, while more agentic approaches were subsequently leveraged, such as comparing a customer record to an insurance contract to see if they were billed at the right rate.<\/p>\n<p>\u201cYou start to layer different use cases on top of one another,\u201d says Rose. \u201cThat\u2019s not to say that it gets everything right \u2013 you still need a human in the loop. But what you want to do is say, \u2018we started at 20% automated, and then 40%, and then 60, 80%\u2019, and kind of grow that over time.\u201d<\/p>\n<p>Going forward, Rose expects future discussions for these models to be around cost and portability. \u201cI think you\u2019re going to see a shift away from these radical leaps and model capability, and more shift towards \u2018how do we make the cost more sustainable that we don\u2019t have to build data centres at the rate we\u2019re building data centres?\u2019,\u201d he says.<\/p>\n<p>\u201cThe last mile is \u2018how do we get these things to run on a laptop or a phone instead of having to run in a data centre?\u2019 The models were trained on a body of data \u2013 essentially every page on the internet and other stuff. It\u2019s not like there\u2019s a tonne more data that hasn\u2019t already been put into them that\u2019s going to lead to some type of breakthrough.\u201d<\/p>\n<p>At <a href=\"https:\/\/ai-expo.net\/northamerica\">AI &amp; Big Data Expo<\/a>, where JBS Dev is participating, Rose is looking forward to the conversations \u2013 and one more controversial opinion he\u2019ll put across is to tell folk to stop buying from SaaS vendors when you can do it yourself. \u201cIt\u2019s not as hard as it sounds,\u201d he says. \u201cAlmost everybody\u2019s got some kind of cloud presence, and that\u2019s where I would start, because the cloud tooling, especially for the big three\u2026 has everything you need to start implementing agentic workloads tomorrow, without new software licenses and new training.\u201d<\/p>\n<p>Once that\u2019s in place, JBS Dev is there for the next steps of the journey.<\/p>\n<p><em>Watch the full interview with Rose below:<\/em><\/p>\n<figure class=\"wp-block-video\"><video height=\"1080\" style=\"aspect-ratio: 1920 \/ 1080;\" width=\"1920\" controls src=\"https:\/\/www.artificialintelligence-news.com\/wp-content\/uploads\/2026\/05\/Joe_Rose_Jbs_Dev_07_05.mp4\"><\/video><\/figure>\n<p><em>Image by <a href=\"https:\/\/pixabay.com\/users\/geralt-9301\/?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=6511448\">Gerd Altmann<\/a> from <a href=\"https:\/\/pixabay.com\/\/?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=6511448\">Pixabay<\/a><\/em><\/p>\n<p>The post <a href=\"https:\/\/www.artificialintelligence-news.com\/news\/jbs-dev-on-imperfect-data-and-the-ai-last-mile-from-model-capability-to-cost-sustainability\/\">JBS Dev: On imperfect data and the AI last mile \u2013 from model capability to cost sustainability<\/a> appeared first on <a href=\"https:\/\/www.artificialintelligence-news.com\/\">AI News<\/a>.<\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>Joe Rose, president at strategic technology provider JBS Dev, wants  [&#8230;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"content-type":"","footnotes":""},"categories":[226],"tags":[],"class_list":["post-22770","post","type-post","status-publish","format-standard","hentry","category-technology"],"acf":[],"_links":{"self":[{"href":"https:\/\/ideainthebox.com\/index.php\/wp-json\/wp\/v2\/posts\/22770","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ideainthebox.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ideainthebox.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ideainthebox.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/ideainthebox.com\/index.php\/wp-json\/wp\/v2\/comments?post=22770"}],"version-history":[{"count":0,"href":"https:\/\/ideainthebox.com\/index.php\/wp-json\/wp\/v2\/posts\/22770\/revisions"}],"wp:attachment":[{"href":"https:\/\/ideainthebox.com\/index.php\/wp-json\/wp\/v2\/media?parent=22770"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ideainthebox.com\/index.php\/wp-json\/wp\/v2\/categories?post=22770"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ideainthebox.com\/index.php\/wp-json\/wp\/v2\/tags?post=22770"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}