--- title: Boost accuracy with the prompt optimizer | daily.dev description: A walkthrough of using Google's Vertex AI Prompt Optimizer to automatically improve a system prompt for generating YouTube video descriptions from scripts. The... canonical: https://daily.dev/posts/boost-accuracy-with-the-prompt-optimizer-ybzq2sjvb twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Boost accuracy with the prompt optimizer | daily.dev og:description: A walkthrough of using Google's Vertex AI Prompt Optimizer to automatically improve a system prompt for generating YouTube video descriptions from scripts. The... og:url: https://daily.dev/posts/boost-accuracy-with-the-prompt-optimizer-ybzq2sjvb og:image: https://api.daily.dev/og/posts/YbzQ2SjVB.png og:image:alt: Boost accuracy with the prompt optimizer 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. # Boost accuracy with the prompt optimizer **[Firebase](https://daily.dev/sources/firebase-blog)** · 13 min read · 0 upvotes · 0 comments ## Summary A walkthrough of using Google's Vertex AI Prompt Optimizer to automatically improve a system prompt for generating YouTube video descriptions from scripts. The author uploaded scripts and target descriptions to a CSV in Google Cloud Storage, then ran the data-driven optimizer with Gemini 2.5 Flash across instructions-and-demo optimization mode. The process took roughly four hours, consumed millions of tokens, cost a couple hundred dollars, and produced a much more consistent, detailed prompt that closely matched the quality of human-written descriptions, including accurate chapter timestamps and curated resource links. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How does the Vertex AI data-driven prompt optimizer improve a prompt automatically? It iterates on a system prompt using paired input and target output data, running computation-based or model-based evaluation metrics (via Gemini) to grade each variant against a baseline generated from the original prompt. Inputs and target outputs are uploaded as a CSV to Google Cloud Storage, then processed through a Colab Enterprise notebook that tests optimization modes like instructions-and-demo to find the best-performing version. _See how teams track prompt-tuning workflows like this one on daily.dev before automating content pipelines._ ### How long and how much does it cost to run the Vertex AI prompt optimizer on real data? For 45 input prompt and description pairs, optimizing with the Gemini 2.5 Flash model took roughly four hours and consumed millions of tokens, costing a couple hundred dollars in that case. The upfront cost is significant, but it can be worthwhile when it produces more consistent, higher-quality automated outputs with less human review needed afterward. _Developers weighing token costs against automation gains follow real-world case studies like this on daily.dev._ ## Similar posts on daily.dev - [How Automated Prompt Optimization Unlocks Quality Gains for ML Kit’s GenAI Prompt API](https://daily.dev/posts/how-automated-prompt-optimization-unlocks-quality-gains-for-ml-kit-s-genai-prompt-api-q8j6xn1gb) · Android Developers Blog · 0 upvotes · 0 comments - [4 tricks I use to get better videos from Gemini Omni every time](https://daily.dev/posts/4-tricks-i-use-to-get-better-videos-from-gemini-omni-every-time-s0ncia4z8) · XDA Developers · 0 upvotes · 0 comments - [Introducing Gemini Omni](https://daily.dev/posts/introducing-gemini-omni-fosehbhac) · DeepMind · 0 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#prompt-engineering](https://daily.dev/tags/prompt-engineering), [#firebase](https://daily.dev/tags/firebase), [#google-gemini](https://daily.dev/tags/google-gemini), [#vertex-ai](https://daily.dev/tags/vertex-ai) [View this post on daily.dev](https://daily.dev/posts/boost-accuracy-with-the-prompt-optimizer-ybzq2sjvb) ```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":"Boost accuracy with the prompt optimizer","url":"https://daily.dev/posts/boost-accuracy-with-the-prompt-optimizer-ybzq2sjvb","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/boost-accuracy-with-the-prompt-optimizer-ybzq2sjvb"},"datePublished":"2026-08-23T12:19:29.599Z","dateModified":"2026-08-23T12:27:23.688Z","description":"A walkthrough of using Google's Vertex AI Prompt Optimizer to automatically improve a system prompt for generating YouTube video descriptions from scripts. The...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/05cfb7cdb59dba4d7ff8ca57f9eca6cf?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/05cfb7cdb59dba4d7ff8ca57f9eca6cf?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Firebase","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":"Firebase","logo":"https://media.daily.dev/image/upload/s--i395oqZH--/f_auto,q_auto/v1787487448/logos/firebase-blog?_a=BAMAMicg0","url":"https://daily.dev/sources/firebase-blog"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/boost-accuracy-with-the-prompt-optimizer-ybzq2sjvb","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,prompt-engineering,firebase,google-gemini,vertex-ai","timeRequired":"PT13M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Firebase","item":"https://daily.dev/sources/firebase-blog"},{"@type":"ListItem","position":3,"name":"Boost accuracy with the prompt optimizer"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/boost-accuracy-with-the-prompt-optimizer-ybzq2sjvb#faq","mainEntity":[{"@type":"Question","name":"How does the Vertex AI data-driven prompt optimizer improve a prompt automatically?","acceptedAnswer":{"@type":"Answer","text":"It iterates on a system prompt using paired input and target output data, running computation-based or model-based evaluation metrics (via Gemini) to grade each variant against a baseline generated from the original prompt. Inputs and target outputs are uploaded as a CSV to Google Cloud Storage, then processed through a Colab Enterprise notebook that tests optimization modes like instructions-and-demo to find the best-performing version. See how teams track prompt-tuning workflows like this one on daily.dev before automating content pipelines."}},{"@type":"Question","name":"How long and how much does it cost to run the Vertex AI prompt optimizer on real data?","acceptedAnswer":{"@type":"Answer","text":"For 45 input prompt and description pairs, optimizing with the Gemini 2.5 Flash model took roughly four hours and consumed millions of tokens, costing a couple hundred dollars in that case. The upfront cost is significant, but it can be worthwhile when it produces more consistent, higher-quality automated outputs with less human review needed afterward. Developers weighing token costs against automation gains follow real-world case studies like this on daily.dev."}}]} ```