SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

Google Gemini Agentic Video Cuts Token Use 88% but Has 4 Silent Failure Conditions

0
·5 views

A developer integrating Google's Gemini Agentic Video feature into a LINE Bot found that the mode reduces token consumption by 88% and costs by 66% compared to standard static video processing. Unlike the default static mode, which loads every frame at a fixed sampling rate, agentic mode lets the model selectively scan only relevant video segments to answer specific queries. Testing on a two-hour Google I/O keynote showed the feature could accurately pinpoint timestamped content, such as pricing discussions and product segments. However, the developer discovered four conditions must be met simultaneously for agentic mode to activate: using API version v1beta1, setting the media_processing parameter to AGENTIC, using a supported model, and configuring the thinking_level setting. Critically, if any condition is unmet, the API still returns a successful 200 response with no error, silently falling back to the more expensive static mode.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Strategy Pattern Can Replace Bloated SQL Queries in Role-Based App Search

A software engineering article on DEV Community argues that complex, role-based SQL search queries should be treated as structured application logic rather than dynamically assembled strings. The piece demonstrates how a legacy approach — appending filters and role conditions to a single StringBuilder method — produces unmaintainable, hard-to-test code that can obscure security boundaries. The author proposes applying the Strategy design pattern twice: one set of strategies governs what data a user is permitted to see based on their role, while another handles the optional search filters a user applies. A companion demo repository illustrates both approaches against the same SQL Server schema, using Spring Boot, Spring JDBC, and Testcontainers for real-database testing. The article covers five distinct user roles with differing visibility rules and ten optional filters, highlighting how the composed approach makes each rule independently addressable and reviewable.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer fixes Next.js portfolio invisible to Google and AI crawlers with 5 tweaks

Full Stack Developer Dhruv Sheladiya discovered his Next.js portfolio was nearly invisible to search engines and AI crawlers like ChatGPT because the homepage relied entirely on client-side JavaScript rendering. Since bots such as Googlebot and AI crawlers often skip or delay JavaScript execution, they were seeing an empty page instead of his actual content. He resolved this by embedding crawlable profile text in a server-side layout component using Tailwind's screen-reader-only class, adding structured JSON-LD schema with sameAs links to his social profiles, and ensuring all those profiles linked back to his site. He also corrected a mismatched Open Graph image — resizing it from 1629×966 px and 1.77 MB down to the standard 1200×630 px at 80 KB using Next.js's built-in Sharp library. The fixes improved his site's visibility in search results and made his identity recognisable to AI-powered search tools.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

AI Writes 47% of Enterprise Code in 2026 as Developer Trust and Adoption Surge

By 2026, artificial intelligence has become deeply embedded in software development, with AI assistants writing 47% of new code in enterprise GitHub repositories and being used daily by 44% of software teams. Developer trust in AI-generated code has grown significantly, with 81% of surveyed developers saying they trust it more than code written by junior colleagues. AI-powered tools have also transformed code review, cutting release delays from several days to under 18 hours for 62% of teams, while AI-assisted test generation has reduced post-release bugs by 38% compared to manual testing. Observability platforms now predict outages with 94% accuracy, and AIOps tools have cut mean recovery times by 43% at major companies. Looking ahead, Gartner projects that 65% of all code will be generated or reviewed by AI by 2028, signalling a fundamental shift in how software is built and maintained.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Google Search Console Launches Multimodal Filter to Track Image-Initiated Searches

Google Search Console has added a multimodal search type filter to its Performance reports, enabling website owners to track traffic from searches that begin with visual input rather than typed text. The filter consolidates data from four entry points: Google Lens, Circle to Search on Android, image uploads to Google Search, and Chrome's right-click image search feature. Google began rolling out the update globally on September 24, 2026, with data appearing as image-initiated search traffic reaches individual sites. The filter is available within both the standard Search results Performance report and the Generative AI features report, requiring no new dashboard. The update gives publishers, e-commerce sites, and local businesses a unified view of visual search behavior that was previously difficult to measure within Search Console.

0 comments Read more at DEV Community