标签：#Automation

Picture the scenario: your invoicing pipeline generates a clean, branded PDF for a German B2B customer. It looks right. It would print fine, email fine, and satisfy anyone who opens it by hand. Then it bounces, because since January 1, 2025, that customer is legally required to receive invoices in a format their software can parse without a human retyping the totals. A pretty PDF isn't enough anymore, and honestly, for a machine, it never really was the point. The part that surprises people who haven't dealt with this yet: the mandate doesn't force you to give up the human-readable PDF. It just requires that PDF to carry a second, structured version of itself, riding along inside it. That format is called ZUGFeRD, with an internationally aligned sibling called Factur-X. If you've never had to build one, it's worth understanding the mechanics before the code, because it's a genuinely clever piece of engineering, not just a compliance checkbox. So how does a single file manage to be both a human-readable invoice and a machine-parseable one at once? What a ZUGFeRD invoice actually is Open a ZUGFeRD invoice in Adobe Acrobat or any PDF viewer and you see a normal invoice: logo, line items, totals, payment terms, nothing unusual. But embedded inside that same file, in its attachments, sits an XML document carrying the exact same invoice data in structured, typed form: invoice number, line items, tax rates, totals, every field an accounting system needs, tagged rather than buried in a paragraph a parser has to guess at. The container format making this possible is PDF/A-3 , the only PDF/A variant that permits arbitrary file attachments while still meeting the archival standard's long-term readability requirements. PDF/A-1 and PDF/A-2 explicitly forbid embedded attachments; PDF/A-3 was built for exactly this use case, which is why every ZUGFeRD file you'll open is, underneath, a PDF/A-3b document with an XML file riding inside it. The embedded XML follows EN 16931, the EU's

AI search is creating an attribution problem for marketers: a brand can help shape an answer in ChatGPT, Google AI Mode , or Perplexity without receiving a visit to its website. That makes rankings, impressions, and click-through rates incomplete indicators of visibility. New research from Wix Studio adds evidence that the content cited by AI systems follows recognizable patterns, while industry discussions increasingly point to measurement frameworks built around citations, answer presence, prompt coverage, and downstream influence. The key shift is not that website traffic has stopped mattering. It is that a click is no longer the only observable outcome of search visibility. When an AI interface summarizes options, recommends a product category, or cites a publisher, users may form an opinion or continue their journey elsewhere. Brands therefore need to separate direct referral traffic from their broader presence in AI-generated answers. What Wix Studio's research shows about AI citations Wix Studio's AI Search Lab research examines citations in answers generated by major AI search interfaces, including ChatGPT, Google AI Mode, and Perplexity. Published summaries describe a dataset of roughly 75,000 AI-generated answers and more than one million citations. Its central finding is that citations are not spread evenly across every kind of web page. Listicles, articles, and product pages account for a disproportionate share of the citations observed in the research. That is consistent with how answer engines retrieve and synthesize material: content that is clear, segmented, easy to scan, and closely matched to a question can be easier to extract into a response. A subsequent Search Engine Land summary of Wix Studio's work discussed a 25,000-URL dataset in which listicles represented a majority of AI citations. The precise mix should not be treated as a universal rule. Wix Studio's analysis covers a defined set of prompts and engines, and results can change with the

CleanSlate is an open-source platform for running coding agents across your local machine and the cloud. Today, it works through an IDE, CLI, and SDK. Agents can understand a codebase, make changes, run commands, browse the web, and verify their work. We are now building longer-running autonomous cloud agents that can continue working without keeping your machine active. CleanSlate supports multiple model providers and is not tied to a single ecosystem. GitHub: https://github.com/TheWariend/CleanSlate Website: https://thewariend.com/cleanslate The project is still early, and I would appreciate honest feedback from developers who use coding agents.

Yelp has confirmed a licensing agreement with OpenAI that will extend Yelp content into AI platforms, including the OpenAI ecosystem powering ChatGPT. The deal positions Yelp’s reviews, ratings, photos and business information within a growing AI-driven local discovery experience, while opening a potential path for users to request quotes from local service providers through ChatGPT. The agreement is more consequential than a new search result format. Yelp is expanding its data-licensing strategy beyond conventional search surfaces, while ChatGPT gains access to a major source of local business content. For people asking an AI assistant where to eat, which contractor to contact or how a nearby business is rated, the quality, freshness and governance of the underlying data will matter as much as the answer itself. What the Yelp and OpenAI agreement covers In its February 2026 earnings and shareholder release , Yelp announced an agreement with OpenAI and described it as part of its AI transformation and strategy to license content for local discovery across AI ecosystems. That is the confirmed foundation of the development. Axios has reported the practical user-facing direction: ChatGPT will surface Yelp reviews, ratings, photos and other business details in responses to local queries. Yelp has also signaled that its Request a Quote capability could be integrated into ChatGPT in the near term, enabling users to initiate an inquiry with a service provider from the AI interface. Capability What the research supports Status Yelp content in ChatGPT Reviews, ratings, photos and other business details are expected to surface for local queries. Reported user-facing outcome of the confirmed licensing agreement Request a Quote in ChatGPT Users may be able to initiate quote requests with local service providers through the AI interface. Signaled for a future rollout Data timing and interface design Reporting describes real-time business data, but exact latency, update frequency

Google is expanding Gemini on two fronts at once: faster, lower-cost models for software and enterprise workflows, and a new robotics family designed for embodied, cross-robot control. The releases include Gemini 3.6 Flash , Gemini 3.5 Flash-Lite , Gemini 3.5 Flash Cyber, and Gemini Robotics 2 with related embodied-reasoning and on-device variants. The clearest immediate enterprise story is the widening choice of models for agentic work. In its official Gemini Flash announcement , Google positions 3.6 Flash as a general workhorse for coding, knowledge work, and multimodal tasks, while 3.5 Flash-Lite is aimed at workloads where response speed and cost efficiency are decisive. The robotics update extends the same broader push beyond software agents into systems that must reason about and act in physical environments. A broader Gemini stack for agentic workloads Gemini 3.6 Flash is generally available through Google's developer, enterprise, and consumer channels. Google says it improves on 3.5 Flash for coding, knowledge-work, and multimodal tasks, while producing around 17% fewer output tokens than 3.5 Flash. That token-efficiency claim matters because output tokens are a material part of both latency and inference spending in multi-step agent workflows. Google lists pricing for Gemini 3.6 Flash at $1.50 per 1 million input tokens and $7.50 per 1 million output tokens . The company describes the model as offering a lower cost per task, a metric that depends not only on token prices but also on how many tokens a task requires to complete. The supplied release information does not provide a price for 3.5 Flash-Lite, so it should not be inferred from 3.6 Flash pricing. Gemini 3.5 Flash-Lite occupies a different role. Google calls it its fastest and most cost-effective subfamily, with a stated output speed of 350 output tokens per second . It is intended for high-throughput agentic workflows where an organization may value quick model responses and high request volume ove

How to Audit Hidden Reminders and Context Usage in Claude Code Logs | Agent Lab Journal Agent Lab Journal Guides Glossary Advanced field guide How to Audit Hidden Reminders and Context Usage in Claude Code Logs Advanced · 45 min read · Local analysis · Updated August 1, 2026 The visible transcript in Claude Code is not necessarily a complete representation of everything recorded around a request. Service messages, internal reminder markers, tool payloads, and usage metadata can exist in session logs without appearing as ordinary chat turns. If you want to know how often ip_reminder occurs—or how input, output, cache creation, and cache read tokens are distributed—you need to inspect the stored records directly and preserve enough structure to avoid misleading totals. In this guide What this audit can establish Concrete investigation case Locate and select one session Preserve an auditable copy Run a quick structural check Build the full local report Interpret reminder and token data Verify the report independently Failure cases and repairs Limitations What this audit can—and cannot—establish This workflow examines one local session stored as JSON Lines (JSONL): a text format in which each line is normally an independent JSON value. It creates a report with: the selected file’s path, size, modification time, and SHA-256 digest; the number of physical lines, parsed records, blank lines, and malformed lines; every record containing the exact, case-sensitive string ip_reminder; the JSON paths at which the marker was found; timestamps and record types when those fields are available; per-record and aggregate input, output, cache creation, and cache read token values; a chronological CSV suitable for a spreadsheet or notebook; a machine-readable JSON report for later comparison. The report shows what is present in the selected file. It does not prove why a reminder was inserted, whether it was transmitted to a model exactly as stored, or how the client’s undocumented inte

Google Gemini offers an AI trip planner that combines travel research, itinerary generation and Google service integrations in one conversational workflow. The capability can surface real-time flight and hotel options, build itineraries around a traveler’s interests and adjust plans as needs change. Although Google is continuing to promote the feature, its official materials position it as an established part of the Gemini ecosystem rather than a newly launched product. The practical appeal is straightforward: trip planning often requires moving among airfare searches, hotel listings, maps, saved locations and notes. Gemini is designed to bring several of those steps together. On Google’s official Gemini AI trip planner page , the company describes prompts such as planning a four-day Tokyo visit around particular interests, then using Gemini to organize a tailored schedule by neighborhood. What Gemini’s travel planner can do Gemini’s travel functionality is framed as a consumer assistant for the research and planning stages of a trip. Users can describe a destination, trip length, interests or preferred travel style in natural language. Gemini can then help turn that input into an itinerary while drawing on relevant Google travel and mapping services. The official descriptions identify several connected capabilities: Real-time flight options through Google Flights. Real-time hotel options through Google Hotels. Customized itineraries organized around a traveler’s requested interests and locations. Plan adjustments during the trip , rather than a fixed itinerary created only before departure. Maps integration for navigation and points of interest along a route. Google’s Gemini Apps support material also confirms that the apps can help plan trips and retrieve live flight information. Maps integration matters because it extends the experience beyond trip inspiration: a user can move from deciding what to do to navigating to places and discovering points of interest whi

This article was originally published on e-shikumi-labo . Hello, I'm Shin from e-Shikumi-Labo. This is the final installment (Part 4) of "Systematized Thinking," where we use AI to build our own tools and independently maintain them. So far, we have discussed creating a prototype that automatically saves Gemini chat logs, converting them to Markdown for Obsidian integration, and elevating it to a safe, fully automated system. In this final installment, we will cover the "countermeasures for downtime due to screen specification changes," an unavoidable issue when operating tools that handle web data, and the core of the "self-reliance" humans should possess in the AI era. 1. The Web Data Extraction Compromise: "You Can't Extract What Isn't on the Screen" During development, there was a time when I thought, "I also want to record the exact date and time (timestamp) when the chat was sent." However, no matter how much I analyzed Gemini's screen structure, the exact timestamp of each utterance did not exist in the HTML. The fundamental rule of web data extraction is: "You cannot extract data that does not exist on the browser screen." As long as you are extracting data from the screen (DOM) rather than via an API, forcing the extraction of something that isn't there will require complex guesswork processes and will instead become a cause of trouble. Understanding this "technical limit," gracefully giving up on what cannot be done, and judging to maintain simplicity is also an important element of tool building. 2. Specification Changes Are Not Defects, But "Fate" As long as you deal with tools that extract data from other people's websites, the time will inevitably come when the tool suddenly stops working one day due to design changes or updates on Google's side. "It was working fine until yesterday, but suddenly it stopped saving." This is not a defect in the tool, but an unavoidable "fate" as long as you depend on someone else's platform. The important thing is not t

This article was originally published on e-shikumi-labo . Hello, I'm Shin from e-Shikumi-Labo. This is Part 3 of "Systematized Thinking," where we use AI to build our own tools and independently maintain them. Last time, I talked about creating a system to automatically output Markdown (.md) files to Google Drive simultaneously with appending to a spreadsheet. With list management in a spreadsheet and a comfortable viewing environment in Obsidian established, it was getting very close to completion as a tool. However, as I continued to use it practically, new challenges emerged on the operational front. This time, I will share the risks I faced while transitioning from a "manual button" to "full automation," and the process of evolving into safe code. 1. I Want to Eliminate the "Hassle of Pressing a Button" During the prototype stage, the system was designed so that logs were saved by pressing a button placed on the screen. However, as long as a human operates it manually, there are inevitably limitations. If you are concentrating on the conversation, you might forget to press the save button and close the screen. If the conversation gets long, you might miss past utterances that are no longer displayed on the screen. "If I have the screen open and am conversing, I want it to automatically save in the background without bothering human hands." Thinking this, I asked the AI to write the code for full automation. 2. The Code the AI Produced: "Patrolling the Screen Every 1.5 Seconds" When I consulted the AI, it immediately presented code for full automation. The mechanism was, "Start a timer every 1.5 seconds, check the entire screen in the background, and send any new utterances." When I actually tried it, the logs accumulated automatically as soon as I conversed without pressing the button, and at first glance, it looked like exceptionally well-done full automation. However, I felt something was slightly off regarding this "monitoring on a 1.5-second cycle." 3. The B

What I wanted I'm building a system where all my daily records live in Notion, so I can point an AI at it and get feedback. Goals, tasks, daily logs, finances — those are all manual entry, and that's fine. But one day it hit me that weight would be nice to sync automatically. The requirements were simple: Every morning, my weight and body fat percentage get appended to a Notion database as one row No manual typing That's it. My scale is a Withings Body Smart. The design I picked first This one: Scale → vendor app → Apple Health → iOS Shortcut → Notion API I chose Apple Health as the hub for these reasons: It doesn't depend on the scale model. As long as the data lands in Health, the same implementation works for any vendor. No server required. A time-based Shortcuts automation handles it end to end — no always-on machine, no cron. Free. No extra subscription. Extensible later. Anything that's already in Health — steps, sleep, heart rate — could be added the same way (if I ever wanted to). Generic, zero cost, extensible. The design looked sound to me. Implementation Here's what the Shortcut looks like: 1. Find Health Samples [Weight] latest, limit 1 2. Get Details of Health Sample [Value] → variable Kg 3. Get Details of Health Sample [Start Date] → variable SampleDate 4. Format Date yyyy-MM-dd → variable Ymd 5. If Ymd == today 6. Text ← build the JSON 7. Get Contents of URL ← POST to the Notion API Step 5 matters. Without it, on a day you don't step on the scale, yesterday's weight gets appended under today's date . Here's the JSON built in step 6: { "parent" : { "database_id" : "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx" }, "properties" : { "Date" : { "title" : [ { "text" : { "content" : "@@YMD@@" } } ] }, "Measured" : { "date" : { "start" : "@@YMD@@" } }, "Weight kg" : { "number" : @@KG@@ }, "Body fat %" : { "number" : @@FAT@@ } } } (My real database uses Japanese property names. What matters is that they match your database exactly.) I write this as a plain string in a

When I joined Entire, I noticed my boss spending a chunk of time every week writing detailed release notes, called Dispatches at Entire. It looked like a painful process. Each Dispatch had to cover changes across several repositories, explain why those changes mattered, credit external contributors, and carefully avoid leaking anything that was not public yet. I offered to take it over. I had solved a similar problem before, so I figured it would be an easy win. I built something similar and simpler at Block While I was at Block, I built a release notes generator for goose . It ran in GitHub Actions after a release workflow completed, checked out the new tag, compared it against the previous one, and handed goose a recipe to inspect the commit diff. Goose organized those commits into features, bug fixes, improvements, and documentation. Each entry got a short description and a PR link. The workflow then updated the GitHub release and posted the announcement to Discord, opening a thread if the notes exceeded the message limit. It was clean and effective, but it solved a very clean problem: one repository, one new release tag, public commit history, and concise output. So when I looked at Entire’s Dispatches, I assumed I could reuse the same playbook. Gather changes, run goose, post the draft. That assumption did not survive contact with reality. But a Dispatch turned out to be more complex A Dispatch spans multiple projects: the Entire CLI, entire.io, EntireDB, external agent integrations, and open source libraries like go-git, go-nuts, git-sync, and ForgeMark. Every project also ships on a different cadence. Some push to main and deploy continuously. Others bundle work into scheduled releases. The CLI maintains separate stable and nightly channels, which means a feature can be available to testers without being part of the latest stable tag. Then there are feature flags. Finding changes was not the hard part because GitHub APIs handle that easily. The hard part was

security #api #domain #subdomaintakeover #defcon #whois #rapidapi #threatintel DEF CON 32 made one thing clear: open-source security chips and hardware keys are having a moment. But while badges get the spotlight, most real-world attacks still start with something far less glamorous — a forgotten DNS record, a dangling CNAME, or a missing DMARC policy. Subdomain takeover remains one of the most reliable paths from "benign misconfiguration" to "account compromise." If your organization owns dozens or hundreds of domains, manual checks do not scale. This is where an API-first domain intelligence tool becomes essential. In this post, we'll use the Domain WHOIS API to automate: WHOIS/RDAP lookups and domain-age checks DNS record enumeration and SSL certificate inspection Subdomain discovery and takeover-risk scoring Email-security validation (SPF, DMARC, DKIM, DNSSEC, MTA-STS) Historical snapshots via /history Why subdomain takeover still matters A subdomain takeover happens when a DNS record points to a third-party service — GitHub Pages, Heroku, AWS S3, Vercel, etc. — that is no longer registered under your account. An attacker can claim the dangling endpoint and suddenly serve content under your brand's domain. Bug bounty programs consistently rank subdomain takeovers as high-severity findings because they enable phishing, session hijacking, and reputation abuse. The root cause is usually an orphaned CNAME that nobody is monitoring. The fix is continuous monitoring. Instead of running dig , whois , and openssl by hand, we can consolidate everything into a single API call. What the Domain WHOIS API returns The API combines several data sources into one response: Capability Use case WHOIS via RDAP Ownership, registrar, creation/expiration dates DNS records A, AAAA, CNAME, MX, NS, TXT records SSL certificate Issuer, expiry, SANs, validity Subdomain discovery Asset inventory and shadow-IT detection Takeover risk Dangling CNAME/A-record scoring Email security SPF, DMARC,

Disclosure: BrowserAct sponsored this piece. The BrowserAct links below are affiliate-tracked — I get...

Every idea gets run through a one-sentence test before it is allowed to count as a real idea at all Most ideas die for one of three specific reasons, not vague lack of enthusiasm An idea only earns a build slot once it has survived contact with a real, repeated problem A maybe-later list holds the rest on purpose, and I check it far less often than people assume The One-Sentence Test I Run Before Anything Becomes an Idea I get more ideas than I could ever build. That is not a boast, it is a liability if I do not manage it, because every one of those ideas feels exciting for about twenty minutes, and excitement is a terrible filter for what is actually worth my evenings. So before an idea is allowed to sit on any kind of list, it has to pass one test: can I describe the smallest useful version of it in a single sentence, with no "and" in the middle. That sounds small, but it kills more ideas than any other step in the process. "A tool that tracks my Claude usage and also shows analytics and also has a community feature" does not pass. "A tool that warns me before I hit my usage limit" passes. The first sentence is a pitch for a platform. The second sentence is a pitch for a Tuesday evening. I want the second kind, because the second kind is the one I actually finish. I did not always work this way. Early on, an idea earned space on my list the moment it sounded interesting, and my list grew into a graveyard of half-described plans that all needed a paragraph to explain. A paragraph is a warning sign now, not a feature. If I need more than one sentence to say what the smallest version does, the idea has not actually taken shape yet, it has just acquired enthusiasm, and those are different things. The test also forces honesty about scope early, before I have sunk any real time into something. An idea that needs "and" is usually two or three ideas wearing a trenchcoat, and pulling them apart at the sentence stage is far cheaper than pulling them apart three weeks into a

OpenAI has formally outlined a national science initiative designed to connect frontier AI models with government research infrastructure, National Laboratories, universities, and working scientists. The program is not a single model launch. Instead, it combines funded access, early product access, scientific campaigns, and an emphasis on fitting advanced AI into real research workflows. The initiative gives concrete form to OpenAI’s stated goal of helping scientists use increasingly capable models to accelerate discovery. In its official announcement on advancing the next era of national science , published July 22, 2026, the company describes a long-term strategy built around the U.S. Department of Energy’s Genesis Mission and collaborations with National Laboratories. The core proposition is that AI can contribute to hypothesis testing, simulations, and experimental work when it is deployed alongside scientific infrastructure and human expertise. That framing matters. OpenAI is positioning frontier models as tools that researchers direct and evaluate, rather than as a replacement for the institutions and specialists responsible for scientific work. What OpenAI is providing to scientific researchers OpenAI’s commitments span several types of access, from coding support for a broad research community to model capabilities and API funding for large campaigns. The announced provisions include: $4 million in Codex access for approximately 2,000 Genesis researchers at national labs and universities. $3 million in API support for two large scientific campaigns. Up to $10 million in API usage for participating researchers who reach a $2.5 million spending threshold. Access to GPT-Rosalind’s bioscience capabilities for national-lab researchers. Early access to selected models and features for trusted national-lab leaders preparing workflows and evaluations. Expanded access to advanced cyber capabilities for national-lab cybersecurity researchers. These commitments indicat

OpenAI has published a post-mortem examining an unusual pattern in its model testing: recurring references to “goblins” and “gremlins” in model outputs. The company’s official post, “Where the goblins came from” , published on April 29, 2026, frames the behavior as an emergent effect of reinforcement learning and human-feedback dynamics, not as a new product feature. Its practical message is more consequential than the metaphor suggests: unexpected model personas can affect the consistency, safety, and reliability that developers expect from AI systems. The published analysis provides the substantive context behind recent attention to a purported “goblin-level” post. Rather than indicating a model launch, OpenAI’s account suggests a narrower but important lesson about how optimization signals can inadvertently reinforce patterns in language models. For organizations using LLMs in production, the relevant question is not whether goblin-like language is amusing. It is whether teams can detect and address unexpected behaviors before those behaviors influence customer-facing, operational, or high-stakes workflows. What OpenAI documented OpenAI said the “goblin” and “gremlin” metaphors appeared during GPT-5.x testing and RLHF training. The company reported a notable increase in goblin-like language during GPT-5.5 testing when Codex was being evaluated. According to the post, the pattern emerged from reward-signal dynamics : persona-like responses were inadvertently reinforced through reinforcement learning and human feedback. That distinction matters. OpenAI does not characterize goblin behavior as a fixed capability or intentional model identity. It describes it as a byproduct that can arise at scale when a training and feedback process favors certain output patterns. The episode is therefore best understood as an alignment and evaluation lesson, rather than evidence of a separate “goblin” model, feature, or policy release. OpenAI also described a mitigation introduced

Google Photos has launched Video Remix , an AI-powered editing feature that applies stylized templates to users' existing video clips. Powered by Gemini Omni , the tool is designed to turn a video into a more cinematic or artistic version through a one-tap workflow inside Google Photos. The feature matters because it brings generative video styling into a consumer photo library and editing workflow rather than requiring users to begin in a dedicated video-generation product. According to Google's official Video Remix announcement , templates can add cinematic relighting, replace backgrounds, and apply artistic treatments including watercolor, raw sketchbook, and oil painting. What Google Photos Video Remix changes Video Remix is built around easy-to-use templates rather than a conventional timeline editor. A user starts with their own clip, chooses a template in the Google Photos Create workflow, and receives a stylized result. That positions the feature as a fast option for personal memories, social posts, and short marketing assets where a full editing process may be disproportionate to the desired output. Google says Video Remix is beginning to roll out to eligible Google AI Plus, Pro, and Ultra subscribers in select countries. The supplied materials identify the subscription tiers, but do not provide feature-specific pricing or a complete country-by-country availability list. Access may therefore differ by market and rollout stage. Area Google Photos Video Remix Broader Gemini Omni context Primary workflow Applies templates to a user's existing video clips in Google Photos Supports wider video generation and editing workflows Documented examples Cinematic relighting, background changes, watercolor, raw sketchbook, and oil painting Style-driven video transformations, including claymation-style demonstrations reported by third parties Access described in supplied research Rolling out to eligible AI Plus, Pro, and Ultra subscribers in select countries Google docume

Google has expanded Gemini Omni into Google Vids for end-to-end AI video generation and editing. The update lets users create clips from text and image references, then make targeted changes to existing footage through a step-by-step conversation. Rather than rebuilding a video after each revision, users can describe an adjustment, supply additional media where useful and refine the result in place. The central development is Omni's use of multimodal and real-world understanding in a Vids workflow. According to Google DeepMind's Gemini Omni overview , the model can work from arbitrary media, including images, text, video and audio, and apply reference-to-video capabilities grounded in world knowledge and physics-like reasoning. In Google Vids, that foundation is intended to make generated and edited scenes more coherent in composition, context and visual behavior. For teams that already use Vids to communicate ideas, training material or internal updates, the change moves AI assistance beyond first-draft generation. It introduces a conversational editing layer that can alter a chosen part of a video while preserving the broader scene and workflow. What Gemini Omni changes in Google Vids Gemini Omni supports both video creation and revision. A creator can begin with a prompt or image reference to generate a clip, or bring in existing footage and specify what should change. Google describes examples such as changing color grading or lighting, replacing backgrounds and removing background elements. This distinction matters because prompt-to-video and video editing have different practical constraints. Generating a new clip can be useful when no footage exists. Editing existing material is more relevant when a team wants to retain an established subject, scene or message while changing selected details. Omni's reference handling is designed to connect those modes rather than treating each request as an isolated output. Workflow How Gemini Omni is used in Vids Supported

I built a small Flask example that turns voicemail into a routing workflow. Code: https://github.com/team-telnyx/telnyx-code-examples/tree/main/voicemail-smart-router-python The app accepts either: a voicemail transcript an uploaded voicemail audio file For audio, it transcribes the voicemail first. Then it uses Telnyx AI Inference to classify the message and decide where it should go. Categories The app classifies voicemails into: urgent billing support sales spam routine Each category maps to a route: urgent -> Slack alert billing -> email support -> ticket queue sales -> CRM lead spam -> blocklist + archive routine -> daily digest Run it git clone https://github.com/team-telnyx/telnyx-code-examples.git cd telnyx-code-examples/voicemail-smart-router-python cp .env.example .env pip install -r requirements.txt python app.py Configure .env : TELNYX_API_KEY=your_telnyx_api_key AI_MODEL=zai-org/GLM-5.2 FALLBACK_MODEL=meta-llama/Llama-3.3-70B-Instruct HOST=127.0.0.1 Optional Slack webhook for urgent messages: SLACK_WEBHOOK=https://hooks.slack.com/... Classify a transcript curl -X POST http://localhost:5000/voicemails/transcript \ -H "Content-Type: application/json" \ -d '{ "transcript": "This is an emergency. Our production system is down and we need help immediately.", "caller_number": "+17177247292" }' Example response: { "category" : "urgent" , "confidence" : 1.0 , "priority" : "high" , "reason" : "The caller reports a production system outage requiring immediate attention." , "suggested_action" : "Escalate immediately to the on-call engineering team." , "route" : "slack" , "routed_to" : "#oncall-alerts" , "routing_status" : "delivered" } Process voicemail audio curl -X POST http://localhost:5000/voicemails/process \ -F "file=@voicemail.wav" \ -F "caller_number=+17177247292" For audio, the app calls: POST /v2/ai/audio/transcriptions using: distil-whisper/distil-large-v2 Then it calls: POST /v2/ai/chat/completions to classify the transcript. Routes included POST /voic

Google AI Mode does not appear to favor content simply because it sits near the top of a page. Research from SALT.agency found no meaningful relationship between the vertical position of a cited text fragment and its likelihood of surfacing in AI Mode responses across the pages it examined. That finding matters for publishers and SEO teams trying to understand how Google AI Mode selects supporting material. The available evidence points away from a universal above-the-fold formula and toward a more familiar discipline: publishing well-structured content that directly addresses the reader's need. The research also identifies a recurring pattern in highlighted material: descriptive subheadings followed by clear opening sentences. What SALT's AI Mode research measured In its research into whether content structure improves AI Mode surfacing , published December 30, 2025, SALT.agency analyzed 2,318 unique URLs cited by AI Mode across travel, e-commerce, and SaaS. The researchers used a Chrome bookmarklet and a 1920 by 1080 viewport to record the vertical location of the first highlighted text fragment on each page. The study recorded average cited-fragment depths of roughly 2,400 to 4,600 pixels across the three verticals. Some cited material appeared much farther down a page, including at depths above 60,000 pixels. Despite those differences, the analysis found no consistent citation advantage for content located near the top of a page. Content factor What the research observed Editorial implication Vertical page position No meaningful correlation between shallow pixel depth and being cited. Do not treat above-the-fold placement as a reliable AI Mode citation tactic. Page layout Elements such as hero images can push cited text farther down the page. Layout can affect where a fragment appears without determining whether it is selected. Headings and opening sentences Highlighted passages often included a descriptive subheading and the sentence immediately after it. Use h