Wissenswertes

Key Takeaways: Google introduced new standards named "Open Knowledge Format" and "Agent Resource Discovery" OKF for Google should work as: a vendor-neutral, agent- and human-friendly standard for representing the metadata, context, and curated knowledge that modern AI systems need. Just markdown — readable in any editor, renderable on GitHub, indexable by any search tool Just files — shippable as a tarball, hostable in any git repo, mountable on any filesystem Just YAML front

Key Takeaways: Elvis emphasizes the importance of owning the orchestrator and the harness for complex AI workflows the orchestration, the harness, routing capabilities, dynamic artifacts/workflows, verifiers, ability to switch/route between agent backends, automations, the skills, and the MCP tools would be the absolute best defense for what happened with Fable this week. The potential downside: high maintenance, might be too costly, and migth be unsustainable The bigger you

Key Takeaways: Perplexity announced some great updates such as introducing Hybrid Agentic Inference and Search as Code Hybrid Agentic Inference enables you to split tasks between a local model running on your machine and frontier models in the cloud. This keeps private data on your device and maximizes token efficiency Only let Perplexity access a local folder and run a local model via a subagent, then switching to more powerful, expensive models in the cloud Search as Code c

Key Takeaways: Tran ran a nice experiment on PDF extraction, especially for tables, using different models/libraries In short: LlamaParse wins on speed and accuracy. It’s the fastest overall and produces the cleanest output, but it requires sending PDFs to LlamaCloud. Marker is the best local option. It’s faster than Docling and handles simple tables well, but it merges columns on dense layouts. Docling is the slowest of the three and prone to hallucinating values on dense ta

Key Takeaways: Google Chrome introduced its MCP and Google Chrome Canary 150.x now supports WebMCP while adding a "agentic-browsing" category to Lighthouse 13.3.0 audits WebMCP + Lighthouse: whether the page exposes a clean accessibility tree which WebMCP tools are registered (imperative + declarative form tools) whether their input schemas validate form coverage - which <𝚏𝚘𝚛𝚖> elements are annotated for agents and which are invisible to them whether your llms.txt follows

Key Takeaways: Screaming Frog released version 24.0 of their crawler tool and they dropped a bomb with lots of incredible features: Local MCP: integrate Screaming Frog crawls into your agentic workflows. Of course there need to be some improvements as you currently cannot control lots of important crawl settings AND cannot create crawl configs using the MCP - still, lots of opportunities analyzing crawl data with Claude Auto-compare + Email notifications: You can now setup ma

Key Takeaways: On 16th of April, Claude Opus 4.7 was released Wojcik cited the AMD report about performance issues with Claude Opus 4.6 BridgeBench reported that accuracy on the hallucination benchmark of Opus 4.6 dropped to 68,3% vs. 83,3% before He outlines that it is a natural rhythm of hype cycles that push providers to allocate infrastructure to newer models taking power away from other models He proposes to adjust flexibly to the time of day: e.g. during end of business

Key Takeaways: Google introduced Gemma 4 , a state-of-the-art model with 256K context, Open-Source on Apache 2.0 license , that can be run locally as it only needs less space: 31B - full reasoning, runs on a workstation 26B MoE - fast, activates only 3.8B params per inference E4B / E2B - runs on a phone, offline, real-time audio + visio Google published their new compression algorithm TurboQuant that opens the door to new local developments and use cases: TurboQuant: Our

Key Takeaways: Grimm on next level crawling possibilities for Agents: Cloudflare /crawl REST API offers crawling based on Cloudflare infrastructure: "𝗖𝗹𝗼𝘂𝗱𝗳𝗹𝗮𝗿𝗲 𝗷𝘂𝘀𝘁 𝗰𝗼𝗺𝗺𝗼𝗱𝗶𝘁𝗶𝘇𝗲𝗱 𝘄𝗲𝗯 𝗰𝗿𝗮𝘄𝗹𝗶𝗻𝗴" REST API with POST request to initiate crawl and GET to retrieve results Sitemap or Spider crawl mode Javascript rendering in headless browser Result as HTML, markdown or JSON with using a prompt against the crawled pages incremental crawling (only

Key Takeaways : Ray & Rudzki stress the risk of heavy AI usage in SEO & content creation as they point out examples of failure: "Mount AI" is the rapid increase of visibility that is followed by a rapid decline as Search algorithms catch up An analysis of success stories of AI content tool from Landwehr revealed that 57% of websites saw their visibility tank , 22% that successfully gained rapid growth in the short term but tanked after Ray highlights: These drops can be rea

Key Takeaways: After n8n, Zapier & others now OpenAI launched its own workflow builder to build AI workflows for OpenAI Models: OpenAI Agent Builder Agent Builder enables users to create multi-Agent environments with tools , guardrails , vector storages , IF-ELSE-flow logics and more in a no-code UI Contrary, Google offers a Python library to create multi-Agent environments: Google SDK LangChain comments: "Not another workflow builder" Visual workflow builders are not “low”