Sign in

André Bandarra

@bandarra.me
379 followers 132 following 148 posts

Chrome Web Developer Relations at Google

PostsRepliesMedia
André Bandarra @bandarra.me · 23/09/2026
⚠️ The loop was closed, but it immediately surfaced a big catch: the agent interrupted the user with a permission modal ("Allow tool actions"). For someone just trying to book a flight, an unexpected modal to submit developer telemetry is jarring and breaks the conversational flow.
110
André Bandarra @bandarra.me · 23/09/2026
✈️ To experiment with closing this loop, I added a fileSupportTicket tool to our flight search demo. When asked to filter by an unsupported aircraft type (Boeing 777), the agent recognized the missing capability and filed a structured bug report with reproduction steps and URL parameters.
110
André Bandarra @bandarra.me · 23/09/2026
🔄 Every software dev cycle relies on a feedback loop: design, build, test, launch, and review. When users hit broken flows on a site, some file support tickets. But when AI agents hit dead ends, they fail silently or apologize in prose. What happens if we give agents a bug-reporting tool? 🧵
131
André Bandarra @bandarra.me · 09/06/2026
Are you building with Angular and wondering how use WebMCP to expose tools to AI Agents? The Angular team has got you covered and published detailed docs on how to use WebMCP with the framework! Read all about it at: angular.dev/ai/webmcp #WebMCP #AgenticWeb
140
André Bandarra @bandarra.me · 11/05/2026
I wrote a Markdown editor on the web that has an AI Agent attached to it. The agent an read, write, edit, read other documents in the workspace, delegate to other agents and more. Check it out at bandarra.me/apps/agent-t... #AgenticWeb #AI
011
André Bandarra @bandarra.me · 09/05/2026
010
André Bandarra @bandarra.me · 09/05/2026
Good thing tomorrow is Sunday. I'm going to stay up late today playing some MIDI music :D
110
André Bandarra @bandarra.me · 06/05/2026
AI agents often feel bolted to sites. Moving the agent loop to the client-side enables tight integration between the agent and your site's user interface, while still allowing usage of big cloud LLMs or delegating tasks to server-side agents Read more at bandarra.me/posts/bringi... #AI #AgenticWeb
000
André Bandarra @bandarra.me · 29/04/2026
Building on the previous video, can AI write code to solve the maze? It turns out it can! Check the video and read more at bandarra.me/posts/webmcp... #WebMCP
072
André Bandarra @bandarra.me · 28/04/2026
Can we solve AI agent latency by changing where logic executes? 🔌 I'm investigating client-side code execution via #WebMCP. Instead of high-latency tool round-trips, the agent generates a local JS algorithm. Great for speed, but security trade-offs remain. Details: bandarra.me/posts/webmcp...
131
André Bandarra @bandarra.me · 21/04/2026
Is your AI agent truly "magical," or just a `while` loop with good PR? 🤖 To build production-grade agents, you have to trade magic for mechanics: 🧠 Brain (LLM) 🛠 Tools 💾 Memory 🔄 Engine Read about the plumbing behind these systems: bandarra.me/posts/demyst... 🦀
021
André Bandarra @bandarra.me · 01/04/2026
Been exploring #WebMCP — it only gives agents tools, no prompts, no injected context. I tried: tool description = hint, return value = full protocol. The agent calls the skill and gets the instructions. Lazy-loaded knowledge. Full write up: bandarra.me/posts/webmcp...
130
André Bandarra @bandarra.me · 11/03/2026
When to use WebMCP vs. MCP? 🤔 Use WebMCP to make your active website "agent-aware" by exposing UI tools directly in the browser. Use MCP for connecting agents to external data and services beyond the tab. Read more at: developer.chrome.com/blog/webmcp-... #WebMCP #MCP #AIAgents
185
André Bandarra @bandarra.me · 27/02/2026
Updated my Prompt API demo! 🚀 Now using WebMCP to expose on-page agents to broader browser agents. This lets a general agent recruit the page as a domain-expert "sub-agent" for hyper-accurate tasks. Links in replies! 👇 #WebMCP #BuiltInAI
111
André Bandarra @bandarra.me · 20/02/2026
I added WebMCP to my personal blog with a simple one-liner! Now users can leave comments with the help of their AI agents. 🤖💬 Join the early preview to help build the agentic web: developer.chrome.com/blog/webmcp-... #WebDev #AI #WebMCP
000
André Bandarra @bandarra.me · 24/03/2025
Ever wondered what really happens when you tweak the temperature on an LLM? 🤔 I built a visualizer to show you! See how probabilities shift & creativity blooms (or shrinks!). Dive in! ✨ bandarra.me/posts/unders... #AI #LLM andreban.github.io/temperature-...
A screenshot of the visualizer tool showing the prompt "Once upon a time, there was a", with temperature 1.0 and top-k 6. It shows the possible next tokens with the probabilities for each.
000
André Bandarra @bandarra.me · 12/03/2025
Yo Dawg, I heard you like to use AI to generate Summaries, so @alexandrascript.com and I have put together a blog post on summarizing your summaries to overcome context window limitations of the Summarizer API and get awesome summaries: developer.chrome.com/docs/ai/scal... #webai #builtinai
Yo dawg, I heard you like summaries, so here's a blog post on summary of summaries
053
André Bandarra @bandarra.me · 29/01/2025
Asked Gemini to make a squircle plush toy, because that's what the word makes me think of.
A plush toy in the shape of a squircle, with eyes and feet.
180
André Bandarra @bandarra.me · 14/01/2025
Great series of posts by @christianliebel.com on web.dev, covering the capabilities and limitations of LLMs running in the browser, and a walkthrough on building an on-device chatbot. #WebAI #PromptAPI #WebLLM #GenAI
A screenshot of the first post in the series. The title of the blog post says "Benefits and limits of large language models".
041
André Bandarra @bandarra.me · 08/12/2024
The Christmas tree is up. Mamba is entertained.
020
André Bandarra @bandarra.me · 21/11/2024
Oh, I didn't know we were doing this :) Here's Doom Fire, implemented using as an Offscreen Canvas demo. 🔥 The demo: doom-fire.com 🔥 Blog post: bandarra.me/posts/Buildi... 🔥 Source code: github.com/andreban/doo.... 🔥 Inspired by: fabiensanglard.net/doom_fire_psx/
022
André Bandarra @bandarra.me · 19/11/2024
I feel there should be a starter pack of people who every now and then post pictures of their cats... or maybe there's one already? Anyway, here's one of Rio using the CalDigit on my desk as a warm pillow for the cold weather.
Rio, the cat, sleeping with his head on top of the warm CalDigit dock.
6574
André Bandarra @bandarra.me · 10/08/2023
Went down the memory lane and found a link to the book I used to learn HTML and writing CGI applications back in around '96 - allperlbooks.com/book/en/using-html…. It was a fancy one because it came with 2 CDs.
000