🪰 Kimi Swarms 100 AI Agents

Hey folks,

The viral AI tool ClawdBot just became Moltbot after Anthropic sent a trademark notice and creator Peter Steinberger was forced to change it. Honestly, I'm not surprised Anthropic wanted distance from this one. The bot has some genuinely useful applications, but it also has massive security holes. When a tool carrying your name can autonomously execute commands with minimal guardrails, you're going to want your lawyers involved.

Meanwhile, Google announced that Gemini 3 is coming to Chrome with auto browse capabilities. The AI can autonomously navigate websites and complete tasks without you clicking through pages manually. Google calls it "the new era of browsing" where the browser itself takes actions on your behalf.

Today in AI:
→ AI assistants can now show interactive dashboards
→ This coding tool splits work across 100 AI assistants
→ How to set up AI Projects so it remembers you

Let's dive in..

📊 AI Assistants Can Now Show You Interactive Dashboards

What if your AI assistant could show you a live dashboard instead of just describing it in text?

The Model Context Protocol just launched MCP Apps, and it solves a frustrating problem. You ask your AI to pull up some data, it shows you hundreds of rows and summarizes what it found. Then you're stuck and if you want to filter by date, you have to ask again. Sort by revenue is another request. Every little change requires a new conversation.

MCP Apps changes that. Now AI tools can show you actual interactive dashboards, forms, and charts right in your chat window. You can click, filter, and manipulate the data yourself while the AI watches what you're doing and helps as needed. Here's what this looks like in practice:

• Sales dashboards: Instead of asking "show me sales by region," you get an interactive dashboard where you can click between regions, drill into specific accounts, and explore the data yourself

• Setup wizards: Forms that update automatically based on what you choose. Pick a cloud provider and the fields change to match their requirements

• Contract review: PDFs appear right in the chat with key clauses highlighted, and you can approve or flag issues without switching apps

• System monitoring: Live metrics that update in real-time as your servers change, not just a static snapshot

The security is built-in. Everything runs in a protected sandbox with approval controls for sensitive actions. What makes this powerful though, is it works the same way across Claude, ChatGPT, and other AI tools. Developers build it once, and it shows up everywhere. No custom code for each platform.

For your team: If you're using AI tools for data analysis or system configuration, this shift from text-only to interactive interfaces makes complex workflows much easier. Instead of asking the AI to describe what it sees, you can explore it yourself.

Together with Attio

Introducing the first AI-native CRM

Connect your email, and you’ll instantly get a CRM with enriched customer insights and a platform that grows with your business.

With AI at the core, Attio lets you:

• Prospect and route leads with research agents

• Get real-time insights during customer calls

• Build powerful automations for your complex workflows

Join industry leaders like Granola, Taskrabbit, Flatfile and more.

👉 Try Attio Pro for free

🪰 Kimi Splits Work Across 100 AI Assistants at Once

Moonshot AI just released Kimi K2.5 with a different approach to writing code: instead of one AI assistant working through your project step-by-step, it can spin up 100 AI helpers working in parallel. Think of it like having a hundred junior developers who each tackle a different piece while one rewrites old code, another writes tests, a third updates documentation, all at the same time.

The performance is solid. On coding tests, Kimi K2.5 scores between GTP-5 and Claude. For general tasks, it matches the top AI models. And it's free to use, costing 5x less than GPT-5 when you run it yourself.

Most AI coding tools work like having one really smart assistant. You ask it to build a feature, and it goes file by file, making changes one at a time. Kimi's approach is more like managing a team. It breaks big projects into smaller tasks, assigns them to different AI workers, and coordinates everything automatically. One checks how your code connects, another writes the tests, a third updates your docs.

The tool can also read diagrams, screenshots, and architecture drawings (not just code). It's available now for anyone to download and use.

AI coding tools are moving from autocomplete to autonomous teams. Tools like Cursor, Claude Code, and GitHub Copilot are all heading this direction, but Kimi makes the team approach explicit with up to 100 workers instead of just one doing everything.

For your team: The next time you evaluate AI coding tools, watch how they handle big changes across multiple files. The tools that can split work across AI assistants will finish projects faster than those working one step at a time.

🎯 How To Set Up AI Projects (Stop Starting from Scratch)

Most people open a blank chat every time they use AI. That's why results feel generic. Here's how to set up Projects so AI actually remembers your context and writes like you want:

1. Go to ChatGPT or Claude and create a new Project (not a blank chat)

2. Add custom instructions that define your voice, role, and goals once
   (AI will remember these permanently)

3. Upload 3-5 examples of your best work (writing samples, reports, emails) so AI learns your style

4. Turn on these features before you start:
   

   • Extended Thinking (makes AI reason deeper instead of giving surface answers)

   • Search (pulls real data instead of making things up)

   • Memory (remembers your preferences across conversations)

🐝 AI Buzz Bits

🔬 OpenAI launches Prism, a free workspace for scientists. The tool helps researchers write scientific papers, find relevant research, build bibliographies automatically, and collaborate in real-time. It's free for all ChatGPT users with unlimited projects. OpenAI wants to do for science in 2026 what it did for software development in 2025.

⚡ Mistral launches Vibe 2.0, a coding agent that asks before it acts. Mistral AI released Vibe 2.0, powered by their Devstral 2 model. The big change: when something's unclear, the AI asks you for clarification instead of guessing. You can also build custom helper agents for specific tasks (like PR reviews or test generation) and invoke them with slash commands.

🔍 Yahoo is back in search with Scout, an AI answer engine. After more than a decade away from search, Yahoo launched Scout across its entire ecosystem (Search, Mail, News, Finance, Sports). It's powered by Anthropic's technology plus Yahoo's knowledge graph of over one billion entities.

🛠️ Tool Spotlight

1. OpenCode — Open-source coding assistant. Works with Claude, OpenAI, or models you run yourself. You're not locked into one vendor, and it's built for people who prefer working in the command line.

2. Windsurf — An AI coding editor where the AI understands your entire project and what you're working on in real-time.

3. Ultracite — Automatically creates coding style rules for your AI tools (Copilot, Cursor, Claude Code) so they write code that matches how your team already writes. Set it up once and your AI assistants follow your team's standards automatically.

4. NativeBridge — Automates testing on real iPhones and Android phones using AI. Your QA team can review builds instantly with just a link, no need to install anything on their computer.

For a full list of 1500+ AI tools, visit our Supertool Directory