Just published my first Substack article:
Building a High-Quality Personal MCP Knowledge Base with Multiple Interfaces
I built eswat2-mcp — a pure Python (FastAPI + FastMCP) knowledge platform that serves as a reliable digital twin of my 40+ years in UI/UX software engineering.
It features the same content through multiple experiences:
• REST API (live on Vercel)
• Full MCP (locally / self-hosted)
• 2D + 3D knowledge graphs
• New bespoke Wiki with full browser history & deep linking
Full article: https://t.co/Xx5XrXAbQ4
Would love feedback from anyone working on personal knowledge infrastructure, AI context systems, or MCP tools.
#MCP #FastAPI #PKM #KnowledgeBase
At the 24H Daytona, the BMW M Hybrid V8 features a refined front design, new yellow icon lights, and enhanced track lighting for night racing. The car debuts the Inception Design, now applied across all of our LMDh cars – uniting the entire BMW M race car family under one look.
Stop over-engineering your context.
LLMs have been extensively trained in how to use filesystems. Even if you’re not making a coding agent, it turns out that’s the most effective and natural way to manage your context.
Here’s our guide to mapping any domain, from financial analysis and support tickets to sales optimization, into a filesystem so you can build better agents ↓
https://t.co/dANfsH7aCh
New on the Anthropic Engineering Blog: Demystifying evals for AI agents.
The capabilities that make agents useful also make them more difficult to evaluate. Here are evaluation strategies that have worked across real-world deployments.
https://t.co/UD0yGglTU0
The most surprising “AI PC” move might not be a laptop at all. 🔧
AMD is launching a dedicated dev box concept: the Ryzen AI Halo Developer Platform.
It is described as a compact SFF desktop PC built around high-performance Ryzen AI Max+ Series processors, with a clear pitch for developers: leadership tokens-per-second-per-dollar.
Availability is targeted for Q2 2026.
This matters because it acknowledges a very real gap in the market for AMD: developers want a local system that can run meaningful models, iterate fast, and stay aligned with the same software story they use elsewhere.
AMD is also tying Halo to developer support across the ecosystem, which lines up with the broader push to unify experiences through ROCm.
And yes, positioning-wise, this is a competitor to Nvidia DGX Spark: a purpose-built, compact developer system designed to make local AI development feel practical, not experimental.
So what: the winners in AI are going to be the platforms that reduce friction for builders. A dev box that optimizes for tokens-per-dollar is a very “developer-first” signal.
Best YouTube Channels to learn AI in 2026:
1. Cole Medin
https://t.co/Fudp9wYUCP
2. Matthew Berman
https://t.co/KUqkIWa3B9
3. Tech With Tim
https://t.co/toAoMfdKB1
4. Two Minute Papers
https://t.co/qPkZE6LuEt
5. Skill Leap AI
https://t.co/RhJVKrgiXC
6. AI Explained
https://t.co/bWjMnPO80p
7. DeepLearningAI
https://t.co/bdwkpuud9O
8. Futurepedia
https://t.co/mEFNilA7t0
9. Tina Huang
https://t.co/ycT8PL0rLl
10. Andrej Karpathy
https://t.co/sDa5sFMpOM
🎄 Holiday gift from the Color.js team: v0.6.0 is out with a host of fixes and improvements!
We’ve also crossed 100 million installs on npm a few weeks ago!
Release notes:
https://t.co/0xrAcVFFyZ
How is AI changing work inside Anthropic? And what might this tell us about the effects on the wider labor force to come?
We surveyed 132 of our engineers, conducted 53 in-depth interviews, and analyzed 200K internal Claude Code sessions to find out.
https://t.co/YLLjs9W9e5
Saturn's famous hexagon.
This atmospheric phenomenon rotates counterclockwise at approximately 320 km/h, completing one rotation every 10 hours and 40 minutes.
The hexagon is approximately 25,000 km in diameter.
Remarkably, while storms on other planets change and disappear, this hexagon remains unchanged for at least 40 years.
The ultimate unlock... Gemini 3 + Gemini CLI has arrived 🚀🚀🚀
❯ npm install -g @google/gemini-cli@latest
Now to avoid melting too many TPUs we're going to be iteratively rolling out the support (it's quite epic):
- ✅ Available Now: Google AI Ultra subscribers and those with a paid Gemini API key.
- 🏢 Coming Soon: Gemini Code Assist Enterprise users.
- 📝 Waitlist: All other users (including Free & Standard tiers) can join the waitlist here: https://t.co/N4Vs3Wa24h
Track waitlist and rollout progress here: https://t.co/7bgZSJsq3l
If you already have access (AI Ultra or paid API key), run /settings, toggle “Preview features” to true and enjoy!
Finally check out our announcement blog: https://t.co/O7GIu7LsPY
See it in action 👇
Claude Code Weekly Roundup
Last week we landed several CLI UX polish updates (better spinner, inline bash feedback), expanded hooks options, and a new frontend-design plugin that levels up Claude's design capabilities.