Newsdesk
Engineering
Data & AI
Industries
Enterprise Systems
Go-to-Market
Longform
DesignIndiaAll stories

Craft & Leadership

07 November 2026

45 new posts across 8 practice and leadership sources — 5 worth your time.

One week of Craft & Leadership, 5 stories, as published.

Architecture

Uber Builds MCP Gateway to Govern Access to 800+ Servers and 5,000+ AI Agent Tools

Uber built an MCP Gateway to standardize how AI agents discover and invoke services across 800+ MCP servers and 5,000+ tools. The system separates tool registration from runtime execution and includes automated API discovery, fine-grained access controls, sensitive data redaction, and response filtering.

Why it matters — Decouple agent tool registration from execution to centralize access control, payload redaction, and runtime discovery across large tool ecosystems.

InfoQ Architecture · 07 November 2026 · Read the original →

Architecture

Article: Systemic Approach to Memory, Knowledge, and Context in Agentic AI Architectures

This article outlines a systemic context engine architecture designed to select, inject, compress, and expire data before passing it to LLMs. It defines an operational framework separating context, memory, and knowledge using tiered storage, curated data, and token budgeting to prevent anti-patterns like context stuffing.

Why it matters — Manage LLM context lifecycle via tiered memory and token budgeting rather than raw prompt context stuffing.

InfoQ Architecture · 07 November 2026 · Read the original →

Testing & Quality

Test and verification technique effectiveness with GPT-6 Astra

Dan Luu analyzes empirical testing performance when LLM agents are directed to use specific software verification techniques versus open-ended defaults. The evaluation notes that unguided agents frequently generate low-value unit tests compared to structured verification methods.

Why it matters — Explicitly constraining LLM verification methods prevents agents from defaulting to superficial unit tests.

Dan Luu · 07 November 2026 · Read the original →

Architecture

The Data Product Platform Navigator

This piece introduces a platform navigator based on the CNCF Platform Engineering Maturity Model to help organizations address low adoption of self-service data platforms. It outlines how to structure self-service data mesh platforms to scale across teams at different maturity levels.

Why it matters — Assess organizational platform maturity against standard models to diagnose and fix adoption bottlenecks in self-service platforms.

Martin Fowler · 07 November 2026 · Read the original →

Opinion

Fragments: October 28

The post details Andrej Karpathy's breakdown of the widening gap in how different groups experience LLM capabilities. It contrasts the public's view of simple chatbots against the modest gains of average professionals and the vertical productivity curve achieved by advanced practitioners.

Why it matters — AI tooling gains are non-linear; organization-wide productivity improvements depend on elevating engineer fluency rather than merely distributing access to tools.

Martin Fowler · 07 November 2026 · Read the original →

We are already building governance layers for tools nobody has figured out how to use.

5 stories, every Saturday

Published here every week. Follow by RSS to get it as it lands.

← Previous issue Next issue →