---
title: "Latest reads"
description: "The engineering internet, summarised so you can actually read it."
---

# Latest reads
> The engineering internet, summarised so you can actually read it.

## Articles

### [Code Quality Q&A With the JetBrains Qodana Team](https://yomu.fyi/post/code-quality-q-a-with-the-jetbrains-qodana-team.md)
- Company: [Jetbrains](https://yomu.fyi/company/jetbrains.md)
- Author: Kerry Beetge
- Published: Sep 22, 2026
- Type: Explainer

The Qodana team answers frequently searched questions about code quality, distinguishing technical debt from poor code quality and offering practices to reduce cognitive complexity in nested loops. They explain that code coverage measures how much production code is executed by unit tests, while mutation testing evaluates test quality by checking whether tests detect deliberate code changes. Unit tests provide the fastest feedback on code changes by catching regressions close to their introduction and making refactoring safer. To track improvement, teams should monitor static analysis trends such as new issues and technical debt, and also look for fewer Sev 1 issues in production and faster mean time to repair. False positives in automated scanning can be reduced by giving the tool more repository context, such as excluding inspections in qodana.yaml or silencing them with inline comments.


### [ClickHouse appoints Mike Scarpelli, former Snowflake and ServiceNow CFO, to Board of Directors](https://yomu.fyi/post/clickhouse-appoints-mike-scarpelli-former-snowflake-and-servicenow-cfo.md)
- Company: [Clickhouse](https://yomu.fyi/company/clickhouse.md)
- Published: Sep 22, 2026
- Type: Announcement

ClickHouse appointed Michael Scarpelli, former CFO of Snowflake and ServiceNow, to its Board of Directors as an independent director and chair of the Audit Committee. Scarpelli previously guided three companies through initial public offerings and said ClickHouse is redefining the category for real-time and agentic workloads. The company has surpassed $350 million in annualized revenue, up from roughly $200 million at the end of 2025, and achieved a $15 billion valuation in its Series D round, with over 4,000 customers including DoorDash, Meta, Tesla, and Visa. ClickHouse now employs nearly 800 people across 27 countries, plans to reach 1,000 by year-end, and derives more than half its revenue from outside the United States as it scales its enterprise motion. The appointment is part of a deliberate build-out of finance and governance capabilities, including the 2025 hire of CFO Jimmy Sexton, who also came from Snowflake.


### [Data liberation: Apache Kafka's native cluster mirroring](https://yomu.fyi/post/data-liberation-apache-kafka-s-native-cluster-mirroring.md)
- Company: [Red Hat](https://yomu.fyi/company/red-hat.md)
- Author: Federico Valeri
- Published: Sep 22, 2026
- Type: Announcement

KIP-1279 introduces native cluster mirroring, embedding cross-cluster replication directly into the Kafka broker. This eliminates external tooling, offset translation tables, and recompression, as destination brokers fetch committed records using the standard fetch protocol and append them byte for byte. The approach preserves offsets, compression, and consumer group state, simplifying failover to stopping the mirror and redirecting clients. The architecture uses mirror fetcher threads inside the broker, extending AbstractFetcherThread, and leverages existing client quota mechanisms. Migration involves creating a new KRaft cluster, setting up a mirror, and starting mirroring, with future work including synchronous mirroring and tiered storage integration.


### [Extend OpenShift Lightspeed with Red Hat Satellite, MCP, and BYOK](https://yomu.fyi/post/extend-openshift-lightspeed-with-red-hat-satellite-mcp-and-byok.md)
- Company: [Red Hat](https://yomu.fyi/company/red-hat.md)
- Author: Jose Manuel Marquez, Josep Andreu
- Published: Sep 22, 2026
- Type: Tutorial

The article demonstrates extending Red Hat OpenShift Lightspeed by integrating the Red Hat Satellite MCP server and bring your own knowledge (BYOK). The Satellite MCP server exposes live operational data, such as kernel versions, CVEs, and errata, through the Model Context Protocol. BYOK enriches the assistant with private enterprise knowledge, enabling retrieval-augmented generation (RAG) to combine internal documentation with infrastructure data. The integration requires installing the OpenShift Lightspeed operator and deploying the Satellite MCP server in the openshift-lightspeed project. Authentication is handled by storing the header value in a Secret, and the MCP server expects clients to include appropriate Accept headers for the text/event-stream media type.


### [Introducing Worker Previews: isolated preview environments for every change your agent makes](https://yomu.fyi/post/introducing-worker-previews-isolated-preview-environments-for-every-ch.md)
- Company: [Cloudflare](https://yomu.fyi/company/cloudflare.md)
- Author: Nanda Syahrasyad
- Published: Sep 22, 2026
- Type: Announcement

Worker Previews gives every Git branch its own production-like environment with an isolated URL, configuration, state, and observability. Developers deploy a Preview using npx wrangler preview, which creates a separate Worker version with its own variables, secrets, and bindings, distinct from production, on a stable URL that updates with each push. Durable Objects and Containers automatically receive new namespaces per Preview, preventing state contamination across branches during migrations or schema changes. Each Preview supports full observability—logs, errors, metrics, and traces—scoped to that branch, enabling agents to autonomously test, catch failures, push fixes, and verify before production. The feature enables an Agent Development Lifecycle where every change is atomic, independently deployable, and revisable before merging to production.


### [Accelerating a ROS 2 Node with an AI Agent and NVIDIA Isaac ROS](https://yomu.fyi/post/accelerating-a-ros-2-node-with-an-ai-agent-and-nvidia-isaac-ros.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Tanya Lenz
- Published: Sep 22, 2026
- Type: Tutorial

GPU acceleration alone does not guarantee fast ROS 2 graphs because messages may still be serialized or copied through CPU memory. NVIDIA contributed the rosidl::Buffer abstraction and CUDA buffer backend to ROS 2 Lyrical, enabling zero-copy transport of GPU-resident payloads between co-located nodes when they share the same host, CUDA device, Linux user, and a supported RMW implementation, with automatic CPU fallback otherwise. An AI coding agent using the migrate-node-to-rosidl-buffer skill can audit an existing CUDA-accelerated node by tracing data movement, running a copy-boundary audit, and implementing a minimal interface-preserving patch. The tutorial demonstrates this on the Depth Anything 3 TensorRT node, updating subscription options and message storage to use CUDA buffer backend without changing the sensor\_msgs/msg/Image message definition. The resulting accelerated workload can be deployed on NVIDIA Jetson AGX Thor, preserving standard ROS 2 messages while avoiding unnecessary serialization and host copies.


### [JetBrains Air: Building a System of Products for Agentic Software Development](https://yomu.fyi/post/jetbrains-air-building-a-system-of-products-for-agentic-software-devel.md)
- Company: [Jetbrains](https://yomu.fyi/company/jetbrains.md)
- Author: Kirill Skrygan
- Published: Sep 22, 2026
- Type: Announcement

JetBrains Air is a system of products for agentic software development, built on the premise that no single model, agent, or service fits every developer, team, or task. It connects products for individual work, team coordination, organizational control, context, and process automation, while remaining open to tools and agents developers choose, including those JetBrains does not build. This openness lets developers select the agent, model, or service suited to a task without forcing the organization to give up context, visibility, or control. The Post notes that every new model, agent, or service an organization adds reduces visibility into its own development work. Because intelligence does not belong to one model, the benefit can extend across supported agents and services.


### [What is AI analytics? Why it only works on governed data](https://yomu.fyi/post/what-is-ai-analytics-why-it-only-works-on-governed-data.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Richard Tomlinson
- Published: Sep 22, 2026
- Type: Explainer

AI analytics applies AI and machine learning to data analysis, enabling natural-language queries and automated insights without manual query-building, shifting from retrospective dashboards toward agentic investigations. Unlike traditional BI, which requires users to navigate predefined dashboards and filters, AI analytics systems understand intent, determine relevant data and business context, and can conduct multi-step investigations on behalf of the user. A key challenge is that AI analytics only works when grounded in governed data: without shared metric definitions, authoritative sources, and permission-aware execution, the system produces untrustworthy results. Trust in AI analytics is a system property, not a model property; it requires governed data, shared business context, permission enforcement, and verifiability of queries and calculations. Better AI models do not eliminate the need for business context; as AI becomes more autonomous, semantic definitions and governance become more critical.


### [Data Ontology defined: The context layer your AI agents are missing](https://yomu.fyi/post/data-ontology-defined-the-context-layer-your-ai-agents-are-missing.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Richard Tomlinson
- Published: Sep 22, 2026
- Type: Explainer

The Post argues that AI agents are breaking a long-standing assumption in enterprise data: that a human analyst can always resolve ambiguity. Richard Tomlinson explains that while analysts rely on tribal knowledge—like which revenue table is trusted or what ‘active customer’ means—agents lack this context. This gap causes inconsistent answers and erodes trust in self-service tools. The conversation examines why previous ontology efforts stalled and what must be true for an ontology to support agents. The ontology is positioned as the missing context layer that companies have not yet built.


### [Resilient Edge AI: Keeping Connected Vehicles Smart Offline](https://yomu.fyi/post/resilient-edge-ai-keeping-connected-vehicles-smart-offline.md)
- Company: [MongoDB](https://yomu.fyi/company/mongodb.md)
- Published: Sep 22, 2026
- Type: Explainer

Connected vehicles face intermittent connectivity in tunnels, rural roads, and underground garages, yet drivers expect reliable AI assistance during those moments. The post argues that offline capability is a requirement, not a fallback, for trustworthy in-vehicle AI. It proposes three principles: design for the moment of need, treat vehicle knowledge as an evolving product asset, and separate availability from connectivity. A reference architecture uses ObjectBox for local persistence and synchronization with MongoDB Atlas, which provides the cloud foundation for shared knowledge and updates. The enduring pattern is to keep critical interactions available at the edge while using the cloud for scale, coordination, and continuity when connectivity returns.


### [From 2D Drawings to 3D Models: Designing an AI Reconstruction Pipeline on MongoDB](https://yomu.fyi/post/from-2d-drawings-to-3d-models-designing-an-ai-reconstruction-pipeline.md)
- Company: [MongoDB](https://yomu.fyi/company/mongodb.md)
- Published: Sep 22, 2026
- Type: Explainer

BIRD is a research framework that converts five 2D perspective drawings into a proportionate, dimensioned 3D building envelope. It uses a dual-branch CNN to separate architectural line segments from noise, a text recognition model to scale them to real-world inches, and a wrapping algorithm to fold the elevation views around the floor plan. The framework achieved a median accuracy of roughly 93% across complete building plan sets from small residences to a multi-story apartment complex. The proposed deployment on MongoDB Atlas stores each project as a single JSON document containing all drawings’ segments, corrections, and final geometry, with S3 for image storage and Atlas Stream Processing for event-driven workflow coordination. MongoDB Search and Vector Search with Voyage AI embeddings enable full-text and semantic search over past projects, allowing drafters to start from corrected models of similar builds.


### [Canary rollouts: upgrade models in production without downtime](https://yomu.fyi/post/canary-rollouts-upgrade-models-in-production-without-downtime.md)
- Company: [Together AI](https://yomu.fyi/company/together-ai.md)
- Published: Sep 22, 2026
- Type: Announcement

Deploying a new model checkpoint in production typically requires a tradeoff: a hard swap exposes all users to regressions at once, while a DIY staged cutover relies on manual traffic shifting. Together AI’s rollout system automates this by moving traffic between source and target deployments through configurable canary steps, such as 5%, 25%, 50%, and 100%. Each step runs health checks before shifting traffic, enforces a propagation wait for routing caches, and drains the source only after traffic has moved. Optional metric gates, like p95 latency, run after each step; if a gate trips, the rollout pauses automatically and waits for human action. In a real canary from Qwen2.5-7B to Qwen3.5-9B, a 137% p95 regression was caught at 10% traffic, allowing reversal with no failed requests.


### [The 2026 CDP Shift: Why AI Agents Are Only as Smart as Your Data Infrastructure](https://yomu.fyi/post/the-2026-cdp-shift-why-ai-agents-are-only-as-smart-as-your-data-infras.md)
- Company: [Twilio](https://yomu.fyi/company/twilio.md)
- Author: Cynthia Goldsworthy
- Published: Sep 22, 2026
- Type: Problem & solution

The 2026 Customer Data Platform Report identifies a critical enterprise challenge: while AI tools grow individually smarter, fragmented customer data causes a worsening, not better, experience. The shift is from computational capacity to contextual continuity, as disconnected pipelines force AI agents to start every interaction from scratch. Three failures arise from fragmented systems: agent amnesia, escalating operational costs from wasted tokens, and trapped intelligence that never reaches active workflows. Modern CDPs have evolved from passive audience builders into API-first context engines that support three pillars: pre-execution context with working memory, unified behavioral and conversational data streams, and infrastructure-level governance enforcing compliance before data reaches AI models. Organizations like Twilio, Trustpilot, and Quartz demonstrate that embedding real-time customer context into the conversation layer outperforms deploying the smartest standalone models.


### [How to design an AI-to-human handoff that preserves context](https://yomu.fyi/post/how-to-design-an-ai-to-human-handoff-that-preserves-context.md)
- Company: [Twilio](https://yomu.fyi/company/twilio.md)
- Author: Jesse Sumrak
- Published: Sep 22, 2026
- Type: Problem & solution

AI-to-human handoffs often fail because context is lost during transfer, forcing customers to repeat themselves. 76% of consumers report that human agents have little to no context about their issue, and 91% of brands have seen customer satisfaction drop due to AI missteps, with 40% attributed to context loss between channels. A cold transfer drops the customer into a new queue with nothing carried over, while a warm transfer briefs the receiving agent beforehand. Five steps to preserve context include defining escalation triggers early, specifying a payload with conversation history and customer profile, generating a three-line summary, maintaining context across channel changes, and writing back learned information to customer records. Measuring repeat-explanation rate and post-handoff resolution rate helps identify whether handoffs are working or if the AI is over-holding conversations.


### [How Stripe is designing Checkout for AI agents](https://yomu.fyi/post/how-stripe-is-designing-checkout-for-ai-agents.md)
- Company: [Stripe](https://yomu.fyi/company/stripe.md)
- Author: Cara Mecozzi
- Published: Sep 22, 2026
- Type: Problem & solution

Stripe implemented WebMCP on its Checkout UIs, which processes around 0.45% of global GDP, to optimize the buying experience for AI agents. Previously, agents relied on slow browser automation, consuming an average of 1.8 million tokens and 38.83 tool calls per transaction, taking over 2.5 minutes. WebMCP exposes MCP tools from the client side, allowing agents to execute actions like get\_order\_summary or fill\_payment\_form instead of interpreting the DOM. The implementation uses progressive tool disclosure, revealing only relevant tools; in evaluations across six models, agents using WebMCP consumed 42% fewer tokens, made 38% fewer tool calls, and completed checkout 39% faster. Stripe reused existing UI state and forms via imperative and declarative APIs, avoiding a separate agent-only code path and keeping human and agent experiences aligned.


### [How UK AISI and EvalEval Are Making Benchmark Results Reproducible](https://yomu.fyi/post/how-uk-aisi-and-evaleval-are-making-benchmark-results-reproducible.md)
- Company: [Hugging Face](https://yomu.fyi/company/hugging-face.md)
- Author: Avijit Ghosh, Jenny Chim, Deep Joshi, Srishti, Matt Kennedy, Irene Solaiman, Jessica McFadyen, Lynn Tan, Coz
- Published: Sep 22, 2026
- Type: Announcement

The UK AI Security Institute (AISI) has begun using EvalEval’s Every Eval Ever (EEE) schema and Evaluation Cards platform to openly share LLM evaluation results. This collaboration, stemming from a joint workshop at NeurIPS 2025, aims to make benchmark results more reproducible by providing standardized configuration and context alongside verified scores. AISI released results for five benchmarks—HealthBench, FrontierMath, Humanity’s Last Exam, SWE-Bench Pro, and Terminal-Bench 2.0—across six frontier models, plus two cyber evaluations using a partially overlapping model set. The release accompanies a paper studying how inference-time compute and evaluation protocol affect performance, revealing that models continued solving tasks when given correctness feedback. By adopting a shared reporting schema, evaluators can compare findings across studies and understand how setup choices influence reported performance, while EvalEval invites model developers, evaluation creators, and researchers to contribute to the shared infrastructure.


### [Confused Deputy: The Old Bug That AI Agents Keep Reintroducing](https://yomu.fyi/post/confused-deputy-the-old-bug-that-ai-agents-keep-reintroducing.md)
- Company: [Auth0](https://yomu.fyi/company/auth0.md)
- Author: Andrea Chiarelli
- Published: Sep 22, 2026
- Type: Explainer

The Confused Deputy problem, first documented in 1988, describes a program with legitimate authority being tricked into misusing that authority on behalf of a less-privileged actor. AI agents are particularly vulnerable because they process instructions and untrusted content through the same natural-language channel, making prompt injection straightforward. The post traces this pattern through CSRF, SSRF, and recent attacks like Clinejection, where an issue-triage bot executed an attacker-supplied command from a crafted issue title. That attack led to a poisoned GitHub Actions cache and a compromised npm package downloaded 4,000 times. The fix involves task-scoped short-lived credentials, capability-scoped permissions, credential isolation from the model, and human-in-the-loop approval for high-risk actions.


### [Data Sovereignty by Design: Five Practical Evaluation Lenses for IT Leaders](https://yomu.fyi/post/data-sovereignty-by-design-five-practical-evaluation-lenses-for-it-lea.md)
- Company: [MongoDB](https://yomu.fyi/company/mongodb.md)
- Published: Sep 22, 2026
- Type: Explainer

Data sovereignty is a market expression, not a universal legal standard, and IT leaders should evaluate it through architecture and governance questions rather than a single label. The assessment covers where data resides, who can access it, how it moves, and whether control is retained, with considerations varying by jurisdiction, industry, and risk posture. Physical location is only one part of a broader control boundary that includes operations, key custody, and access paths. The article proposes five technical lenses for evaluation: jurisdiction-native data placement, in-use encryption, customer-held keys, network isolation and identity-based access, and forensic-grade audit evidence. These lenses are practical reference points for architecture decisions, not a legal test or certification checklist, and organizations should assess their own requirements with qualified counsel.


### [The Headless Dashboard Initiative](https://yomu.fyi/post/the-headless-dashboard-initiative.md)
- Company: [Resend](https://yomu.fyi/company/resend.md)
- Published: Sep 22, 2026
- Type: Announcement

Resend announced a headless dashboard initiative to make all dashboard capabilities available programmatically via the MCP, CLI, and API. Recent releases include the Email Metrics API for deliverability and engagement data, a Headless Webhook API for listing events and replaying deliveries, and a Cancel Broadcast API. The Duplicate Broadcast API copies any broadcast into a draft, and the Share Email API provides expiring read-only email links. The team aims to make the dashboard best-in-class while committing to full headless operation for every future entity. Developers can start building today by connecting the MCP server or using the CLI.


### [Transformers now runs llama.cpp quants](https://yomu.fyi/post/transformers-now-runs-llama-cpp-quants.md)
- Company: [Hugging Face](https://yomu.fyi/company/hugging-face.md)
- Author: Marc Sun, Arthur Zucker, Lysandre
- Published: Sep 22, 2026
- Type: Announcement

Transformers now supports running GGUF quantized models locally, reusing llama.cpp’s ggml kernels for efficient inference on Apple Silicon. The GGUF format packages weights, tokenizer, and chat template in one file, with quantization levels like Q4\_K\_M that trade precision for memory. Users can load a checkpoint from the Hub with from\_pretrained and generate text using the standard Transformers API, or serve it via an OpenAI-compatible endpoint. The initial focus is on the Qwen3.5 architecture, with benchmarks on a MacBook Pro M2 Max showing token-generation rates close to llama.cpp, including prefill time. The integration uses packed quantization kernels and generation-loop optimizations to reduce overhead, though it is currently limited to single conversations on MPS without batching support.


[Newer posts](https://yomu.fyi/posts/page/36.md) · [Older posts](https://yomu.fyi/posts/page/38.md)