---
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

### [Small Talk With Prasun Kumar, CEO and Founder of Oppex AI](https://yomu.fyi/post/small-talk-with-prasun-kumar-ceo-and-founder-of-oppex-ai.md)
- Company: [Jetbrains](https://yomu.fyi/company/jetbrains.md)
- Author: Mitsal Athaya
- Published: Sep 23, 2026
- Type: Explainer

Oppex AI builds AI agents that help developers resolve production incidents by gathering context before the engineer is paged. The agents collect logs, cloud metrics, database health, affected customers, recent code changes, and past incident history, then present a recommendation. Because developer logs are not plain English, the team constructs a knowledge graph of how the entire system connects. They harden the agent by pitting it against a chaos monkey that breaks test systems without revealing the method, forcing the agent to learn. The 15-person team has launched the product with early enterprise customers and aims to bring mean time to resolve under 10 minutes, eventually moving toward autonomous system healing.


### [\*\*Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization\*\*](https://yomu.fyi/post/know-who-spoke-when-build-real-time-multi-speaker-ai-with-nvidia-nemot.md)
- Company: [Hugging Face](https://yomu.fyi/company/hugging-face.md)
- Author: Francesco, Ivan Medennikov, Taejin Park, Jagadeesh Balam, Adi- margolin, Maryam Motamedi
- Published: Sep 23, 2026
- Type: Announcement

NVIDIA Nemotron 3 Diarization is an open-weight, 100M-parameter model that ranks #1 on VoiceArena’s Diarization-Bench leaderboard with a 14.72% Diarization Error Rate (DER). It classifies who spoke when in conversations, handling overlapping speech and supporting up to eight speakers across live and recorded audio. The model uses a Sortformer approach with arrival-order speaker labeling, a 31-layer Transformer encoder, and two memory mechanisms for streaming inference. It accepts 16 kHz single-channel audio and outputs a floating-point tensor of speaker-activity probabilities at a configurable time step. Developers can combine its timestamps with automatic speech recognition to produce speaker-attributed transcripts, with recommended input-buffer latencies ranging from 0.32 to 30.4 seconds.


### [100 Exercises to Learn Rust, Updated](https://yomu.fyi/post/100-exercises-to-learn-rust-updated.md)
- Company: [Jetbrains](https://yomu.fyi/company/jetbrains.md)
- Author: Daria Kolodzey
- Published: Sep 23, 2026
- Type: Tutorial

The post announces a major update to the free JetBrains Academy course ‘100 Exercises to Learn Rust,’ which runs inside RustRover. The course teaches Rust through hands-on exercises building a ticket management system across eight chapters from ownership to futures, assuming fluency in another language. The update adds a rebuilt introduction, an FAQ, a final challenge turning the system into an async REST API by editing Cargo.toml and verifying with a requests.http file, visible test assertions, and a dev/release profile experiment. A ‘fast track’ lets experienced users skip to the final challenge to quickly see which areas need revisiting. The post advises using AI for explanations, not code generation, and turning off AI code completion during exercises to maximize learning from getting stuck and working out solutions.


### [What Dropbox has learned from deploying AI at company scale](https://yomu.fyi/post/what-dropbox-has-learned-from-deploying-ai-at-company-scale.md)
- Company: [Dropbox](https://yomu.fyi/company/dropbox.md)
- Author: Dropbox Team
- Published: Sep 23, 2026
- Type: Explainer

Dropbox has learned that deploying AI at scale requires examining workflows end to end, addressing new bottlenecks like too many code reviews or infrastructure strain. The company uses a framework measuring speed, effectiveness, quality, and impact, with proxy metrics including pull request throughput and change failure percentage. Dropbox reports around 70% AI-generated code, comparable to top technology peers, and its pull request throughput ranks in the top 5% of a custom benchmark while change failure percentage is at the 75th percentile. The company connects AI token usage to engineering value through its internal agent service Nova, and it uses ground-up initiatives like boot camps and aligned incentives to drive adoption. Human skills like problem solving, judgment, and leadership become more critical as AI lowers the cost of producing software, and Dropbox aims to extend AI workflows beyond engineering to product, design, and customer experience teams.


### [ClickHouse release 26.9](https://yomu.fyi/post/clickhouse-release-26-9.md)
- Company: [Clickhouse](https://yomu.fyi/company/clickhouse.md)
- Published: Sep 23, 2026
- Type: Announcement

ClickHouse 26.9 ships with 56 new features, 135 performance optimizations, and 464 bug fixes. Among the highlights are conditional LIMIT boundaries that control output based on ordered stream values, incremental refreshes for append-only materialized views, and disk spilling for high-cardinality DISTINCT queries that automatically spills when memory usage reaches half the limit. The CREATE TOKEN feature issues time-limited credentials with a default lifetime of 30 minutes and restricted privileges, reducing the impact of leaked keys. Arithmetic operations between DateTime and Time types now support configurable overflow behaviors—throw, saturate, or ignore. PromQL support expands in a private preview on ClickHouse Cloud, bracket syntax simplifies JSON subcolumn access, and queries for min, max, and count receive performance improvements.


### [How to Guide Your Language Flow](https://yomu.fyi/post/how-to-guide-your-language-flow.md)
- Company: [Apple](https://yomu.fyi/company/apple.md)
- Published: Sep 23, 2026
- Type: Announcement

Probe guidance, a novel method for guiding flow matching models, is introduced. It uses the frozen internal states of an existing diffusion model to construct a guidance signal, operating on a similar principle as autoguidance but removing the need for an additional forward pass at inference time. The method provides a reliable path ensuring the weak and strong models share similar dynamics. Probe guidance sets a new state-of-the-art performance on unconditional generation when benchmarked on continuous diffusion language models. Applied to a 1.7B diffusion language model, it consistently improves multiple choice question answering benchmarks.


### [How to serve trillions of tokens for trillion-parameter coding agents](https://yomu.fyi/post/how-to-serve-trillions-of-tokens-for-trillion-parameter-coding-agents.md)
- Company: [Modal](https://yomu.fyi/company/modal.md)
- Published: Sep 23, 2026
- Type: Explainer

Serving inference for coding agents requires trillion-parameter sequence models like Moonshot AI’s Kimi K2.6 to process trillions of input and output tokens at extremely high relative and absolute performance. To be economically viable, these services must achieve large fractions of hardware peak performance, using contemporary matrix math accelerators like Tensor Cores at petaFLOP per second scale. Modal optimized by batching multiple sequences together, caching KV state, and patching SGLang with NVFP4 micro-scaling on Blackwell GPUs like the B200. Per-replica performance improved 2.8x per user and 5.6x across users, transforming ruinously expensive into a price-competitive service. One deployment processed hundreds of billions of tokens daily and trillions total, showing that careful engineering and benchmarking are key to scaling inference.


### [Rollouts and Security Review](https://yomu.fyi/post/rollouts-and-security-review.md)
- Company: [Cursor](https://yomu.fyi/company/cursor.md)
- Published: Sep 23, 2026
- Type: Announcement

Cursor launched two bots for the last mile of shipping code: Rollouts and Security Review. Rollouts watches every change as it deploys and reports health per environment, with verdicts of verified healthy, regression detected, or inconclusive. It reads the diff and systems a pull request touches, writes a monitoring plan as a PR comment, and tracks deploy events per environment. Security Review reads every pull request in the context of the codebase and posts one review comment reporting exploitable bugs, including injection, authentication bypasses, and secrets. Both bots are available today on Teams and Enterprise plans, with usage credits for the next 10 days.


### [How to test that a dozen MongoDB drivers, in a dozen languages, conform to a dozen specs](https://yomu.fyi/post/how-to-test-that-a-dozen-mongodb-drivers-in-a-dozen-languages-conform.md)
- Company: [MongoDB](https://yomu.fyi/company/mongodb.md)
- Published: Sep 23, 2026
- Type: Problem & solution

MongoDB develops a dozen native client library drivers in a dozen languages, each idiomatic to its language but needing consistent API behavior. Instead of relying on English specs or Cucumber tests, engineers created a YAML-based domain-specific test language interpreted by runners in each driver. The Unified Test Format, built on this YAML language, now contains nearly 1,500 files testing conformance to specs covering retries, transactions, client-side encryption, and more. A natural experiment showed four of five drivers had lower nonconformance bug rates after adopting YAML tests; Node.js improved after fixing its test runner. The IEEE ISSRE 2026 accepted a paper on this approach, which has saved over 22,000 lines of test-runner code across seven drivers.


### [Conversations in the AI Era: Manchester Meetup Recap](https://yomu.fyi/post/conversations-in-the-ai-era-manchester-meetup-recap.md)
- Company: [Twilio](https://yomu.fyi/company/twilio.md)
- Author: Ryan Britnell
- Published: Sep 23, 2026
- Type: Announcement

Twilio and Ciptex hosted the Conversations in the AI Era meetup in Manchester on September 3rd, 2026, drawing developers, tech leads, and founders from the local scene. Darragh O’Connor demonstrated using a SIP trunk to connect callers directly into a livestream, emphasizing solving the problem with the simplest tool. Matthew Duggan of Ciptex presented an architecture where AI and human agents share conversation state and tools, treating handoff as a change of ownership rather than a recovery path. Paul D’Ambra of PostHog discussed how cheap code generation made reviewing code the bottleneck, leading them to build AI review agents that substantially dropped their revert rate. The next meetup is scheduled for November 4th, 2026.


### [Collected, stored, and useless: The lifecycle of most customer data](https://yomu.fyi/post/collected-stored-and-useless-the-lifecycle-of-most-customer-data.md)
- Company: [Twilio](https://yomu.fyi/company/twilio.md)
- Author: Jesse Sumrak
- Published: Sep 23, 2026
- Type: Problem & solution

The post examines the typical lifecycle of customer data in companies, revealing that most data is captured but never effectively used. A survey of 660 business leaders found that 48% struggle to connect customer data across channels, making it the top obstacle over budget or headcount. Data is often captured in volume but siloed across systems, duplicated into separate records for the same customer, and reliant on stale overnight batch updates. This leads to AI agents working with outdated profiles, causing 71% of consumers to abandon conversations when the AI fails to recognize them. The post proposes shortening the path between capture and use through real-time data synchronization, exemplified by Twilio’s Conversations and Conversation Memory products.


### [Introducing Support for Local AI Models in the Antigravity SDK](https://yomu.fyi/post/introducing-support-for-local-ai-models-in-the-antigravity-sdk.md)
- Company: [Google](https://yomu.fyi/company/google.md)
- Author: Sachin Kotwani, Taylor Mullen
- Published: Sep 23, 2026
- Type: Announcement

The Antigravity SDK now supports local AI models, starting with Gemma 4 26B A4B via Google AI Edge’s LiteRT, enabling offline agentic assistance on local GPU and RAM with a recommended machine of 24GB VRAM or unified memory. Local execution eliminates API costs and rate limits while keeping code and requests on-device, offering privacy and offline resiliency. The SDK supports hybrid workflows where a cloud model like Gemini 3.8 Flash plans while local Gemma 4 instances execute, as in a code auditing task where 97.2% of tokens (3,322 of 3,417) ran locally. A built-in example generates a live-updating CLI resource monitor using psutil and rich from a single prompt, with the agent autonomously writing the script, creating requirements.txt, and testing the code entirely offline. The SDK also supports any OpenAI-compatible server like Ollama, LM Studio, or vLLM via LocalOpenAIAgentConfig without changing agent orchestration.


### [Database Branching: A Developer's Guide to Git-Style Workflows](https://yomu.fyi/post/database-branching-a-developer-s-guide-to-git-style-workflows.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Databricks Staff
- Published: Sep 22, 2026
- Type: Explainer

Database branching creates isolated database environments from a known parent state using copy-on-write, where branches share unchanged data and only store modifications. For a large parent database like one that is 40 GB, two branches may require just 5.6 MB of additional storage instead of 80 GB with traditional full copies. Developers can test schema migrations like adding a NOT NULL column to a table with millions of rows against realistic data and constraints, CI pipelines provision per-pull-request branches for isolated tests, and AI agents experiment across short-lived environments without risking the parent database. Safe operation relies on protecting parent branches from writes and deletions, restricting sensitive data in disposable branches, automating creation and cleanup, and keeping environments reproducible through version-controlled configuration. Databricks Lakebase provides database branching, enabling teams to adopt Git-style workflows for development, CI/CD, and AI agent use cases.


### [AI Agent Observability: Making Silent Production Failures Visible and Actionable](https://yomu.fyi/post/ai-agent-observability-making-silent-production-failures-visible-and-a.md)
- Company: [Salesforce](https://yomu.fyi/company/salesforce.md)
- Author: Scott Nyberg
- Published: Sep 22, 2026
- Type: Announcement

Agentforce Health Monitoring (AHM) addresses silent availability failures in production AI agents that traditional monitoring misses because agents can appear healthy even when unreachable. The team unified telemetry from approximately five or six locations across sessions, reasoning steps, tool calls, errors, and escalations into a coherent representation. End-to-end alerting latency was reduced from roughly 20 minutes to several minutes by converting Data 360 ingestion to streaming and simplifying query complexity. AHM provides 16+ built-in metrics including agent availability, error rate, response/escalation rate, engagement, and latency, with a drill-down interface connecting alerts to session context. The roadmap includes cost, TTFT/TTLT, toxicity, feedback metrics, RAG Quality Monitoring integration, and eventually a runtime insight agent for root-cause analysis and automated remediation.


### [The Genie One MCP is now Generally Available](https://yomu.fyi/post/the-genie-one-mcp-is-now-generally-available.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Ben Tripp, Sydney Sundell
- Published: Sep 22, 2026
- Type: Announcement

Databricks has announced general availability of the Genie One Model Context Protocol (MCP) server, now a managed service within Unity Gateway with centralized governance, fine-grained policies, and audit logging across every invocation. The MCP provides a single governed interface for any AI agent to access structured and unstructured data from Genie One, grounded in governed business context from Genie Ontology, eliminating agent sprawl and inconsistent answers. It exposes tools for asking questions, getting query results, checking incremental progress, and steering responses, with an optional MCP App for real-time interactive visualizations and Genie Ontology citations. Connected agent clients automatically interpret domain semantics via Genie Ontology across relational data and document repositories without custom connectors. Use cases include enriching slide creation with richer data, bringing customer usage context to outreach agents, and integrating business definitions into developer workflows, as noted by GetYourGuide.


### [Genie One MCP: Give any AI Agent the Right Business Context](https://yomu.fyi/post/genie-one-mcp-give-any-ai-agent-the-right-business-context.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Krishna Rupanagunta, Cynthya Peranandam, Siddesh Pore
- Published: Sep 22, 2026
- Type: Explainer

Business leaders often have access to plenty of data but still cannot get a reliable answer to a simple question like why net revenue declined 8% at their largest account last week. The answer may span sales, finance, promotions, inventory, and account data, with each source potentially correct in isolation yet different in definitions, detail, relationships, and authority. Without the right business context, general-purpose agents cannot reliably determine what metrics mean, which sources are authoritative, how data should connect, or what each user is allowed to see. Genie One MCP gives governed context to any MCP-compatible AI agent, connecting assistants like ChatGPT, Claude, Microsoft Copilot, and coding agents to Genie One. This lets them answer business questions using approved definitions, trusted data relationships, and permission-aware access controls, with traceable, permission-aware answers grounded in governed business context.


### [Enabling Private High-Performance Production AI Inference with NVIDIA Confidential Computing](https://yomu.fyi/post/enabling-private-high-performance-production-ai-inference-with-nvidia.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Tanya Lenz
- Published: Sep 22, 2026
- Type: Explainer

NVIDIA Confidential Computing (CC) enables secure AI inference by using memory-encrypted confidential VMs, confidential GPUs, and encrypted NVLink. Running inference frameworks like TensorRT LLM in a CC environment introduces performance overhead because secure execution changes assumptions about memory movement, timing, and multi-GPU communication. To quantify this overhead, the post presents a controlled methodology comparing CC-on and CC-off performance on a workload with long input context, extended output generation, and low concurrency. With CC-aware adaptations—including pageable memory selection, asynchronous readback, GPU %globaltimer for autotuning, and CC-aware multi-GPU communication—TensorRT LLM retained 96.1–98.2% of baseline throughput on eight NVIDIA B200 GPUs while keeping per-token latency overhead below 5%. The post concludes that security configuration and inference optimization should be treated as a single deployment problem for organizations moving private inference into production.


### [Topology-Aware Workload Scheduling with NVIDIA Topograph](https://yomu.fyi/post/topology-aware-workload-scheduling-with-nvidia-topograph.md)
- Company: [NVIDIA](https://yomu.fyi/company/nvidia-developer-blog.md)
- Author: Elizabeth Goodman
- Published: Sep 22, 2026
- Type: Explainer

AI factories are power-limited systems where GPU workload placement is a key optimization; poor placement fragments topology domains and forces traffic across shared links, reducing throughput and raising job costs. NVIDIA Topograph discovers cluster topology from cloud APIs or on-premises fabric systems, normalizes it into a common model, and publishes it in formats each workload manager expects, such as Kubernetes node labels, Slurm topology configuration, or Slinky ConfigMaps. It uses five components: API Server, Node Observer, Node Data Broker, Provider, and Engine to keep the topology view current as the cluster changes. The post covers deploying Topograph with Helm on Kubernetes, as a system package on Slurm, and as a Helm chart on Slinky, with support for cloud providers including Google Cloud, Lambda, Nebius, Nscale, and OCI. Topograph works alongside Dynamic Resource Allocation and KAI Scheduler to enable topology-aware gang scheduling, improving AI factory efficiency and tokens per watt.


### [How Fountain rebuilt its data plane on ClickHouse Cloud to power Cue, the Frontline Superintelligence](https://yomu.fyi/post/how-fountain-rebuilt-its-data-plane-on-clickhouse-cloud-to-power-cue-t.md)
- Company: [Clickhouse](https://yomu.fyi/company/clickhouse.md)
- Published: Sep 22, 2026
- Type: Announcement

Fountain rebuilt its data plane on ClickHouse Cloud to power Cue, its agentic AI platform for frontline hiring, moving off a multi-vendor BigQuery and S3/Iceberg batch stack. The migration cut analytics latency from three hours to under two minutes, with most queries returning in under a second, at roughly 66% lower cost. Data from 13 Postgres production databases and four MongoDB databases streams into ClickHouse Cloud via ClickPipes managed CDC, landing as Iceberg and Parquet in S3 before reaching BigQuery and ClickHouse. The new architecture uses two symmetrical pipelines—one for Postgres, one for MongoDB—each running a short path: source, ClickPipes, materialized views, and Cue. The system now carries 8.48 billion CDC rows replicated continuously by ClickPipes and 10.3 billion analytics rows across 15 deployments, served by over 2,000 incremental materialized views.


### [We just shipped support for the ugliest part of HTTP: Vary](https://yomu.fyi/post/we-just-shipped-support-for-the-ugliest-part-of-http-vary.md)
- Company: [Cloudflare](https://yomu.fyi/company/cloudflare.md)
- Author: Zaidoon Abd Al Hadi
- Published: Sep 22, 2026
- Type: Announcement

The Vary response header, often called the ugliest part of HTTP, is now supported in Cache Rules on every Cloudflare plan. Vary tells caches which request fields may affect a response, but not which differences actually matter, leading to cache fragmentation or incorrect responses. For example, a server might deliver different image formats to different browsers, and a cache ignoring Vary risks serving the wrong bytes. The origin still names the request headers that may affect a response, but Cloudflare lets users decide how to handle each one. This support addresses the challenge of balancing correct caching with cache efficiency.


[Newer posts](https://yomu.fyi/posts/page/35.md) · [Older posts](https://yomu.fyi/posts/page/37.md)