---
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

### [Prioritize your messages at scale with the Twilio Traffic Optimization Engine](https://yomu.fyi/post/prioritize-your-messages-at-scale-with-the-twilio-traffic-optimization.md)
- Company: [Twilio](https://yomu.fyi/company/twilio.md)
- Author: Will McKenzie
- Published: Oct 1, 2026
- Type: Announcement

Twilio’s Traffic Optimization Engine is a new suite for global messaging deliverability and prioritization, tackling carrier network strain from soaring volumes like the 6.9 billion messages during Cyber Week 2025. The suite automates months of manual planning, working across networks, senders, geographies, and use cases to deliver high-priority messages on time. Real-time Routing uses patented algorithms to optimize traffic across 4,800+ carrier connections every 75 seconds, while Market Throughput manages rate limits at the country and sender-type level to prevent throttling. Traffic Shaping creates dynamic lanes to prioritize one-time passwords over promotional messages, and Multitenancy fairly distributes throughput among ISV teams. Insights provides real-time analytics on deliverability and capacity allocation, and the overall system uses traffic shaping and priority queueing based on carrier, sender, and message type capabilities.


### [VM Sandboxes: Full computers for agents](https://yomu.fyi/post/vm-sandboxes-full-computers-for-agents.md)
- Company: [Modal](https://yomu.fyi/company/modal.md)
- Published: Oct 1, 2026
- Type: Announcement

Modal has launched VM Sandboxes in general availability, providing agents with full Linux VMs accessible via a single flag. The VMs support Docker stacks, local databases, dev servers, and other real-machine features that agents increasingly require, as demonstrated by Ramp’s internal agent driving the entire software development lifecycle. Modal built a custom runtime on the Rust-based Cloud Hypervisor, layering innovations in host filesystems, lazy image loading, memory bursting, and snapshotting to preserve the same APIs and sub-second cold starts as their existing gVisor Sandboxes. Early customers like Linear, Legora, and Snorkel have launched over 20 million VMs, using them for coding sessions, legal agent evaluations, and agent simulations that require multiple containers and real-world environments. Modal recommends using VM Sandboxes when workloads need Docker, FUSE filesystems, or niche kernel features, while gVisor remains the default for compatible workloads.


### [Runtime Roundup: VM Sandboxes, Multi-node clusters, and more](https://yomu.fyi/post/runtime-roundup-vm-sandboxes-multi-node-clusters-and-more.md)
- Company: [Modal](https://yomu.fyi/company/modal.md)
- Published: Oct 1, 2026
- Type: Announcement

Modal announced several new features at its inaugural Runtime conference. VM Sandboxes provide agents with full Linux computer access, supporting Docker stacks, databases, graphical environments, and kernel experimentation, already used at Linear, Legora, and Snorkel for coding agents and evals. Sandbox Sidecars run alongside main sandboxes with isolation but share a host, allowing agent code in the main sandbox while trusted operations run in a sidecar the agent cannot access, avoiding network round trips in execution-heavy loops. Modal Clusters, now generally available via @modal.clustered, enable multi-node jobs for training or serving trillion-parameter models, powering post-training at Decagon and real-time inference at Runway. Sticky Sessions maintain client state by routing to the same container using a session token, critical for voice model serving, while Endpoint Candidates allow mirroring production traffic to compare quality, latency, throughput, and cost before promoting.


### [Sidecars: A low-latency trust boundary for Sandboxes](https://yomu.fyi/post/sidecars-a-low-latency-trust-boundary-for-sandboxes.md)
- Company: [Modal](https://yomu.fyi/company/modal.md)
- Published: Oct 1, 2026
- Type: Announcement

Modal introduces Sidecars, isolated containers that run alongside a main Sandbox on the same host, establishing a trust boundary between trusted and untrusted code. Sidecars enable communication across trust boundaries at least 3x faster than using separate Sandboxes, reducing latency for operation-heavy agent workloads. The feature addresses the security risk of running an agent’s harness and tool calls together, which creates the “lethal trifecta” of private data, untrusted content, and external communication. Existing solutions either move the harness to a remote service (adding network latency) or rely on egress controls that lack flexibility. Sidecars allow trusted services such as credential injection, proxy traffic, and harness logic to run locally alongside the Sandbox, with independent network policies and dynamic creation via the Modal SDK.


### [RLTL;DR: Self-Improvement by Internalizing Self-Generated Feedback](https://yomu.fyi/post/rltl-dr-self-improvement-by-internalizing-self-generated-feedback.md)
- Company: [Apple](https://yomu.fyi/company/apple.md)
- Published: Oct 1, 2026
- Type: Announcement

Standard reinforcement learning with verifiable rewards (RLVR) fails when tasks are so difficult that agents have no chance of success and no teacher models exist. RLTL;DR addresses this by having the policy write its own TL;DR insight after each failed attempt, conditioning the next rollout on all previous insights. Backpropagation on the in-context insights internalizes a direct task-to-insight mapping. On challenging tool-calling and coding datasets filtered to Pass@128=0, standard GRPO training stays flat at 0-1% Pass@1, while RLTL;DR achieves 14-31% Pass@1 with insights in context during training and 12-13% without at eval time. Reducing to SFTL;DR, training on only 4k (task, insight) tuples without rollouts, recovers almost full performance of RLTL;DR and classical SFT, demonstrating a promising compacted training paradigm.


### [How Much of a Harness Does a Strong Agent Need for Autonomous ML Engineering?](https://yomu.fyi/post/how-much-of-a-harness-does-a-strong-agent-need-for-autonomous-ml-engin.md)
- Company: [Apple](https://yomu.fyi/company/apple.md)
- Published: Oct 1, 2026
- Type: Problem & solution

Recent autonomous machine learning engineering agents increasingly rely on elaborate machinery such as multi-agent orchestrators and dedicated retrieval subagents, motivated by progress stagnation over long-horizon cycles and limited Large Language Model primitives. This paper finds that under an equal time budget and the same frontier LLM backbone, open-source state-of-the-art harnesses provide no advantages over a single session of a minimal-harness coding agent baseline that uses direct read, write, and bash primitives. The authors argue via large-scale systematic ablation studies that the machinery layers become redundant in the coding agent setting. The backbone, not the harness complexity, is the primary driver for performance. Consequently, the effort spent elaborating hand-crafted harnesses around strong models yields poor returns for current MLE benchmarks.


### [Read Restrictions and Catalog Labels: Unifying governance across engines and catalogs](https://yomu.fyi/post/read-restrictions-and-catalog-labels-unifying-governance-across-engine.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Daniel Weeks, Ryan Blue, Andrei Tserakhau
- Published: Sep 30, 2026
- Type: Explainer

The Apache Iceberg community has adopted two additions to the Iceberg REST Catalog — read restrictions and catalog labels — each addressing distinct governance challenges. Read restrictions standardize delegated enforcement: a catalog evaluates policy for a principal and returns column-projection actions and row-filter expressions from a bounded vocabulary for a trusted engine to enforce. Catalog labels enable portable governance by allowing catalogs to exchange opaque key-value metadata at table and column level, so a consuming catalog applies native policies without per-user calls to the source. Read restrictions suit engine-to-catalog access where the source catalog trusts the engine and policies fit the standard vocabulary, whereas catalog labels fit federation between heterogeneous catalogs needing scalable governance and independent audit trails. These building blocks, along with centralized enforcement via cross-engine ABAC, advance the vision of unified governance across the open lakehouse.


### [How to scale agentic applications without creating AI sprawl](https://yomu.fyi/post/how-to-scale-agentic-applications-without-creating-ai-sprawl.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Julia Powell, Sirui Sun
- Published: Sep 30, 2026
- Type: Problem & solution

As agentic applications move from answering questions to taking actions, enterprises risk AI sprawl from duplicated integrations, inconsistent policies, and fragmented context. Databricks asserts scaling requires three pillars: choice across models and tools, context grounded in governed enterprise data and business semantics, and control through consistent permissions, policies, observability, and cost management. Its platform provides Agent Bricks for unified agent fleet building and governance, Omnigent as a common layer above agent harnesses, and Unity Gateway as a centralized control plane for models, agents, and tools. Context is shared via Unity Catalog for data and AI asset governance and Genie Ontology for common business concepts, letting teams reuse governed context across applications. Control uses Unity Gateway for access policies, budgets, and traces and Databricks Sandbox for isolated execution, with MLflow enabling evaluation and workflow inspection to maintain reliability as agents scale.


### [A practical guide to cost optimization with Lakebase Postgres](https://yomu.fyi/post/a-practical-guide-to-cost-optimization-with-lakebase-postgres.md)
- Company: [Databricks](https://yomu.fyi/company/databricks.md)
- Author: Benjamin Nwokeleme, Firas Farah, Jen Darrouzet
- Published: Sep 30, 2026
- Type: Tutorial

Lakebase uses separated storage and compute to reduce costs: branches share storage, autoscaling matches compute to demand, and scale to zero suspends compute after inactivity. Read replicas and high availability add compute without duplicating storage. For synced tables from Lakehouse, sync only the active working set via materialized views and choose the appropriate sync mode—snapshot for volatile data, triggered for incremental, continuous for real-time—and binpack tables into one pipeline to share compute. Right-size initial compute by focusing on the working set, and set PITR retention and snapshot schedules based on actual recovery needs. Snapshot storage is priced at $0.090/GB-month and PITR storage at $0.200/GB-month, both cheaper than regular branch storage. Compute and storage usage can be tracked through system.billing.usage, with separate SKUs for compute, branch storage, PITR, and snapshots.


### [Introducing SCIM Provisioning in ClickHouse Cloud](https://yomu.fyi/post/introducing-scim-provisioning-in-clickhouse-cloud.md)
- Company: [Clickhouse](https://yomu.fyi/company/clickhouse.md)
- Published: Sep 30, 2026
- Type: Announcement

ClickHouse Cloud introduces SCIM 2.0 provisioning, making the identity provider the single source of truth for user membership and role assignments. Previously, SAML SSO handled authentication but left user invites and role updates as manual tasks, creating security risks. SCIM provisioning automatically syncs user create, update, and delete operations from the IdP using standard SCIM 2.0 endpoints, maps IdP groups to custom ClickHouse roles via stable external IDs, supports up to two API keys for rotation, and provides audit logging for all provisioning actions. The feature requires an active SAML connection, is limited to verified domain SSO users, excludes system roles like Admin, and revokes all keys if SAML is disabled. It is generally available for Enterprise and BYOC organizations, validated with Okta and Microsoft Entra ID, and eliminates manual invite and deprovisioning tasks.


### [pg\_clickhouse & chdb updates: Encoding, nesting, and types](https://yomu.fyi/post/pg-clickhouse-chdb-updates-encoding-nesting-and-types.md)
- Company: [Clickhouse](https://yomu.fyi/company/clickhouse.md)
- Published: Sep 30, 2026
- Type: Announcement

pg\_clickhouse v0.11.0 and the chdb extension v0.1.2 continue their focus on cross-database compatibility between ClickHouse and PostgreSQL with several enhancements derived from header-only C libraries. A new check\_encoding server option provides error handlers for invalid byte sequences in text columns: fail, remove, replace, and truncate, with replace displaying the Unicode replacement character for UTF-8 databases. ClickHouse Interval types such as IntervalMillisecond are now imported as PostgreSQL interval columns, and expressions like start\_at + duration are pushed down to ClickHouse. Unflattened Nested columns are mapped to two-dimensional text arrays by default, or to arrays of composite types when a structure is specified; parameterized JSON columns are mapped to jsonb with proper ORDER BY pushdown. Additional type mappings include Int128 and UInt256 to numeric, BFloat16 to float4, and spatial types; the deprecated clickhouse\_raw\_query function was removed, and PostgreSQL 13 support was dropped.


### [Anyone can start building verified knowledge with Stack Internal](https://yomu.fyi/post/anyone-can-start-building-verified-knowledge-with-stack-internal.md)
- Company: [Stack Overflow](https://yomu.fyi/company/stack-overflow.md)
- Author: Caroline Thomas
- Published: Sep 30, 2026
- Type: Announcement

Stack Internal, now generally available, enables anyone to create a free Starter workspace, connect their top three data sources, and collaborate with a couple of teammates to capture and store verified knowledge. The platform evaluates knowledge using five signals—provenance, recency, expertise, corroboration, and human validation—and automatically flags conflicting information at response time to convert team know-how into validated institutional memory. It automatically detects subject matter experts based on topics, and when a critical knowledge gap arises, it notifies the right expert directly in Slack for validation, allowing confirmation or updates within that conversation. An MCP server pushes trusted, verified knowledge into AI tools like Claude, Codex, and Cursor, grounding responses in the organization’s own know-how rather than model guesses. Governance features include custom roles and permissions, full activity tracking with self-serve exportable audit trails, and a knowledge health dashboard displaying metrics like Helpful Answer Rate and Daily Active Users.


### [Getting ready for 2026 results: A look back on Developer Survey findings](https://yomu.fyi/post/getting-ready-for-2026-results-a-look-back-on-developer-survey-finding.md)
- Company: [Stack Overflow](https://yomu.fyi/company/stack-overflow.md)
- Author: Erin Yepis
- Published: Sep 30, 2026
- Type: Explainer

Drawing on 15 years of data, the post compares 2024 and 2025 Developer Survey findings across AI evolution, work settings, and community demographics ahead of the 2026 results. AI tool usage grew from 62% in 2024 to 79% in 2025, while agent adoption jumped from 31% in 2025 to 59% in an April 2026 pulse survey, with ChatGPT and GitHub Copilot as primary gateways. Among learners, AI tool use rose from 63% to 73.3%, but favorable sentiment dropped from 72.2% to 52.8% and confidence in output accuracy declined, signaling growing skepticism. Work settings evolved, with fully remote falling from 38% to 32.4% and hybrid models gaining ground, yet 82% of respondents retained some remote flexibility in 2025. The post raises open questions for the 2026 survey, including whether the gap between AI use and enthusiasm persists and how agent delegation affects developer workflows.


### [Memory safety for Postgres extensions in C/C++](https://yomu.fyi/post/memory-safety-for-postgres-extensions-in-c-c.md)
- Company: [Clickhouse](https://yomu.fyi/company/clickhouse.md)
- Published: Sep 30, 2026
- Type: Problem & solution

Postgres MemoryContext allocation combined with setjmp/longjmp error handling bypasses C++ destructors, leading to undefined behavior when C++ code runs inside a Postgres extension. C++ exceptions also bypass PG\_CATCH, and an uncaught exception in a background worker aborts the process, forcing the postmaster to restart everything due to potential shared memory corruption. ClickHouse’s four Postgres extensions each used a different approach: pg\_clickhouse replaced clickhouse-cpp with a new C library, pg\_re2 scoped C++ to a single file, pg\_chdb isolated libchdb in a helper executable, and pg\_stat\_ch uses a terminate handler. pg\_chdb’s helper process runs outside the Postgres backend, so a libchdb crash fails only the calling operation without triggering cluster crash recovery. A long-term plan for pg\_stat\_ch moves its C++ dependencies into a helper executable, with PR #129 as the first step to isolate C++ from Postgres code.


### [How Deterministic Controls Turn AI Output Into Reliable Prompt Templates](https://yomu.fyi/post/how-deterministic-controls-turn-ai-output-into-reliable-prompt-templat.md)
- Company: [Salesforce](https://yomu.fyi/company/salesforce.md)
- Author: Scott Nyberg
- Published: Sep 30, 2026
- Type: Explainer

Salesforce’s Assistive Authoring aimed to reduce prompt-template creation from 25 minutes to under a minute by using Agent Graph. A router node dispatches natural-language requests to specialized sub-agents for creation, refinement, or clarification while state variables carry context. Deterministic controls prevent the LLM from corrupting routing, record identity, or structured output through guarded transitions and a server-side transpose layer. A discovery service normalizes available data fields into an allowlist, and the LLM can only select from validated references. A human-in-the-loop gate ensures the agent never saves or executes, presenting output as an accept-or-reject diff only.


### [Cloudflare Impact reaches $100 million in donations](https://yomu.fyi/post/cloudflare-impact-reaches-100-million-in-donations.md)
- Company: [Cloudflare](https://yomu.fyi/company/cloudflare.md)
- Author: Patrick Day
- Published: Sep 30, 2026
- Type: Announcement

Cloudflare’s Impact programs have reached $100 million in donated services, protecting journalists, governments, election bodies, and public schools. The programs began as free services for developers and small businesses but evolved after discovering that some free customers doing essential work faced large attacks. Project Galileo, launched in 2014, now includes over 3,500 domains in 120 countries and blocked over 38.5 billion cyberattacks in 2025; a 2026 report found participants faced vulnerability exploit attempts at seven times the average rate. Other programs include the Athenian Project for election websites in 33 U.S. states and eight countries, Cloudflare for Campaigns with over 530 websites, and support for critical infrastructure like public schools and Ukraine. Cloudflare is expanding beyond cybersecurity to help public interest groups adapt to AI, including a hackathon for journalists that produced winning tool AIdas for studying AI bias, and a non-profit startup program.


### [Beyond pass or fail: The new era of chaos engineering](https://yomu.fyi/post/beyond-pass-or-fail-the-new-era-of-chaos-engineering.md)
- Company: [Red Hat](https://yomu.fyi/company/red-hat.md)
- Author: Paige Patton
- Published: Sep 30, 2026
- Type: Problem & solution

Binary pass/fail uptime testing no longer suffices for cloud-native deployments because a cluster can remain operational while dropping packets, breaching SLOs, or suffering performance degradation. Krkn addresses this with a trifecta of telemetry: a weighted Resiliency Score, continuous HTTP Health Checks, and KubeVirt Checks. The Resiliency Score is a 0–100% macro-health metric that evaluates live Prometheus data against pre-defined SLOs using custom weights, allowing engineers to prioritize business-critical components like authentication or payment gateways. HTTP Health Checks probe application endpoints at configurable intervals, capturing exact downtime start and end times alongside HTTP status codes, enabling teams to track self-healing duration. KubeVirt Checks monitor virtual machine instances via real-time SSH connectivity and record node, IP address changes, and exact down-time, with a final post-chaos check that flags VMs remaining unreachable after the experiment concludes.


### [Backdoors in LLMs: Why model scanning isn't enough](https://yomu.fyi/post/backdoors-in-llms-why-model-scanning-isn-t-enough.md)
- Company: [Red Hat](https://yomu.fyi/company/red-hat.md)
- Author: Uri Shaket
- Published: Sep 30, 2026
- Type: Explainer

Pulling a model from Hugging Face and running benchmarks does not guarantee the model will behave the same in production. Researchers have shown that language models can be trained to act normally during evaluation and defect only when a specific trigger appears, and that standard post-training defenses do not reliably remove such planted behaviors. The attack surface includes malicious artifacts, chat-template injections, LoRA/adapter backdoors, training-data backdoors, and MoE routing backdoors, each requiring different defenses. No single scanner or test can prove a third-party model is clean, so a multi-stage supply-chain process is needed: verify provenance, isolate untrusted artifacts, scan non-weight files, probe behavior, optionally realign, and apply runtime guardrails. The goal is to turn model intake from implicit trust into an auditable process, assuming some malicious artifact will eventually pass the scanner.


### [Small Talk with Emiliano Bonilla, CEO and Co-Founder of Synnax](https://yomu.fyi/post/small-talk-with-emiliano-bonilla-ceo-and-co-founder-of-synnax.md)
- Company: [Jetbrains](https://yomu.fyi/company/jetbrains.md)
- Author: Mitsal Athaya
- Published: Sep 30, 2026
- Type: Announcement

Emiliano Bonilla, CEO of Synnax, built the company after witnessing rocket engine testing operations ruin multi-million-dollar hardware because Excel and Python could not handle real-time control. Synnax is a platform that replaces decades-old aerospace tools, capturing sensor data in real time and tightening the loop between collection and action. The team built its own real-time programming language, Arc, in about six months after finding existing tools unsuitable. Synnax currently focuses on aerospace but is expanding into cryogenic control for quantum computing, oil and gas, and manufacturing. The company uses JetBrains IDEs, citing their debuggers and merge conflict resolution as critical to balancing speed and quality in mission-critical software.


### [Cut your AI spend with AI Gateway's Auto Router](https://yomu.fyi/post/cut-your-ai-spend-with-ai-gateway-s-auto-router.md)
- Company: [Cloudflare](https://yomu.fyi/company/cloudflare.md)
- Author: Harrison Harnisch
- Published: Sep 30, 2026
- Type: Announcement

Cloudflare released Auto Router in public beta through AI Gateway. Set the model to “cloudflare/auto” and the gateway automatically routes each request to a model capable enough for the task, filtering by credentials, access controls, and provider health. A multi-head classifier on Workers AI assigns probabilities across 14 task categories and rates complexity, ambiguity, stakes, and context dependence on a 1-5 scale. A separate scoring matrix combines those signals with model benchmark results and token prices, applying an adaptive cost penalty and switching costs for cached contexts in long sessions. On an internal general knowledge-work benchmark of 97 tasks with three samples each, Auto Router achieved 86.6% success at $0.0084 per success, costing 35% of Claude Opus 5.5 and 80% of GPT-6 Sol while remaining free during beta.


[Newer posts](https://yomu.fyi/posts/page/27.md) · [Older posts](https://yomu.fyi/posts/page/29.md)