https://thehallwaytrack.news/Searchable intelligence from the world's AI conferences — talks, quotes, and who said what.en-usWed, 07 Oct 2026 21:00:23 +0000https://thehallwaytrack.news/talks/yt-ltqhimoeueyhttps://thehallwaytrack.news/talks/yt-ltqhimoeueyTue, 06 Oct 2026 23:00:23 +0000Google DeepMind releases new audio models including TTS, live translation, and Gemini Live upgrades “Our new text-to-speech models are capable of doing really compelling voice personalization, which means that you can create a voice and guide it to have specific emotions, specific resonance, and even to incorporate things like pauses.” Google DeepMind announced three new audio capabilities at Google I/O: expressive text-to-speech with emotion and resonance control, live translation across 100 languages from video or audio input, and expanded Gemini Live models with function/tool call support for developers. These updates extend Ge…https://thehallwaytrack.news/talks/yt-ya0-o69jkp8https://thehallwaytrack.news/talks/yt-ya0-o69jkp8Tue, 06 Oct 2026 23:00:16 +0000Enterprises must redesign operating models around agents, not just add agents to existing workflows “Where do I best insert agents working side by side with humans to ultimately optimize humans?” A Salesforce leader describes the shift from ad-hoc agent building to a deliberate operating model redesign, where technologists are embedded directly with business leaders rather than siloed in IT. The Co-op organization model places data and AI roles alongside business decision-makers daily. Becom…https://thehallwaytrack.news/talks/yt-xtpynregjq4https://thehallwaytrack.news/talks/yt-xtpynregjq4Tue, 06 Oct 2026 22:30:02 +0000Speculative decoding cuts LLM inference cost but only when GPU memory has spare capacity An Akamai engineer explains speculative decoding — using a smaller draft model to speculatively generate tokens that a larger target model then validates in one pass — as a way to reduce inference latency and cost. The key practical finding is that it only pays off when GPU memory is underutilized; …https://thehallwaytrack.news/talks/rss-61299ed5https://thehallwaytrack.news/talks/rss-61299ed5Tue, 06 Oct 2026 21:43:00 +0000TypeSafe's Jev model uses RLCD to deliver reliable machine-native intelligence for software automation TypeSafe CEO Diogo Almeida argues that text-optimized LLMs are poorly suited for real-world automation decisions, and introduces Jev, a model built around calibration and reliability via reinforcement learning from calibrated decisions (RLCD). The core thesis is that AI systems should become more en…https://thehallwaytrack.news/talks/rss-a31b24b0https://thehallwaytrack.news/talks/rss-a31b24b0Tue, 06 Oct 2026 21:32:34 +0000llm-mistral 0.16 adds reasoning model support including Mistral Large 4 Simon Willison released llm-mistral 0.16, adding support for reasoning models via his LLM CLI tool, coinciding with the release of Mistral Large 4. This extends the LLM ecosystem's support for the latest reasoning-capable models from Mistral. While useful for developers using the LLM tool, this is a…https://thehallwaytrack.news/talks/yt-odxmhoduykyhttps://thehallwaytrack.news/talks/yt-odxmhoduykyTue, 06 Oct 2026 21:00:12 +0000Adecco used Salesforce AgentForce to save recruiters 35-40% time and place 20,000 more people “AI had to happen with people and not to people.” Adecco CEO Danny shared at Dreamforce 2026 how deploying Salesforce AgentForce across core recruiting workflows freed recruiters from administrative tasks, resulting in 35-40% time savings and 20,000 additional job placements year-to-date. Adecco also co-developed 'Our Potential,' an AI-native platf…https://thehallwaytrack.news/talks/rss-bb6b5493https://thehallwaytrack.news/talks/rss-bb6b5493Tue, 06 Oct 2026 20:37:53 +0000Google releases EmbeddingGemma 2 under Apache 2.0, enabling portable open-weight embeddings “I don't think it makes sense to use a closed, proprietary, hosted-only model.” Google released EmbeddingGemma 2 under the Apache 2.0 license, which Simon Willison highlights as particularly important for embedding models. The open license solves a critical vendor lock-in problem: proprietary embedding models force costly re-computation of stored vectors when vendors deprecate …https://thehallwaytrack.news/talks/rss-3c606a59https://thehallwaytrack.news/talks/rss-3c606a59Tue, 06 Oct 2026 20:18:19 +0000Mistral releases 1 trillion parameter Large 4 model, reclaiming competitive position “it's great to see Mistral put out a model that's back to being maybe about 6 months behind the frontier” Mistral has released a preview of Mistral Large 4, a 1 trillion parameter MoE model with 49B active parameters trained on 3,800 NVIDIA Grace Blackwell GPUs. The model scores 38 on Artificial Analysis, a massive jump from Mistral Large 3's score of 9, placing it just behind DeepSeek 4.1 Flash. Open w…https://thehallwaytrack.news/talks/yt-ukua1a0kdfchttps://thehallwaytrack.news/talks/yt-ukua1a0kdfcTue, 06 Oct 2026 20:00:07 +0000AI agent demand is causing 10-100x user growth that database infrastructure cannot keep pace with “the number of daily active users has doubled, tripled, 10 or 100 times. This is largely driven by the demand for AI agents. The infrastructure simply cannot keep up with this development.” A PlanetScale engineer at AI Engineer conference argues that AI agent adoption is driving unprecedented database scaling challenges, with some platforms seeing 10-100x daily active user growth that existing infrastructure cannot handle. The talk reframes Facebook's old 'move fast and break things' e…https://thehallwaytrack.news/talks/rss-bf211d23https://thehallwaytrack.news/talks/rss-bf211d23Tue, 06 Oct 2026 19:57:04 +0000Google DeepMind releases EmbeddingGemma 2, an open lightweight multimodal embedding model Google DeepMind has released EmbeddingGemma 2, an open and lightweight multimodal embedding model. The release signals Google's continued investment in open model weights for the embedding/retrieval space, competing with models like OpenAI's text-embedding series and open alternatives. No content bo…https://thehallwaytrack.news/talks/rss-48daff0ahttps://thehallwaytrack.news/talks/rss-48daff0aTue, 06 Oct 2026 19:19:15 +0000AWS AgentCore and OpenClaw enable persistent-memory AI assistants via single CloudFormation deployment “AgentCore memory, a capability of Amazon Bedrock AgentCore, turns disposable chats into durable knowledge.” AWS published a reference architecture for building context-aware AI assistants using AgentCore runtime and the open-source OpenClaw framework, addressing the persistent-memory gap in stateless LLM assistants. The system accumulates structured, tagged memories across sessions and deploys entirely vi…https://thehallwaytrack.news/talks/rss-149df84bhttps://thehallwaytrack.news/talks/rss-149df84bTue, 06 Oct 2026 19:07:31 +0000Simon Willison integrates Parseable observability platform with Datasette OpenTelemetry traces using Codex. Simon Willison documents a TIL on connecting Parseable, a new Rust-based observability platform, to Datasette's newly added OpenTelemetry support. He used Codex (AI coding assistant) to help configure the integration. The post is primarily a developer how-to with minimal AI industry signal beyond ca…https://thehallwaytrack.news/talks/rss-240166d7https://thehallwaytrack.news/talks/rss-240166d7Tue, 06 Oct 2026 19:07:02 +0000NVIDIA DOCA GPUNetIO enables GPUs to control networking directly, eliminating CPU bottlenecks “GPU applications increasingly need networking and data movement to behave like first-class GPU-controlled operations rather than host-driven services.” NVIDIA's DOCA GPUNetIO framework unifies GPU-initiated networking across its software stack, allowing GPUs to own network transactions without CPU intermediation. This reduces latency on the critical path for distributed GPU applications. While technically significant for AI infrastructure engineers…https://thehallwaytrack.news/talks/yt-srxhjodsztchttps://thehallwaytrack.news/talks/yt-srxhjodsztcTue, 06 Oct 2026 19:00:07 +0000GitHub Copilot agent converts meeting transcripts into project backlogs and risk assessments automatically “it gives you a competitive edge” A Microsoft MVP demonstrates a 'delivery factory agent' built on GitHub Copilot that transforms raw meeting transcripts and technical proposals into structured task lists, project scopes, risk registers, and upsell opportunities. The workflow targets delivery leads who currently lack systematic trac…https://thehallwaytrack.news/talks/yt-c9xh7bmkbqehttps://thehallwaytrack.news/talks/yt-c9xh7bmkbqeTue, 06 Oct 2026 18:48:43 +0000Microsoft IQ gives enterprise agents grounded context via four intelligence sources for meaningful outcomes “Models give agents intelligence but contexts give that intelligence meaning” Microsoft introduced 'Microsoft IQ,' a framework that connects enterprise agents to four distinct context sources—including Web IQ and Foundry IQ—enabling agents to go beyond generic responses and deliver grounded, business-specific outcomes. The demo showcased a supply chain autopilot built on Azur…https://thehallwaytrack.news/talks/yt-muqovz7thu0https://thehallwaytrack.news/talks/yt-muqovz7thu0Tue, 06 Oct 2026 18:30:35 +0000Agent quality demands both pre-production evals and production observability to manage LLM non-determinism risks “grading matters because LLMs are inherently non-deterministic” Braintrust's Hussein frames agent quality around two pillars: offline evaluations before deployment and continuous observability in production, treating them as two sides of the same reliability problem. The core argument is that LLMs' non-determinism—their source of power—creates brand, compliance,…https://thehallwaytrack.news/talks/rss-d9a88140https://thehallwaytrack.news/talks/rss-d9a88140Tue, 06 Oct 2026 18:20:58 +0000Mistral Large 4 released, benchmarks criticized as saturated by frontier models “The benchmark is saturated. Frontier models are tested with an armadillo in fishnet tights jaywalking on Mars.” Mistral released Mistral Large 4, prompting community commentary on benchmark saturation in frontier model evaluation. Simon Willison used the release as an opportunity to run a whimsical SVG generation test across Claude, GPT, Gemini, and Mistral models. The post is more notable for the benchmark c…https://thehallwaytrack.news/talks/rss-e964055chttps://thehallwaytrack.news/talks/rss-e964055cTue, 06 Oct 2026 17:35:39 +0000datasette-atom 0.11a0 released for compatibility with latest Datasette alphas A minor compatibility fix for the datasette-atom plugin to work with the latest Datasette alpha releases. This enabled an upgrade of datasette.io to Datasette 1.0a41. No meaningful AI industry signal.https://thehallwaytrack.news/talks/yt-teerpybuiemhttps://thehallwaytrack.news/talks/yt-teerpybuiemTue, 06 Oct 2026 17:00:19 +0000An OSS reference stack for self-improving agents combines online monitoring with offline evaluation cycles. “I mean the transition from "I don't write prompts" to "I have cycles."” LangFuse co-founder Marc Klingen describes an emerging OSS reference stack where teams close the loop between production agent monitoring and offline dataset evaluation. He argues that combining online observability with offline experimentation is now table stakes for serious agent development, repl…https://thehallwaytrack.news/talks/rss-643b4690https://thehallwaytrack.news/talks/rss-643b4690Tue, 06 Oct 2026 16:35:12 +0000Databricks claims lakehouse architecture outperforms traditional data warehouses in 2026 benchmarks Databricks published 2026 benchmark data arguing the lakehouse architecture has surpassed the traditional data warehouse after two decades of warehouse dominance. This is primarily a vendor marketing piece rather than a neutral research finding. Relevant as infrastructure context but not a critical …https://thehallwaytrack.news/talks/yt-5ktfyeogarkhttps://thehallwaytrack.news/talks/yt-5ktfyeogarkTue, 06 Oct 2026 16:31:25 +0000LangChain achieved 64% cost reduction in their coding agent via model routing with no quality loss “we were able to see a 64% reduction in median cost per thread with no measurable change in quality” LangChain PM Sydney demonstrates a 4-step model routing framework that selects cheap vs. expensive LLMs per task, tested on their Open SWE coding agent. The experiment yielded a 64% median cost reduction with no measurable quality degradation, validated via A/B test on live traffic. The pattern — cl…https://thehallwaytrack.news/talks/rss-929785fbhttps://thehallwaytrack.news/talks/rss-929785fbTue, 06 Oct 2026 16:13:47 +0000NVIDIA releases AICR v1.0 for open, verifiable GPU cluster configuration on Kubernetes NVIDIA published AICR v1.0, a specification for stable and verifiable GPU-accelerated Kubernetes cluster configurations. It addresses the complex dependency problem where dozens of components across kernels, drivers, runtimes, and operators must be compatible. This matters for AI infrastructure team…https://thehallwaytrack.news/talks/yt-anpss6huqk0https://thehallwaytrack.news/talks/yt-anpss6huqk0Tue, 06 Oct 2026 16:10:20 +0000Google releases EmbeddingGemma 2, a sub-billion open model unifying text, image, video, and audio embeddings on-device. “a picture of a cat, the word cat, and the sound of a cat are all mapped close to each other in a shared high-dimensional embedding space” Google launched EmbeddingGemma 2, an open model with up to 740M parameters that maps text, images, video, and audio into a single unified embedding space, setting a new benchmark for sub-billion parameter models. It runs entirely on-device with no API calls required, enabling private multimodal sear…https://thehallwaytrack.news/talks/rss-60bdcf15https://thehallwaytrack.news/talks/rss-60bdcf15Tue, 06 Oct 2026 16:00:06 +0000Meta open-sources authenticated NTP via NTS at nts.meta.com, making time verifiable “That work made time precise. This work makes it verifiable.” Meta has deployed Network Time Security (NTS, RFC 8915) on its public time service, adding cryptographic authentication to NTP so clients can verify time packets were not tampered with in transit. The stateless implementation is fully open-sourced via Meta's Time library on GitHub. This is a foundat…https://thehallwaytrack.news/talks/rss-a444153ahttps://thehallwaytrack.news/talks/rss-a444153aTue, 06 Oct 2026 16:00:00 +0000Atlassian and OpenAI expand partnership to embed frontier models into enterprise workflows Atlassian and OpenAI are deepening their partnership to connect frontier AI models with enterprise knowledge stored in Atlassian's suite of tools. The integration aims to help teams move from information retrieval to action, spanning planning, building, and delivery workflows. This signals OpenAI's …https://thehallwaytrack.news/talks/rss-3bab1efbhttps://thehallwaytrack.news/talks/rss-3bab1efbTue, 06 Oct 2026 16:00:00 +0000IFCO shares operational lessons scaling a large dbt project on Databricks for supply chain data. IFCO, a reusable packaging company, describes how they manage a large-scale dbt project on Databricks. The content focuses on performance, visibility, and debugging in data pipelines. This is a niche data engineering case study with minimal relevance to AI conference intelligence.https://thehallwaytrack.news/talks/rss-6becfe32https://thehallwaytrack.news/talks/rss-6becfe32Tue, 06 Oct 2026 15:53:28 +0000AWS publishes guidance aligning customers with ISO/IEC 42005:2025 AI impact assessment standards. “translat[ing] AI capability into practical deployment across firms, sectors, and public services.” AWS published a blog post walking enterprise customers through ISO/IEC 42005:2025, the international standard for AI system impact assessments, citing $581.69B in global AI investment as context for urgency. The post highlights AWS tooling—including an ISO/IEC 42001:2023 AIMS implementation guide an…https://thehallwaytrack.news/talks/rss-c64f29f6https://thehallwaytrack.news/talks/rss-c64f29f6Tue, 06 Oct 2026 15:50:23 +0000AWS outlines four-layer governance model for shared SageMaker HyperPod ML compute clusters AWS published a best-practices guide for administering SageMaker HyperPod clusters shared across multiple ML teams, covering organization, project, cluster, and workload governance layers. The post addresses the non-technical challenge of multi-team compute governance: capacity allocation, access co…https://thehallwaytrack.news/talks/rss-70b427dfhttps://thehallwaytrack.news/talks/rss-70b427dfTue, 06 Oct 2026 15:47:02 +0000AWS adds Studio UI to manage HyperPod Spaces, eliminating CLI dependency for ML developers. “reducing the time from cluster access to productive development to a few clicks” AWS now lets data scientists create and manage SageMaker HyperPod Spaces directly from the Studio browser UI, removing the need for CLI or kubectl commands. The update targets the gap between infrastructure admins (who prefer CLI) and data scientists who want a visual workflow. It reflects a broader…https://thehallwaytrack.news/talks/rss-e00f5f8dhttps://thehallwaytrack.news/talks/rss-e00f5f8dTue, 06 Oct 2026 15:45:42 +0000AWS launches Amazon Nova 2.5 Sonic for real-time voice AI agents via Bedrock AgentCore AWS published a tutorial demonstrating a voice travel concierge built on Amazon Bedrock AgentCore, the newly referenced Amazon Nova 2.5 Sonic speech-to-speech model, and Bedrock Knowledge Bases. The architecture uses Model Context Protocol (MCP) to loosely couple the AI agent to backend systems, wit…https://thehallwaytrack.news/talks/yt-35vj67zoyeyhttps://thehallwaytrack.news/talks/yt-35vj67zoyeyTue, 06 Oct 2026 15:43:27 +0000No meaningful content extracted from AWS Summit Milan 2026 captions The content consists only of garbled caption artifacts with no substantive information. No announcements, quotes, or AI-relevant signal could be extracted. This item should be excluded from the newsletter.https://thehallwaytrack.news/talks/yt-x6l4lpa0-nyhttps://thehallwaytrack.news/talks/yt-x6l4lpa0-nyTue, 06 Oct 2026 15:30:27 +0000Tessl proposes 'harness engineering' as a new discipline for building agent-powered software factories “a software factory is any agent-based system where all of the end product you build and deliver to users is created by agents” Drew Knox from Tessl introduced 'harness engineering' as a new core discipline for teams scaling beyond individual coding agents toward fully autonomous software factories. A software factory is defined as an agent-based system where all deliverables are produced by agents while dev teams shift focu…https://thehallwaytrack.news/talks/rss-56c07856https://thehallwaytrack.news/talks/rss-56c07856Tue, 06 Oct 2026 15:17:39 +0000Claude Opus 5.5 composed surprisingly competent retro game music via a custom text format “I wonder if the ability to compose competent music is similar to the 3D graphics thing - a new capability for text models that emerged in the past few months?” Simon Willison prompted Claude Opus 5.5 to design a text-based music format and build a browser-playable artifact, resulting in a polished retro game music player with six original tracks. The quality was unexpectedly high, leading Willison to speculate that music composition may be an emergent capa…https://thehallwaytrack.news/talks/rss-e02cad9fhttps://thehallwaytrack.news/talks/rss-e02cad9fTue, 06 Oct 2026 15:00:23 +0000Apple launches iPhone Duo, a foldable smartphone with multi-posture form factor for developers Apple's October 2026 developer newsletter announces iPhone Duo, a new foldable iPhone supporting tent, book, and clamshell modes, with developer resources including app submission tips and Group Labs. The content is a developer housekeeping newsletter with no AI-specific signal. While a new Apple fo…https://thehallwaytrack.news/talks/rss-d48d573ehttps://thehallwaytrack.news/talks/rss-d48d573eTue, 06 Oct 2026 15:00:00 +0000NVIDIA introduces Green Contexts for fine-grained GPU resource sharing between concurrent workloads “Controlling how GPU resources are shared between them remains difficult.” NVIDIA's Developer Blog introduces Green Contexts, a mechanism to control GPU resource sharing between multiple concurrent components in a single process. This addresses a real pain point in AI inference deployments where latency-sensitive operators and throughput-oriented background kernels compete…https://thehallwaytrack.news/talks/rss-3142dbe4https://thehallwaytrack.news/talks/rss-3142dbe4Tue, 06 Oct 2026 15:00:00 +0000Salesforce Agentforce splits agent reasoning into probabilistic LLM and deterministic business logic layers “Agentforce splits the agent brain into two layers. The first is a generative layer, the LLM, that does all of the language and reasoning. The second is a deterministic layer.” Salesforce's Principal Architect of Ethical AI Practice explains how Agentforce routes all consequential agent decisions through a deterministic layer (Agent Fabric, Flow, Apex) rather than letting LLMs decide independently. This dual-layer architecture — probabilistic reasoning up front, determinis…https://thehallwaytrack.news/talks/rss-514fdf92https://thehallwaytrack.news/talks/rss-514fdf92Tue, 06 Oct 2026 14:22:41 +0000Banning open-weight models will hurt US AI competitiveness while increasing long-term cyber risk “I feel we're going to end up making policy decisions that both limit American AI competitiveness and increases long-term cyber risk.” Nathan Lambert argues the Western discourse on open-weight AI model cyber risks is dangerously one-sided, locked on a lose-lose policy path that restricts American competitiveness without actually stopping bad actors. He specifically calls out Anthropic's GLM-5.3 cyber risk report as 'solipsistic' f…https://thehallwaytrack.news/talks/yt-v0d-7doxunuhttps://thehallwaytrack.news/talks/yt-v0d-7doxunuTue, 06 Oct 2026 13:36:55 +0000NVIDIA GTC Berlin keynote announcement with no substantive content available The source is a caption transcript containing only the word 'Heat' repeated twice, with no actual content from the NVIDIA GTC Berlin keynote. There is no extractable AI signal or information from this submission. The event itself may be significant, but this content provides nothing to report.https://thehallwaytrack.news/talks/rss-549d1d1dhttps://thehallwaytrack.news/talks/rss-549d1d1dTue, 06 Oct 2026 12:45:00 +0000Meta's ads MCP server integrates with Databricks to connect enterprise customer data to ad campaigns Meta is bringing its advertising MCP server to the Databricks platform, enabling enterprise advertisers to apply their customer intelligence data directly to ad campaign targeting. This represents a practical MCP adoption use case outside of developer tooling, extending the protocol into the martech…https://thehallwaytrack.news/talks/rss-8ab2e904https://thehallwaytrack.news/talks/rss-8ab2e904Tue, 06 Oct 2026 12:00:00 +0000OpenAI frontier model solves open math problems, releases formal Lean proofs on GitHub OpenAI released results showing an internal frontier model making progress on genuine open problems in mathematics, with formal Lean proof formalizations published on GitHub for public verification. This moves beyond benchmark performance to actual contributions to unsolved mathematical problems, re…https://thehallwaytrack.news/talks/yt-u3-vqzao2vmhttps://thehallwaytrack.news/talks/yt-u3-vqzao2vmTue, 06 Oct 2026 10:22:16 +0000YDUQS used Databricks to achieve 90% productivity gains across 1.3M students “we can bring context to our IAIS, to our agents, who are IAIS who understand our business, who understand our students, who understand our employees and our teachers” Brazilian education group YDUQS (1.3M students) migrated to Databricks to resolve data silos and legacy architecture issues, achieving 90% productivity gains in data analysis. The deployment includes Unity Catalog governance, AI/BI dashboards, and AI agents contextualized on student and employee dat…https://thehallwaytrack.news/talks/rss-93f34817https://thehallwaytrack.news/talks/rss-93f34817Tue, 06 Oct 2026 10:00:00 +0000OpenAI and Ironclad are training AI agents on complex legal contracting workflows OpenAI has partnered with Ironclad, a contract management platform, to train and evaluate AI agents on complex contracting workflows. This represents a concrete enterprise deployment of computer use capabilities in professional legal work. The collaboration signals OpenAI's push to validate agentic …https://thehallwaytrack.news/talks/yt-dc4-wolz-tahttps://thehallwaytrack.news/talks/yt-dc4-wolz-taTue, 06 Oct 2026 07:00:33 +0000Azure launches flexible spend-based savings plans for database cost optimization “It's a flexible spend-based commitment that gives you meaningful discounts across eligible Azure database services without locking you into a specific service, region, or configuration.” Microsoft announced Azure Savings Plans for databases, a spend-based commitment model offering discounts across eligible Azure database services for 1 or 3-year terms. The offer is flexible across service tiers, regions, and configurations, automatically applying the greatest available discount each…https://thehallwaytrack.news/talks/rss-807faee2https://thehallwaytrack.news/talks/rss-807faee2Tue, 06 Oct 2026 06:44:39 +0000Falcon-Emirati LLM is trained to understand Emirati dialect, culture, and nuance. Hugging Face highlights Falcon-Emirati, a language model adaptation targeting the Emirati Arabic dialect with cultural and linguistic nuance. This signals growing momentum in region-specific LLM development beyond generic multilingual models. It underscores a broader industry trend of localizing fou…https://thehallwaytrack.news/talks/rss-cf46e538https://thehallwaytrack.news/talks/rss-cf46e538Tue, 06 Oct 2026 06:28:43 +0000Reflection AI launches Beam, a 501B/23B-active US-trained open MoE model for coding and agentic work Reflection AI has ended its stealth period by launching Beam, a 501B-total/23B-active Mixture-of-Experts model trained from scratch in the US, targeting coding, agentic, and scientific workloads. While benchmarks place it behind frontier open models like GLM 5.3, Kimi K3, and DeepSeek V4.1 Flash, it…https://thehallwaytrack.news/talks/yt-fwis44-xyywhttps://thehallwaytrack.news/talks/yt-fwis44-xyywTue, 06 Oct 2026 02:00:02 +0000OpenAI Codex now runs on mobile and can be self-hosted via open-source application server “You can host Codex anywhere thanks to its open source code and application server” OpenAI engineers Charlie Guo and Gabriel Chua presented a 'Cooking with Codex' masterclass at AI Engineer, showcasing community-built projects including a synthesizer plugin, video editing in Premiere, game development, and a self-driving golf cart. They announced Codex is now available on mobile de…https://thehallwaytrack.news/talks/yt-c5u-xbben-ghttps://thehallwaytrack.news/talks/yt-c5u-xbben-gTue, 06 Oct 2026 01:30:12 +0000PayPal is integrating with OpenAI and Google agent commerce protocols to enable AI-initiated payments “PayPal Enterprise Payments is building bridges for merchants so they can accept payments from many different places.” PayPal Corporate Payments (formerly Braintree) is building integrations with two emerging agent commerce protocols: OpenAI's Agent Commerce Protocol (for ChatGPT) and Google's Universal Commerce Protocol, which the speaker notes are nearly identical in specification. The implementation uses MCP apps…https://thehallwaytrack.news/talks/yt-hfeczxdnyvuhttps://thehallwaytrack.news/talks/yt-hfeczxdnyvuTue, 06 Oct 2026 01:00:25 +0000Braintrust workshop teaches isolating signal from AI agent trace noise using observability tooling Braintrust solutions engineer Doug Guthrie led a hands-on AI Engineer workshop on using Braintrust's evaluation and observability platform to monitor AI agents. The core challenge he framed is extracting actionable signal from the large volumes of trace data agents generate. The session was practica…https://thehallwaytrack.news/talks/yt-zuzvnkfsmtmhttps://thehallwaytrack.news/talks/yt-zuzvnkfsmtmTue, 06 Oct 2026 00:30:15 +0000AI agents should operate with zero credentials, secured via sandboxes and MCP gateways “Agents do a lot more than I thought...than I expected them to do.” Docker engineer Jim Clark argues that as AI agents take on longer autonomous tasks with less human supervision, credential management becomes a critical security concern — his answer is zero standing credentials for agents. The talk covers agent harnesses, sandboxes, and MCP gateways as the architec…https://thehallwaytrack.news/talks/rss-6950ea13https://thehallwaytrack.news/talks/rss-6950ea13Tue, 06 Oct 2026 00:05:33 +0000Apple iPhone Duo launches October 23, 2026, requiring developer app updates. Apple announced iPhone Duo availability on October 23, 2026, with developer guidance for app optimization using Xcode 27.1. Starting April 2027, App Store submissions will require iPhone Duo screenshots. This is a mobile hardware/developer story with minimal AI signal.