iCreatAI News
Back to main site
Skip to content
#Tutorials
Nov 13Fri
  1. Hacker News AI40

    AI Security Wiki

    The AI Security Wiki is a reference for securing enterprise AI adoption, covering governance, risk management, secure engineering, monitoring, vendor evaluation and assurance, with pages reusable as templates that specify the audit evidence each activity should produce. It positions itself as the implementation layer—control objectives, evidence, audit procedures and test cases—that works alongside frameworks such as FINOS AIGF, ISO/IEC 42001, NIST CSF, PCI DSS and CERT-In rather than replacing them. The repository includes a 682-case test library, regional compliance hubs for the GCC, EU and other jurisdictions, framework crosswalks, role-based playbooks and downloadable templates.

  2. OpenRouter Announcements60

    Reranking in RAG: How to Fix Bad Retrieval with a Reranker

    This guide explains that vector search often ranks near-miss chunks first because embeddings compress whole chunks into single vectors, and shows how adding a cross-encoder reranker over the top 30 candidates raised precision@1 from 81% to 97% on a 274-question synthetic benchmark at roughly $0.10 per 1,000 queries. It compares 11 rerankers (voyageai/rerank-3-lite and rerank-3 were the most accurate dedicated models and cheapest, while LLM rerankers like openai/gpt-6-luna hit 99.6% but took about 4 seconds per call), covers candidate-count tuning, input_type settings, and silent failures such as cohere/rerank-v3.5's 4,096-token limit. It includes a Python example using the embeddings and rerank APIs and notes that real-world gains depend on your own documents and evaluation set.

    Why it matters: Offers a measured, benchmark-backed walkthrough of adding reranking to a RAG pipeline, including cost, latency and failure-mode comparisons useful for teams tuning retrieval.

  3. Hacker News AI18

    Show HN: AI Math Kit – Open Problems. Explore Honestly

    AI Math Kit is a browser extension popup that presents 217 open math problems (number theory, calculus, and cross-field mysteries such as the Riemann Hypothesis and P vs NP) alongside 242 plain-language guides to AI and formalized-math breakthroughs, plus prompt templates and an honesty pledge for exploring with AI. It offers Chinese, English, and Japanese interfaces, caches payment status locally, and unlocks full problem explanations, breakthrough details, and prompts via an optional Pro in-app purchase (ExtPay). The developer states the content is introductory paraphrase and does not claim AI has proven any conjecture.

  5. ElevenLabs Blog22

    7 customer service skills every human and AI agent needs (with examples)

    This guide outlines seven customer service skills—communication, problem-solving, conflict resolution, empathy, adaptability, personalization, and product knowledge—and explains how both human and AI agents apply each, citing a Zendesk study that more than 50% of customers switched to competitors after a single bad experience. It describes implementation steps including defining CX goals, choosing an agentic AI platform, architecting the agent workflow, and testing before deployment, and notes Gartner's projection that at least 70% of customers will start their customer journey using an AI interface by 2028. It also covers training and feedback through metrics such as CSAT, first contact resolution, and average handle time, and distinguishes hard from soft skills.

  6. vLLM Blog72

    DeepSeek-V4.1-Flash on Vera Rubin NVL72: 58x Throughput vs. GB300 NVL72 at 387 tokens/s

    On a single NVIDIA Vera Rubin NVL72 rack, vLLM serves DeepSeek-V4.1-Flash at 1.7M tokens/s per GPU with 387 tokens/s per user, 58x GB300 NVL72, reaching 2.52M at peak throughput and 1,810 tokens/s for a single user. The gains come from kernels tuned for Rubin's split memory domains and FP4 tensor cores, specialized low-latency decode kernels, and a rack-wide KV cache and Engram tables shared over NVLink. The post details four optimization groups and reports a 96.5–97% prefix hit rate at high concurrency with workload-aware routing.

    Why it matters: It offers a detailed, measurement-backed look at how Rubin-specific memory and MMA features translate into serving throughput and latency for a frontier open model.

  7. GitHub Blog · AI & ML58

    Build once, run on any agent: A practical guide to agent skills

    GitHub's guide explains agent skills as versioned directories centered on a SKILL.md file that capture repeatable task instructions shareable across agents, while the environment still supplies tools, credentials, and permissions. Using examples like Proof Agent for adversarial verification and Data Forge for a multi-skill data pipeline, it recommends keeping project configuration in a manifest, sharing execution paths between skills, and placing human checkpoints and CI checks in workflows. It also covers packaging skills for clients such as Claude Code, Cursor, and GitHub Copilot, and notes that applications suit workflows needing stable interfaces, shared state, audit trails, or permissions enforced outside the model.

  8. AWS Machine Learning Blog47

    How Natera designed an agentic AI system for lab ops with Amazon Bedrock AgentCore

    Natera's lab engineering team built SWARM (Systematic Workflow Analytics and Resolutions Manager) on Amazon Bedrock AgentCore to capture and deploy institutional knowledge for lab instrument support, using two runtimes (swarm-agent and swarm-hive) that share a versioned library of 33 skills. The system reduced average first response time from 17.4 hours to roughly 8 minutes (about 130x faster) and delivered an automated first response on 96.2 percent of tickets, measured over a 90-day production evaluation period. The post details the architecture, including a custom document pipeline with Amazon Titan Text Embeddings and Amazon Bedrock Knowledge Bases, keyless OIDC-based deployment, Jira integration via EventBridge and Amazon SQS, and a continuous evaluation loop.

  10. AWS Machine Learning Blog40

    Four Amazon Bedrock inference choices that affect cost

    The post explains how four request-level Amazon Bedrock settings—prompt cache duration, service tier, reasoning effort, and maxTokens—can change token processing and billing rates without altering the model or prompt. It reports tests showing that a 1-hour cache pays off from the second request per hour when gaps exceed 5 minutes, that Flex on Nova 2 Lite processed the same tokens at half the Standard rate in us-west-2, that lowering effort from high to low cut median output tokens by 169% on an agent test, and that a 155-token maxTokens fence avoided 1,893 output tokens in one test. It recommends confirming model support, changing one setting at a time, and verifying effects through response metadata, CloudWatch metrics, invocation logs, and billing data.

  11. Google Developers Blog54

    Catch Go bugs that fuzzing misses with Gemini and a constraint solver

    The author describes a workflow that compiles Go functions to RISC-V assembly, has Gemini 4 Argon translate the surviving runtime.panicBounds traps into Z3 constraints, and replays any satisfying input through go test to produce a real reproducer. The pipeline caught three planted bugs and three latent bugs in skill-lint, and the post explains why go test -fuzz missed two crashes because fuzz inputs usually have spare slice capacity, fixed by clipping capacity to length with a full slice expression. It also notes that an unsat result only means "maybe correct," since the LLM-written solver model is not a sound proof system.

  12. AWS Machine Learning Blog36

    Tokenomics at scale: Track and cap Amazon Bedrock spend per IAM principal with CUR 2.0 and AWS Budgets

    This AWS post describes how to get per-identity Amazon Bedrock cost visibility and enforceable spend limits using only AWS-native features: IAM principal and session tags for attribution, Cost Explorer or CUR 2.0 exports (with the caller identity option) for reporting, and AWS Budgets with budget actions that attach a Deny IAM policy when a threshold is crossed. It notes that budget data refreshes up to three times a day, cost allocation tags are not retroactive, and a Deny action on a shared role blocks everyone using it.

  13. Databricks Blog28

    Eclipse Aerospace simplifies flight-data ingestion with Databricks

    Eclipse Aerospace worked with Databricks to replace a multi-stage AWS Lambda and Avro pipeline with an Arrow-native PySpark custom data source that wraps its existing Rust flight-data decoder, reading proprietary telemetry files directly in Spark on serverless compute. The approach, enabled by the Python Data Source API available from Databricks Runtime 15.4 LTS, cut latency, removed duplicate Avro storage and a Lambda tier, and brought decoding into the Spark UI and job metrics. Eclipse reports the connector can process records within a minute of upload and commonly handle multi-gigabyte batches in parallel in under two minutes.

  14. Databricks Blog22

    How Constellation is building a governed data foundation on Databricks

    Constellation, the largest US nuclear fleet operator, is building a governed Enterprise Data Platform on Databricks to give data consistent shared meaning across domains, choosing Databricks partly because it interoperates with existing architecture rather than requiring a rip-and-replace. The platform rests on conformed semantic definitions, a semantic knowledge layer with selective Data Vault modeling, and Unity Catalog for lineage, fine-grained access and ownership of meaning, with early use cases including a nuclear engineering assistant built with Databricks Genie, outage planning, and sales intelligence.

  16. AWS Machine Learning Blog34

    Building a self-service LLM gateway with Amazon Bedrock and LiteLLM

    This post describes a pattern for a self-service LLM gateway that runs LiteLLM on AWS Fargate to expose Amazon Bedrock models through an OpenAI-compatible API, using Bedrock Application Inference Profiles for per-model quota isolation and scoped IAM access. It covers configuration of developer tools, LiteLLM model mapping with per-token costs for budget enforcement, CDK constructs for inference profiles, Langfuse observability, CloudWatch TPM alarms, and lessons learned from scaling to several hundred developers.

  17. AWS Machine Learning Blog34

    Optimizing cost per task and proving agent ROI on Amazon Bedrock

    This AWS post argues that model cost should be measured per completed task rather than per token, citing four hidden cost drivers—tokenizer efficiency, output verbosity, agentic loop efficiency, and reasoning overhead—and an illustrative example where a model with double the list price costs 65% less per task. It describes levers for reducing cost per task (prompt caching, intelligent prompt routing, output constraints, batch inference) and Bedrock features such as CountTokens, the usage block, and invocation logging for measuring token and turn counts. For ROI, it recommends documenting and measuring a human baseline before launch, instrumenting workflows as experiments with run records, joining cost to outcomes via Athena, and using a five-dimension scorecard with green/yellow/red thresholds.

  18. Wiz Research22

    ASPM in the AI Era Part 3: Fixing Risk at the Speed of Code

    The final installment of Wiz's series argues that AppSec's bottleneck is remediation, not detection, citing Verizon DBIR data that median patch time has risen to 47 days from 32. It describes how Wiz Code maps cloud findings to code owners via VCS connections and correlated data, and how the Wiz Green Agent generates tailored remediation plans that can open pull requests or hand off to coding agents like Claude Code. Wiz Workflows then routes tickets and alerts to Jira, ServiceNow, Slack, or Teams with full scanner and runtime context.

  19. Databricks Blog25

    How H6.ai builds AI-powered financial automation with Lakebase Postgres

    H6.ai automates accounts payable on top of customers' existing ERPs by combining AI-assisted invoice understanding with deterministic, auditable policy execution. To make regression testing reliable despite constantly changing financial data, it gives each customer a dedicated Lakebase Postgres database and uses copy-on-write branches to freeze snapshots, run each test on its own disposable branch, and judge results with deterministic assertions plus a panel of AI judges. The company says branching provides isolation, frozen state, cheap disposable copies and schema-safe replays, and it plans to use Lakehouse//RT queries on Lakebase tables for read-only analytics without copying data.

  20. Hacker News AI38

    Show HN: LessonFolk – Learn AI from an AI tutor, through MCP (open source)

    LessonFolk is a free, open-source site that teaches AI through short conversational lessons with an AI tutor inside the AI chat app you already use, tracking your progress. It offers 15 courses in a recommended order, from AI Foundations to How AI Models Really Work and Context Engineering for Long-Running Agents, with a short level check to skip what you already know. The courses and code are open source, and users can download or delete their data at any time.

  21. Databricks Blog58

    How Block powers collaborative AI agents with Buzz and Unity Gateway

    Block developed Buzz, an open source, model-agnostic collaboration space where employees and AI agents work together in group discussions, focused conversations, and one-on-one exchanges, with users choosing the model or coding harness behind each agent. Databricks Unity Gateway provides the model access layer, letting Buzz users select Databricks as their inference provider and choose from frontier models from OpenAI, Anthropic, and Google plus select open-weight models through a single integration. Block has also been experimenting with multi-agent workflows, such as a more capable model coordinating agents powered by smaller, more efficient models, and developed an internal benchmarking approach for these agent teams.

  22. PyTorch Blog38

    Fast by Default: Compile-Time Heuristics for Helion Kernels

    Helion now uses compile-time heuristics that analyze a kernel's IR and dataflow to classify it as matmul, reduction, or pointwise and emit block sizes, warps, and stages without autotuning. On H100, the computed configs reach 1.18x hand-written FLA Triton for linear attention, 1.27x vLLM's handwritten CUDA reduction kernels, and 1.09x torch.compile max-autotune on a general reduction suite. The heuristics are merged into Helion and enabled by default on H100 and B200 with the Triton backend.

  23. Hacker News AI25

    Learning shaders when AI can write them for you

    This is the prologue of a course on shaders, presenting a planet drawn by one vertex shader and one fragment shader that you can spin by dragging the canvas. The course breaks the planet down into data, vertices, rasterization, fragments and output, and in the last lesson you rewrite it yourself. It notes the shading appears in bands rather than a smooth gradient, and says you can start from any chapter in the contents.

  24. AWS Machine Learning Blog36

    From prototype to production: Engineering agents with Agent DLC

    AWS authors propose the Agent Development Lifecycle (Agent DLC), which ties six activities—Define, Build, Evaluate, Release, Observe, and Feed Back—to an explicit "gold standard" of criteria, thresholds, and approved examples, with an inner loop (score, inspect the path, change one thing, re-evaluate) and an outer loop that updates the standard from production feedback. A reference HR policy assistant shows chunk-level retrieval precision of 1.00 on all three test questions while fact-level precision was 0.59, 0.53, and 0.09, and a prompt-only change improved one question but worsened another, which the authors attribute to missing source coverage. The post also describes a benefits-enrollment answer that scored 0.90 groundedness yet made a false claim about the user's contact preference, and outlines a one-week AWS getting-started plan using Amazon Bedrock AgentCore Evaluations and Observability.

  25. AWS Machine Learning Blog48

    Fine-grained access control for Amazon Bedrock Managed Knowledge Base tools

    This post describes an architecture in which all employees share one agent and knowledge base, but each user can search only the documents they are allowed to view, using Amazon Bedrock Managed Knowledge Base ACL awareness exposed as MCP tools through Amazon Bedrock AgentCore Gateway. The design keeps the LLM from deciding who the user is: a request interceptor injects the verified email from the JWT into userContext, Policy in AgentCore checks it against the token's email claim, and a resource-based policy denies searches from principals other than the gateway execution role. It also covers prerequisites, ACL metadata files, tool schema configuration with parameterOverrides, and a verification script that tests each defense.

  26. AWS Machine Learning Blog52

    Build composable, interoperable multi-agent systems with A2A on AWS

    This AWS post presents a deployed reference architecture in which a LangChain-based enterprise orchestrator in one AWS account calls a Strands Agents-based compliance SaaS agent in another account using the Agent-to-Agent (A2A) protocol, with neither side depending on the other's framework or SDK. It covers cross-framework interoperability, bidirectional agent-to-agent callbacks, and OAuth authentication in both directions using separate Amazon Cognito user pools, with deployment from three Jupyter notebooks and an end-to-end test.

  27. AWS Machine Learning Blog34

    Migrate to Amazon Bedrock Managed Knowledge Base with your coding agent

    AWS added a knowledge base migration guide to the amazon-bedrock skill in the Agent Toolkit for AWS, which lets a coding agent plan a move from a customer-managed knowledge base to Amazon Bedrock Managed Knowledge Base, starting with read-only discovery and a plan you review line by line. On approval, the agent creates the new knowledge base and data sources alongside the original or writes a migrated infrastructure as code template, while ingestion and updating the application's Retrieve calls remain with your team. The post covers why AWS recommends the move, prerequisites, installation, planning, updating retrieval calls, and validating retrieval before cutover.

  28. Claude Blog40

    The Claude Tag guide for startups

    Anthropic's guide describes how eight high-growth startups use Claude Tag (beta) in Slack, covering common use cases such as on-call triage, turning threads into pull requests, a company knowledge base, self-serve data questions, and routine reporting. It includes example channels, connectors, starter instructions, and reported results such as a PR opened 21 minutes after a Slack mention and 134 merged PRs at NavigateAI. A getting-started checklist notes Claude Tag is in public beta on Team and Enterprise plans and requires an org Owner and Slack admin to connect.

Nov 12Thu
  1. Hacker News AI58

    Tests as data: Ensuring AI-generated code does what we want

    The author reports that developers at Thoughtworks distrust their tests because AI coding agents refactor both code and tests as features are added, and argues AI-generated tests are often voluminous, hard to read, coupled to implementation via mocks, and may confirm current behavior including bugs. He proposes customer tests written as data in a DSL (tables, structured scenarios, textual projections of UI and state) interpreted by a small test runner that AI can generate, drawing on Fit, Specification by Example, Go's tabular tests, and Approval Testing.

  2. Hugging Face Blog38

    The Definitive Fine-Tuning Guide for 3D Generation Models

    This blog provides fine-tuning scripts and Docker containers for adapting TRELLIS.2 to custom 3D assets, runnable locally or on Hugging Face Jobs. Using LoRA adapters on only the sparse-structure and shape flow DiTs, trained on 188 GSO footwear objects, the authors report the mean F-score on nine held-out objects rising from 0.276 to 0.803, Chamfer distance falling from 0.174 to 0.033, and normal consistency rising from 0.530 to 0.804. The texture flow DiT is frozen because the dataset lacks the separate material maps that stage requires.

  3. PyTorch Blog40

    Integrating SGLang into Spyre in Under a Week with AI

    A two-person team used AI coding agents to integrate SGLang with IBM Spyre in under a week, producing a plugin of roughly 1,500 lines built on existing PyTorch Spyre support and SGLang's extension points. SGLang delivered about 3.6× the throughput of the reference PyTorch inference implementation using torch-spyre and was on par with vLLM on the measured workloads for Granite 3.3 8B and Gemma 4 26B A4B. Prefix sharing through RadixCache worked without any Spyre-specific changes.

  4. Hacker News AI22

    Writing with AI – Tales and Tips

    Researcher Juri Opitz describes how asking ChatGPT or other LLMs for feedback on drafts yields overwhelming lists of issues that recur even after fixes, with some suggestions useful for fluency and grammar but others stripping his natural voice or distorting meaning. He advises writing the first draft without AI, using it in carefully dosed amounts, challenging conceptual critiques by asking what would change the AI's view, and never blindly pasting suggested rewrites, noting human feedback remains most useful.

  5. Hacker News AI38

    Don't build your own harness: lessons from building Paywalls AI Editor

    RevenueCat describes how its Paywalls AI editor initially manipulated a custom JSON paywall schema through a bespoke harness, producing results that lacked taste despite model, prompt and skill tweaks. The team switched to letting the agent edit a JSX representation of the paywall with familiar Read, Write, Edit and compiler tools, which set a record on their blank-paywalls eval two days after the prototype and let them cut over half the system prompt and almost all skills. After rollout to 100% of users on September 2nd, positive feedback across roughly 46K requests grew 83% and negative feedback fell 40%, with gains concentrated in large edits, new paywall creation and anything requiring taste.

  6. PyTorch Blog58

    Training a 1T-Parameter MoE on EU public infrastructure with PyTorch and TorchTitan

    The EUROPA consortium (Domyn, Fraunhofer IAIS and IIS) describes scaling pretraining of a ~951.9B-parameter MoE model (32.8B active per token) on JUPITER, the EuroHPC exascale supercomputer at Forschungszentrum Jülich, raising throughput from 90 to ~1,300 tokens/sec per GPU across 384 NVIDIA GH200 GPUs per DP replica. Key optimizations built on PyTorch and TorchTitan include FlexShard placement for the Muon optimizer, activation-checkpoint offloading to the Grace CPU, and running FSDP's all-gather setup on the compute stream via unshard_async_op, freeing ~14 GiB of HBM per GPU. The fixes were upstreamed to PyTorch and TorchTitan, with training reaching ~1,300 tokens/s per GPU on 1,536 GPUs across 4 DDP replicas.

  7. Hacker News AI60

    Show HN: A free course from engineer to AI engineer, one Git tag per episode

    A free YouTube-based course, "From engineer to AI engineer," teaches experienced software engineers with no generative-AI background to build a retrieval-augmented generation (RAG) pipeline in eight episodes, with a second seven-episode series, "Prove it works," building an evaluation harness on the same repository. Each episode is one commit and one Git tag (s1e1–s1e8 and s2e1–s2e7), so checking out a tag reproduces the repository exactly as that episode leaves it. The series uses TypeScript, SQLite, curl and an HTTP server, answering with claude-sonnet-5-5 over a fictional 26,000-word corpus (Lantern Supply Co.) that includes deliberate traps such as near-identical product names and a policy changed in later release notes.

  8. Hacker News AI50

    Why AI Coding Agents Lose Context After the PR, and How to Keep It

    AI coding agents can write and test code but lose the delivery evidence produced after a pull request—failed builds, security findings, latency regressions—unless it is carried back into the workflow. The post describes Harness's Code Context Engine (from its Augment Code acquisition) for code context and its Software Delivery Knowledge Graph for delivery context, citing Augment's benchmark of 80%, 71%, and 30% improvements and Harness's internal estimate of about 12K tokens per graph query versus 250K+ via raw APIs. It presents a CVE workflow combining both, notes the two-way link between them is planned rather than shipped, and covers governance via policy as code, RBAC, and approval gates.

  9. Hacker News AI58

    What it takes to operate a SaaS when AI agents write all the code

    A solo founder describes running DocSpring, a document automation SaaS, with AI agents handling much of the coding, bug fixing, Sentry error triage, performance work and CI, coordinated across 49 Claude Code sessions. He reports shipping long-deferred features such as Word templates, a Page Designer and API v2, and clearing his inbox and error queues, while noting that agents can drift into unwanted behavior changes, as when a date fix expanded into an unplanned date-parser project. He concludes that he still sets priorities, approves changes and decides what to promise customers, and that AI does not by itself protect the business from competitors using the same tools.

  10. AWS Machine Learning Blog36

    Build autonomous AI agents with Amazon Bedrock and Spring AI

    This post shows how Java teams can build autonomous AI agents on the JVM using Spring AI and Amazon Bedrock AgentCore, with a single-agent travel assistant demonstrating tool calling, streaming, and memory, and a multi-agent intelligent document processing system with five specialist agents, a SpEL rules engine, and alerting. It covers the AgentCore runtime contract (@AgentCoreInvocation for POST /invocations and GET /ping), @Tool methods, structured output to Java records, memory advisors, and production patterns including Resilience4j retries, twelve-factor configuration, least-privilege IAM, Bedrock Guardrails, and Micrometer observability. Deployment uses an ARM64 image pushed to Amazon ECR and provisioned through modular Terraform, with an optional Amazon EKS and IRSA path.

  11. OpenRouter Announcements58

    Self-Hosted LLM vs API: When Does Self-Hosting Pay Off

    At October 8, 2026 prices, a 27B model on one rented RTX 3090 breaks even against API pricing at roughly 6–12% load, while DeepSeek V4.1 Flash on four rented H200s never beats the cheapest zero-data-retention API endpoint. The post computes break-even load as GPU hourly rent divided by the API cost of the same hour of tokens, using live per-provider prices and benchmark throughput, and notes that published self-hosting costs near a cent per million tokens mostly count cached input tokens that are also nearly free via API. It also covers privacy options such as provider.zdr and EU/US in-region domains, and suggests self-hosting a steady base load while sending peaks to the API.

  12. OpenRouter Announcements36

    How to Benchmark LLMs

    A guide to benchmarking LLMs by writing tasks with script-checkable answers, grading with code, and comparing models on identical tasks via OpenRouter using graded accuracy, billed cost, and request time. On 60 synthetic tasks, GPT-6.1 Sol passed 180/180 trials while DeepSeek had the lowest response cost at 157/180 passes, and agreement with the public Artificial Analysis ranking was weak (Spearman correlation 0.109). The article includes a downloadable Python companion package with offline checks, configuration examples, and guidance on repeats, missing answers, and run comparison.