iCreatAI News
Back to main site
Skip to content
#Tutorials
Nov 4Wed
  1. Hacker News AI62

    MCP vs. API vs. CLI: How should AI agents connect to your app?

    This post argues that an MCP server should be the first choice for making a web app agent-friendly, because MCP standardizes both server and client behavior, lets the client handle OAuth so the model never sees the token, and can route destructive-action confirmations directly to the human via elicitation. REST APIs remain better for scripts, bulk jobs, data exports and integrations that run without an agent, while CLIs suit agents with a shell working with local files. The author recommends putting business logic in operations and exposing them as thin adapters over MCP, REST and optionally a CLI, using the open-source Linkly link shortener as the example.

  2. Hacker News AI22

    Snowflake's AI_PARSE_DOCUMENT got an agentic mode

    A practitioner tested Snowflake's new AGENTIC mode for AI_PARSE_DOCUMENT on a SharePoint-fed Cortex Search pipeline, finding it costs a flat 8.18 credits per 1,000 pages versus 0.68 for OCR (about 12x) and only paid off on slide decks, where it returned chart, diagram, table and infographic summaries absent from OCR text. Two undocumented behaviors surfaced: image summaries live in each page's images array rather than the page content, and the mode parameter must be a literal, since a CASE expression silently returns NULL. The author settled on AGENTIC for PPTX and OCR for everything else, appending only the four useful image types to page text.

  3. Lovable Blog50

    How we built Lovable, and why we keep rebuilding it

    Lovable's founder recounts how the platform evolved from the 2023 GPT Engineer project into a product where users have created more than 70 million projects, repeatedly reinventing its agent design—from complex async subagents back to a predictable tool chain, then to "knowledge files" with retrieval instead of fine-tuning. He describes stateful sandboxes that keep development servers running for faster edit-check cycles and a framework called Camille for forking and merging entire running computers. He closes with three open problems: automating the full software lifecycle, making all types of software seamless to build, and building a better software stack.

  4. Hacker News AI58

    Webflow wanted $2,500 for one language. AI rebuilt our whole site in a weekend

    Rezi.ai describes moving its ~515-page site off Webflow to an Astro 7, React 19 and Tailwind CSS 4 stack hosted on Netlify, with Claude Opus 5.5 doing the heavy lifting, after Webflow quoted $2,500 a year to add a single language on top of a roughly $30,000-per-year enterprise contract. The team first tested the approach on two smaller brands, then kept the Rezi.ai rebuild visually identical using a script that compares every page to the Webflow original, and reports desktop PageSpeed rising from 38 to 99 and mobile from 57 to 95.

  6. Hacker News AI58

    A private AI box in one apply

    Conn3x now has circuits for Ollama and Qdrant, and its Ollama repo includes an example that builds a private retrieval-augmented generation box on a DigitalOcean droplet with a single conn3x apply. The setup runs Ollama and Qdrant on loopback only, with SSH as the sole open port, and the post walks through storing a note, querying it by meaning, and having qwen2.5:0.5b answer. The author reports 101 resources applied in 2 minutes 36 seconds, a clean second plan, and a 41-second destroy.

  7. MarkTechPost58

    NVIDIA Shows 3 Ways to Speed Up MoE Inference on DGX Spark, Lifting Ling-3.0-flash Throughput 42.5%

    NVIDIA's technical blog describes three stacked optimizations for running Ant Group inclusionAI's 124B-parameter Ling-3.0-flash (5.1B active) on a single DGX Spark at batch size one: Humming MXFP4 MoE kernels, an FP8 LM head, and built-in NEXTN speculative decoding. Across all 1,319 GSM8K questions, decode throughput rose 42.5% (37.81 to 53.89 tokens/s) and mean latency fell 24.5%, with accuracy essentially flat (96.89% to 96.82%). NVIDIA notes that combining the FP8 LM head with MTP requires a shared LM-head module, otherwise the framework silently falls back and throughput drops below the unoptimized baseline.

  9. Hacker News AI34

    An easy way to try local AI with braindead-local-intelligence

    Solène introduces braindead-local-intelligence, a project that runs podman containers for an OpenAI-compatible LLM server, an image generation server, SearXNG web search and Open WebUI behind a single token-protected address, loading only one model at a time and unloading it after a few minutes idle. It requires a Linux system with rootless podman 5.0 or newer, targets AMD GPUs (tested up to 12 GB VRAM), and offers a documented list of ready-to-use text, vision and image models. Setup involves adding the user to the render and video groups, running bin/ia install and ia doctor, then ia start, with the web interface at http://127.0.0.1:3000.

  10. ElevenLabs Blog18

    What is call transcription? A practical business guide

    Call transcription converts phone or video calls into written text, typically via AI speech-to-text models, either after recording or in real time. Businesses use transcripts for searchable records, quality assurance and coaching, compliance documentation, and CRM logging. Legality depends on jurisdiction, with some regions requiring one-party consent and others all-party consent.

  11. Hacker News AI36

    How Our AI Agents Write a Help Center

    Velu describes building a help center generator that needs only a web address and a login, not source code access, by having agents use the product the way a new employee would. The system evolved from a single agent into a seven-phase pipeline—research, planning, writing, editing, mock data, screenshots and videos, and publishing—splitting work whenever an agent grew too large. Haiku handles high-volume simple browser work while Sonnet handles planning, writing and editing, with an editor's rules stored in a customizable SKILL.md.

  12. Google Developers Blog58

    Cost per task: Using cheap classifiers to route across Gemini models and thinking levels

    The article argues that cost per successful task, not price per token, governs production economics for coding agents, and reports a 50-question auto-graded benchmark routing tasks with TypeSafe's Jev classifier across Gemini configurations. A frozen regex and prompt-length heuristic scored 93.8% on 16 development questions but fell 23.2 points to 70.6% on 34 held-out tasks, while Jev agreed with hand labels on all 34 and 48 of 50 overall; capping maxOutputTokens at 300 saved nothing on easy tasks and broke 7 of 25 hard ones. A router sending tasks to Flash-Lite first and escalating Jev-flagged complex ones to Flash on low thinking solved all 50, as did Flash on low thinking alone, and the article notes caching favors keeping long context-heavy turns on one model.

  13. Meta Engineering34

    F3 Lightning: How Meta Accelerates Ads ML Features to Production at Scale

    Meta introduces F3 Lightning, an infrastructure layer on top of its Facebook Feature Framework (F3) that decouples feature releases from shared execution so each of the 20,000+ features in an ads ranking request can be compiled, validated, deployed, and rolled back independently. To preserve F3's efficiency, it uses stable computation identifiers, a single shared execution plan, one shared memory buffer per request, and dense index mapping for direct array access. Meta reports comparable extraction features reach production about 3.6x faster than prior F3 with no increase in infrastructure compute cost.

  14. Hacker News AI58

    Solve the cold start problem with an AI outreach system

    Dhruv Dang outlines a Claude-based direct outreach system for a B2B SaaS startup that sources leads from law firm websites, classifies each lead with a script using explicit yes/no signals to derive priority, and emails founders a daily report with personalized draft emails. The system stores leads and scraped sites in Claude docs, runs a scheduled prompt at 9am, and notes that cold email is regulated under CASL in Canada and CAN-SPAM in the US.

  15. AWS Machine Learning Blog38

    Building ultra-low-latency voice AI applications on AWS

    This AWS post explains how Cartesia, an AWS Partner, uses State Space Models (SSMs) and AWS infrastructure to deliver sub-100ms text-to-speech (Sonic) and speech-to-text (Ink) at global scale. It details two deployment patterns for Sonic: an Amazon SageMaker AI model package via AWS Marketplace running in-VPC, and self-managed containers on EC2 GPU instances, and describes Cartesia's voice models operating inside Amazon Connect.

  16. AWS Machine Learning Blog34

    Build a semantic layer for agentic AI with EDB Postgres AI and Amazon Bedrock AgentCore

    This post shows how to build a semantic layer for agentic AI using EDB Postgres AI's AIDB extension, where business meaning lives in Postgres COMMENT statements, join paths come from foreign keys, and governance uses GRANT and row-level security through the built-in MCP endpoint. A Strands Agents agent on Amazon Bedrock AgentCore queries the layer via AgentCore Gateway and an ALB to answer customer 360 questions, with two Postgres roles receiving different results for the same prompt based on column-level grants. AIDB also captures OpenTelemetry traces for each SQL execution as rows in the aidb_otel schema, queryable with ordinary SQL.

  17. AWS Machine Learning Blog22

    How Gen Digital built Norton Revamp’s AI-curated News Engine with Amazon Bedrock AgentCore Web Search

    Gen Digital describes how its Norton Revamp professional reputation platform, launched in March 2026, uses Web Search on Amazon Bedrock AgentCore as a managed, MCP-compliant connector to ground personalized news feeds, content suggestions and goal recommendations in cited, real-time web results. The post details an event-driven architecture using Amazon Nova Lite, Bedrock and an AI Learning System for ranking, plus reported results including 99.9 percent uptime and 3x faster feature implementation.

  18. AWS Machine Learning Blog36

    ServiceNow on-behalf-of token exchange with Bedrock AgentCore Gateway

    This post explains how to configure Amazon Bedrock AgentCore Gateway for an on-behalf-of (OBO) token exchange with Microsoft Entra ID so an AI agent can call the ServiceNow REST API while preserving the calling user's identity. It covers creating Entra application registrations for the agent and a ServiceNow API connector, configuring a ServiceNow inbound OAuth integration with an OIDC provider, and setting up the gateway with MCP version 2025-11-25, an outbound OAuth client using the JWT authorization grant, and a target with the token exchange grant (scope plus requested_token_use: on_behalf_of). The post states that native OBO support removes the need for custom AWS Lambda interceptors to perform token exchanges.

  19. AWS Machine Learning Blog25

    Bring structured and unstructured data together with Amazon Quick

    This AWS post walks through connecting an Amazon S3 bucket as a knowledge base, enriching a Product_Sales.csv dataset with semantic descriptions, and combining both in a Space that a custom chat agent queries through a single conversational interface. The agent uses text-to-SQL for structured questions, cites sources across both data types, and can generate PowerPoint, Excel, and Word artifacts from chat prompts. It also covers prerequisites, sharing, best practices, security considerations, and cleanup.

  20. Hacker News AI60

    LLMjacking: The AWS, Google Cloud and Azure settings that cap stolen AI keys

    Google's threat team reports that attackers using stolen cloud and AI keys ran up over $100,000 in a day, and up to $1 million, on victims' model bills, entering mainly via stolen IAM credentials, service account keys and GitHub tokens, plus leaked AI keys, open Ollama servers and a LiteLLM bug. The piece converts that advice into exact settings for AWS, Google Cloud and Azure, noting that Bedrock invocation logging and Google's Data Access logs are off by default, that budget alerts do not stop spend, and that tokens-per-minute quotas are the only ceiling that works in minutes. It also finds that the LiteLLM CVE number cited by GTIG is publicly listed as a different bug, and that no public advisory covers the /key/regenerate flaw.

  21. Hacker News AI33

    Show HN: 8 years of "Who's Hiring?" in charts (hiring boom, bust, and AI trends)

    The author used Codex to run semantic analysis on eight years of Hacker News "Who's hiring?" posts, from pre-COVID through October 2026, retrieved via the HN API. Five findings: AI or machine learning appeared in 50% of 2026 posts (up from 17% in 2019); 53% of identified employers appeared in only one of the 94 monthly threads; frontend roles fell from 16% to 3% of posts; staff-level posts rose from 77 in 2019 to 277 in 2026; and only 41 of 2,932 posts in 2026 explicitly named junior roles, versus 163 in 2019. The post also lists 20 other trends.

Nov 3Tue
  2. Hacker News AI38

    Rooting an Echo Plus 2 with AI

    The author used OpenAI's Codex coding agent to help root an Amazon Echo Plus 2, starting from a MediaTek BootROM exploit (Kamakiri via MTKClient) to install an app, then chaining the CVE-2019-2215 Binder use-after-free with an epoll/iovec trick to overwrite selinux_enforcing and core_pattern and gain UID 0. Root returned automatically after cold boots, succeeding 50 out of 50 stock reboots, and the device runs Kiosk Satellite with Home Assistant and Music Assistant on the stock audio stack. The author notes the unit's October 2026 kernel still lacks the 2019 Binder fix, and that a software update could close the route.

  3. Hacker News AI38

    2M databases and counting: the architecture behind Sauna's AI teammate

    Sauna, a proactive AI work assistant launched in April 2026, gives every session, agent, connected inbox and calendar its own database, totaling over 2 million databases with about 90,000 created daily. The piece details how Sauna uses Turso Sync for chat histories and synced copies of inboxes and calendars, after earlier attempts with Cloudflare Durable Objects and Litestream, and explains how per-database isolation supports sharing, fast local search and retention. It also covers Sauna's plans to consider moving its in-house file system to Turso and to experiment with MVCC.

  4. Hacker News AI40

    A Human Corrected the AI. Then the Data Changed

    Zingg describes how adding a simple steward approval flag to its incremental entity resolution and MDM platform raised harder questions about what a human decision locks in, how long it holds, and who may reopen it when underlying data changes. The post lays out design principles (members stay put, no automatic merges, growth allowed), walks through concrete edge cases such as a father-and-son household split, and argues that the system being corrected should not control when a human decision expires. It closes with general guidance for any system where agents and people share decisions over time.

  5. Hacker News AI38

    AI agents hallucinate prices: LLM grounding with live data

    Tests show that GPT-6 Luna and Claude Opus 5.5 each matched the live price for only 2 of 15 products from memory, and built-in web search improved exact matches but still produced wrong prices with source links on every run. The article argues that LLM grounding requires live data, a clear status when a check fails, and code rather than the model writing each price into the answer, and it builds a Python agent using Apify's E-commerce Scraping Tool to fetch prices from Amazon, Walmart, Best Buy, Target and Shopify stores.

  6. Hacker News AI62

    AI Red Team Academy: Free, open-source course on breaking AI systems

    AI Red Team Academy is a free, open-source course on offensive security testing of AI systems, spanning 8 modules and 60+ hours of content with hands-on Docker labs covering topics from prompt injection to supply chain attacks. The curriculum covers frameworks such as MITRE ATLAS, OWASP Top 10 for LLM Applications, and NIST AI 100-2, and includes labs using tools like garak, PyRIT, and promptfoo. All labs run locally via Docker and require no cloud API keys or external services.

  7. Hugging Face Blog58

    Transformers Model Integration Contracts

    Transformers introduces Model Integration Contracts (MICs), which pair a shrunk fixture model built from a real checkpoint's configuration with a runnable usage path and human-reviewed baselines, run on cheap CPU runners in PRs. Replaying frozen MICs over five months of commits showed a small UMT5 fixture flagged a decoder-attends-to-future-tokens regression from July 30 to September 28, 2026, which a forward pass alone would have missed. The tmic CLI and agents help draft contracts, and the MIC CI job currently reports failures without blocking PRs.

  8. Hacker News AI38

    Code-First Internal Tools vs. Low-Code in the Age of AI

    OpenRun, an open source self-hosted GitOps platform for deploying web apps and internal tools declaratively, argues that AI coding tools have removed most of the speed advantage of low-code platforms' visual editors, since generating a small app from a prompt now takes about as long as assembling it on a canvas. The post contends that low-code downsides remain—proprietary app formats, editor-only editing, limited testing and refactoring, per-user pricing and security features gated to higher tiers—while code-first apps benefit from reviewable pull requests, easier maintenance and no lock-in to a tool or model. It also notes that code-first internal tools still need a platform for hosting, SSO, access control, audit logs, database credentials and deployment, which is the gap OpenRun targets.

  11. NVIDIA Technical Blog25

    Keep Distributed Training Running with Automated Workload Recovery

    As distributed AI training scales, failures across hardware and software dependencies increasingly disrupt training progress, making fast, reliable recovery essential. The material notes that a stalled training rank can trigger an NCCL timeout, a GPU node can stop responding to the scheduler, or a straggler can slow the entire job, and that each disruption is manageable on its own.

  12. ElevenLabs Blog20

    How are synthetic voices created for natural speech?

    Synthetic voices are built by training neural AI models on large datasets of real human speech, then generating audio from text or cloning a specific speaker from recorded samples. Modern systems sound natural because they replicate prosody and interpret context rather than stitching together individual words or syllables. Voices can also be designed from a text description alone, without any recording.

  13. Hacker News AI60

    AI evals for agents: what they are and how to write them

    This guide defines the vocabulary of AI evals (task, trial, grader, transcript, outcome, suite) and explains how agent evals differ from LLM evals: they grade a whole run, including tool calls, arguments, resulting state and final reply, not just one response. It covers the three grader types (code, model/LLM judge, human), the distinction between capability and regression evals, and pass@k versus pass^k, then walks through writing an agent eval in seven steps with a worked refund example.

  14. Hacker News AI38

    AI assistants read a Shopify store 10x more often than people click through

    A 30-day analysis of a headless Shopify fashion storefront found that ChatGPT and Claude fetched pages 1,264 times for users while GA4 recorded only 122 visits from AI answers, with ChatGPT alone reading pages about nine times per visit. The logs also showed assistants repeatedly requesting discontinued products that returned soft 404s, Claude probing standard Shopify paths the storefront doesn't serve, and roughly 10,600 requests from fake AI crawlers that failed IP verification and mostly probed for secrets. The authors note their own dashboard initially overstated ChatGPT arrivals by more than tenfold due to utm tag handling in referer headers, and recommend checking 404 status codes, adding redirects for standard Shopify paths, and verifying bots by IP.

  15. Hacker News AI58

    AI video editing in DaVinci Resolve and MCP – workflow demonstration and test

    A test of DaVinci Resolve Studio 21.1's AI Assistance via an MCP server let Claude or Codex drive real editing actions, including building an edit from multiple takes, adding jump cuts and zooms, generating titles, duplicating timelines for branded variants, and exporting a verified 4K file. A first run with GPT-6 Astra (Medium) cut 23:58 of raw footage to about 7:14 in roughly 15 minutes 30 seconds, but Resolve's built-in transcription missed words and hurt cut accuracy. Supplying a Deepgram transcript with word-level timestamps let the AI rebuild a cleaner timeline in about 4 minutes 47 seconds, and estimated costs for a full run were $0.10–$0.15 for transcription plus $13–$19 for editing and generation work.

  16. Hacker News AI34

    Sunscreen, a stylus, and 16 hours of pair programming with an AI

    A kernel developer describes using Claude to add handwriting recognition to KDE's plasma-keyboard on-screen keyboard, after finding Qt Virtual Keyboard's handwriting mode is gated behind a compile-time flag requiring proprietary engines. Over roughly 15–16 hours of active work he produced four commits (~5,000 lines across 32 files) including a handwriting mode, editing keys, scribble-to-correct, and 79 tests, using PhatWare's GPL-3 WritePad engine (6.75% character error rate) after a self-trained model failed at 82.93%. He credits the AI with speed and fluency across unfamiliar stacks, but argues a human was load-bearing for catching wrong assumptions, insisting on measurement over plausible fixes, and product judgment.

  17. AWS Machine Learning Blog38

    How Circle K accelerated DevOps incident triage to under 15 minutes with Amazon Bedrock

    Circle K built an agentic AI triage system on Amazon Bedrock that investigates Grafana alerts across AWS resources, OpenSearch logs, and GitHub code, then posts a structured root-cause summary to Slack. Initially built on Amazon Bedrock Flows, the system was migrated to a single agent on Amazon Bedrock AgentCore Runtime with AgentCore Gateway brokering data source access. Circle K reports mean triage time dropped from 45–60 minutes to under 15 minutes, a 67–75 percent reduction based on internal incident-response metrics.

  18. AWS Machine Learning Blog40

    Agentic engineering intelligence with Amazon Nova Multimodal Embeddings and Amazon Bedrock AgentCore

    This post describes a multi-agent engineering assistant built on Amazon Bedrock AgentCore, Amazon Nova Multimodal Embeddings, and the open source Strands Agents SDK that assembles evidence for aerospace part substitution decisions. A planner routes queries to Catalog, Evaluation, and Knowledge specialists, with deterministic Python tolerance-rule scoring, Nova Multimodal Embeddings for multimodal retrieval and photo cosine similarity, and a supervisor that synthesizes results only when multiple specialists contribute. The engineer remains the decision-maker, and the system drafts justification memos and Integration Impact Assessments with audit trail entries aligned with AS9100 sections 8.4 and 8.5.6.

  19. AWS Machine Learning Blog36

    Build efficient agents with knowledge distillation on Amazon SageMaker AI

    This post explains how to use sequence-level knowledge distillation (SeqKD) to fine-tune a smaller shopping agent on Amazon SageMaker AI, using a frontier teacher (GLM-5.3 on Amazon Bedrock) to generate multi-step tool-use trajectories that are filtered by ShoppingBench's programmatic scorer and used for serverless supervised fine-tuning of a Qwen3.5-4B student. The distilled student reached an average Absolute Success Rate of 0.541 on 324 held-out ShoppingBench problems, versus 0.297 for the base model and 0.540 for the greedy teacher, while IFEval general instruction-following dropped roughly 6 to 8 points. The post also covers per-step trajectory encoding, LoRA training configuration, a low-cost model-evaluation pre-check, and serving the base model and student adapter as two inference components on one real-time endpoint.