SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World

Programming

LatestTopDiscussed
0
ProgrammingDEV Community ·

ChatGPT Hits 1 Billion Weekly Users, Pushing Marketers to Track AI Referral Traffic

OpenAI announced in August 2026 that ChatGPT has surpassed 1 billion weekly active users, marking a significant milestone in AI-driven consumer behavior. As more people use AI assistants to research topics and evaluate options, marketers are being urged to monitor whether these interactions generate measurable website visits and conversions. Separately, Perplexity was reported by TechCrunch in June 2025 to have processed around 780 million queries in May 2025, underscoring the scale of research activity across AI platforms. Experts recommend that website owners create dedicated analytics segments for AI-platform referrals to establish reliable traffic baselines rather than relying on broad platform statistics. The priority is accurate measurement first, so businesses can determine whether AI-sourced visits are reaching key pages and leading to meaningful outcomes.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

GitHub Repo Indexes GPT-6 Astra Robotics Demos Across Simulation and Real-World Tasks

A public GitHub repository called Awesome-Astra-Embodied-AI has catalogued reported demonstrations of GPT-6 Astra's role in robotics across 35 cases spanning simulation, real-world deployment, reinforcement learning, and real-to-sim workflows. The index shows that Astra does not function as a single fixed controller but fills different roles across cases, including high-level planning, task inference, direct pose output, and RL environment construction. In some demonstrations, separate systems like GEAR-SONIC or pretrained policies such as FluxVLA handle low-level execution, meaning capabilities cannot be attributed solely to Astra. Real-world cases cover tasks like plug insertion, cucumber slicing, and mobile manipulation, while simulation entries include Unitree G1 navigation and a pen-spinning task trained with PPO in Isaac Lab. The repository's own case notes flag constraints such as pregrasped starting states and kinematic replays, which limit how broadly the demonstrated results can be interpreted.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Tencent Open-Sources BrowserSkill to Let AI Agents Use Your Logged-In Browser Safely

Tencent has released BrowserSkill, an open-source tool that allows AI coding agents such as Cursor, Claude Code, and Codex to operate within a user's real, already logged-in browser without disrupting their workflow. The tool consists of two local components — a command-line daemon called bsk and a browser extension — communicating entirely over local connections with no cloud routing involved. Agents interact with the browser through a dedicated Agent Window kept separate from the user's normal tabs, and must explicitly request and return any existing tab they need to access. Because the extension runs inside the user's actual browser profile, agents can access authenticated sessions without requiring credential sharing or separate test accounts. When a task requires human input, such as solving a captcha, the agent can pause and hand control back to the user before resuming.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Uber Engineering Explains Its Strategy for Preventing Retry Storms

Uber has published a technical blog post detailing how its engineering team defends its systems against retry storms. Retry storms occur when large numbers of failed requests are automatically retried simultaneously, potentially overwhelming backend services. The post outlines the specific mechanisms and strategies Uber employs to mitigate this cascading failure risk at scale. The article was shared on Hacker News, where it received five upvotes at the time of reporting.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

LLM Routing Tools Compared in 2026: Bifrost Leads With 11-Microsecond Overhead

Model routing tools have evolved to dynamically direct AI inference requests across multiple LLM providers based on cost, latency, and provider health. Production applications that send all prompts to a single frontier model typically overspend by 40–80% on simple queries while remaining vulnerable to provider outages and rate limits. Bifrost, an open-source AI gateway built in Go by Maxim AI, tops the 2026 rankings by delivering sub-millisecond gateway overhead of just 11 microseconds at 5,000 requests per second. Other notable tools include open-source proxy LiteLLM, classifier-based RouteLLM, and hosted aggregator OpenRouter, each addressing different routing needs with distinct trade-offs. Engineering teams evaluating routing platforms are advised to assess six key dimensions: routing logic expressiveness, runtime latency, failover capability, provider coverage, governance controls, and deployment flexibility.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Ex-OpenAI Researcher Launches Jev, a Decision-Only AI Model That Cannot Generate Text

Diogo Almeida, the OpenAI researcher behind RLHF and InstructGPT, has launched Jev through his startup TypeSafe AI, releasing it in early access on 15 September. Unlike large language models, Jev is designed solely to make structured decisions, returning one of three output types — a Choice, a Score, or a probability — each paired with a calibrated confidence value. The model completes tasks in 70 to 500 milliseconds and charges $0.042 per million input tokens, with no fee for output due to its non-autoregressive architecture. Almeida argues that chat was the wrong target for advancing AI automation, and that reliable, fast decision-making with trustworthy confidence scores is what software pipelines actually need. TypeSafe AI raised $40 million in funding and built the team from alumni of OpenAI, Google Brain, Meta, and FAIR, with Jev's benchmarks claiming it runs up to 193x faster and 444x cheaper than frontier language models on comparable workflows.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Poor database indexing caused 'not found' API queries to take up to 30 seconds

A maritime tracking API was found to be dramatically slower when returning empty results than when returning actual data, a problem traced to misconfigured TimescaleDB compression settings. On September 9, 2026, queries for a vessel's latest port event took 28 milliseconds when data existed but 2 to 2.5 seconds when it did not, while ETA lookups for unmapped vessels could stall for 30 seconds or more. Over a single night, 'not found' requests consumed 71% of all database time spent on the ETA endpoint. The root cause was that compressed data chunks lacked usable indexes on the columns being queried, forcing the database to perform a sequential scan across hundreds of compressed chunks for every miss. Grouping API metrics by route template rather than raw path revealed the pattern, showing that slow responses were consistently tied to empty results rather than specific vessels or endpoints.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Critical MikroTik RouterOS Flaws Allowed Full Takeover Without Credentials

CERT Polska disclosed a two-vulnerability exploit chain called MikroTrick in September 2026, targeting MikroTik RouterOS devices reachable via SSH over the internet. The first flaw, CVE-2026-67276, is an SSH public-key authentication bypass caused by a logic error that ignores the RSA exponent during key verification, allowing unauthenticated access. The second, CVE-2026-86060, is a privilege escalation triggered by a malformed username, granting full administrative control once an attacker is authenticated. Both vulnerabilities carry critical CVSS scores between 9.2 and 9.8, and active exploitation was recorded from September 2, 2026 — one day before MikroTik released patched versions across its 6.x and 7.x release trains. Unpatched devices exposed to the internet should be treated as potentially compromised, and administrators are advised to upgrade immediately or restrict SSH access to trusted addresses as an interim measure.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer builds custom voice-to-report app to cut his doctor wife's late work nights

A software developer built a mobile app for his physician wife after noticing she regularly stayed late at work typing up echocardiography reports following patient appointments. The reports consist of roughly fifty measurements — including chamber dimensions, valve data, and flow velocities — which she previously had to manually enter into structured tables after each shift. An initial attempt using off-the-shelf speech-to-text tools failed because the output was unstructured text, not formatted medical data, saving no time overall. The developer then built a domain-specific solution that understands cardiology terminology, assigns dictated numbers to the correct report fields, auto-calculates derived values like ejection fraction, and exports a finished Word and PDF document. A standout feature lets her tap any measurement to replay the audio from one second before she dictated it, making it easy to verify individual values without scrubbing through the full recording.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Workflow Engines vs. Orchestrator Skills: What Actually Sets Them Apart

A workflow engine is a program that holds a procedure and calls a coding agent only when intelligence is needed, inverting the traditional orchestrator model where the AI model itself drives the process. Unlike orchestrator skills, which store procedures as Markdown in a conversation session, workflow engines run procedures as validated programs in their own processes with persistent run state on disk. This architectural difference allows workflow engines to validate control flow at load time, catching errors like missing nodes before a run begins rather than mid-execution. Engines can also run independent steps in parallel by computing dependency layers, whereas orchestrator skills process phases serially due to their conversational nature. Platforms like Archon and Microsoft Conductor are cited as examples of this engine-based approach, though the article notes not all tools marketed as workflow engines truly qualify.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Developer fine-tunes 27B model to write Atlassian Forge apps, releases weights openly

A developer has fine-tuned the Qwen3.8-27B language model to generate functional Atlassian Forge applications, releasing the weights publicly on Hugging Face under the Apache-2.0 license. The base model previously passed only 15% of Atlassian identifier probes and produced zero valid Forge manifests out of 25 attempts, but after fine-tuning those figures rose to 69% and 14 of 25 respectively. Training was conducted as a rank-32 LoRA over approximately 26 hours on a single Apple Mac Studio M3 Ultra with 96 GB of memory. The model was trained on real Forge apps, official documentation, OpenAPI specs, and Atlassian Community answers covering Jira, Confluence, and Jira Service Management. The developer claims this is the first open-weights model specifically trained to write Forge apps and the first Atlassian-tuned release accompanied by base-versus-tuned validation evidence.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Content-Addressed Tool Shed Eliminates Redundant Builds in Ephemeral Environments

A developer building an ephemeral compute system identified a key inefficiency: every new isolated environment was re-downloading and recompiling the same tools, such as ffmpeg, from scratch. To fix this, they introduced a content-addressed cache called the Tool Shed, which stores derived build artifacts keyed by a hash of their inputs, including pinned package versions and the base environment. When a new environment launches, the system checks the Tool Shed first and reuses existing artifacts if the key matches, skipping redundant downloads and builds. Crucially, the Tool Shed is treated as a disposable optimization rather than a source of truth — the recipe remains authoritative, and the cache can be fully deleted without data loss. This approach also enables offline launches once artifacts are cached, restoring the intended meaning of ephemeral to mean only task state is discarded, not toolchains.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Why Saying No to Features May Be a Product Team's Best Decision

A product strategy article published on liamnugent.me argues that the most critical decisions in product development are about what not to build. The piece contends that teams often underestimate the long-term costs of adding features, including maintenance burden and increased complexity. By choosing restraint, product teams can stay focused on core value and avoid bloating their offerings. The article has gained attention on Hacker News, sparking discussion around prioritization and product discipline.

0 comments Read more at Hacker News
0
ProgrammingHacker News ·

Snapdrop Lets Users Share Files Across Devices Instantly Without an Account

Snapdrop is a file-sharing tool that allows users to transfer files between devices with no setup or account required. The service works instantly, removing common friction points associated with traditional file-sharing methods. It was shared on Hacker News as a community project, attracting early attention from the developer community. The tool appears aimed at users seeking a quick, hassle-free way to move files across different devices. Details about supported platforms and underlying technology were not disclosed in the original post.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

AI-to-AI Payments Could Form the Backbone of a New Machine Economy

As autonomous AI agents grow capable of making independent decisions, the financial industry is exploring what happens when these systems need to transact with one another. Major institutions including the IMF, Visa, and Mastercard are already examining or developing frameworks for machine-to-machine payments, with Mastercard launching its Agent Pay for Machines initiative for programmatic, high-frequency micropayments. Unlike traditional payment systems built around human authorization, AI-driven transactions demand programmability, speed, micropayment support, and strong identity verification. Blockchain and programmable digital assets are emerging as leading candidates to underpin this infrastructure, given their auditability and interoperability. The shift signals a broader evolution in commerce, where algorithms, not people, may increasingly act as economic agents initiating and completing financial transactions.

0 comments Read more at DEV Community
0
ProgrammingHacker News ·

Flet 1.0 Released: Python Framework for Cross-Platform App Development

Flet has launched version 1.0, a framework that allows developers to build cross-platform applications using Python. The release was shared on Hacker News, where it garnered community attention. Flet aims to simplify app development by enabling Python code to run across multiple platforms without requiring separate codebases. Further details and documentation are available on the official Flet website at flet.dev.

0 comments Read more at Hacker News
0
ProgrammingDEV Community ·

Developer Builds Independent APK Platform to Bypass Play Store Restrictions

A developer has launched Aeroapk (aeroapk.store), an independent Android app catalog designed as an alternative to the Google Play Store. The platform allows developers to publish their APKs directly to users, with each listing sourced exclusively from the original app creator. All submissions undergo a safety review before approval, and approved listings remain live until the developer decides to update or remove them. The project was motivated by common frustrations with traditional app stores, including slow review cycles, opaque ranking algorithms, and sudden policy changes. The creator plans to expand the catalog and refine the review process based on developer feedback as the platform grows.

0 comments Read more at DEV Community
← NewerPage 1165 of 3223Older →