标签：#EV

Learn more about Paramount's planned acquisition of Warner Bros. Discovery — a historic Hollywood megadeal valued at $111 billion — as it continues to develop.

The familiar story with Xteink’s tiny e-readers plays out once again with its new X4 Pro: upgraded hardware held back by frustrating software. The company’s latest tiny e-reader adds a touchscreen and lighting, so choosing it over a Kindle or a Kobo feels like less of a compromise — until you use it. The new […]

Tesla has revealed its upcoming Cybercab design with a built-in SpaceX Starlink V5 dish.

It starts at just under $25,000, but the hybrid won't arrive until next year.

I first slipped on Halliday's original smart glasses at CES 2025. I was not a fan. The glasses had a tiny, movable display window embedded into the frame that was incredibly finicky to see, and my 30-minute demo left me with achy eyeballs. It was an interesting concept with terrible execution, especially compared to the […]

Nintendo does a lot of weird things, but Splatoon might just be one of the strangest - or at least the boldest. The series started as an attempt to muscle into the ultra-competitive online shooter space, but in a way that feels a world away from the violence and bombast of Call of Duty or […]

New gaming gear is always promising to shave milliseconds off latency or response time. The improvements usually aren't obvious. But the Glorious GHS InfinitePlay wireless headset, on the other hand, could save you far more than just milliseconds: It has the fastest hot-swap battery system I've ever used, and the headset doesn't die when you […]

Halliday’s G2 glasses can listen to and summarize your workplace meetings—with no video involved.

Building AI Agents That Don't Hallucinate: Structured Workflows, Guardrails, and Per-Step Evaluation How we replaced fragile prompt chains with typed schemas, validation gates, and evaluation at every step — 94% task success vs 60% baseline The Prompt Chain Trap January 2024. We built a "research agent" — 12 prompts chained together: Decompose question → 2. Search planning → 3. Execute searches → 4. Extract facts → 5. Synthesize → 6. Fact-check → 7. Format → ... It worked 60% of the time. The other 40%: Step 3 returned malformed JSON → Step 4 crashed Step 5 hallucinated citations → Step 6 missed it Step 7 output wrong format → Downstream consumer failed No visibility into which step failed Debugging meant reading 12 LLM calls' worth of logs. Adding a step broke three others. The Shift: Agents as Typed Workflows We moved from prompt chains to structured workflows with: Pydantic schemas for every step input/output Guardrails that validate and auto-retry Explicit state machine (not implicit chaining) Evaluation harness per step (not just end-to-end) ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │ Decompose │──▶│ Search │──▶│ Extract │──▶│ Synthesize │ │ Question │ │ Planning │ │ Facts │ │ Answer │ │ │ │ │ │ │ │ │ │ In: Query │ │ In: Plan │ │ In: Results │ │ In: Facts │ │ Out: SubQ[] │ │ Out: Steps │ │ Out: Fact[] │ │ Out: Answer │ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ │ │ │ │ ▼ ▼ ▼ ▼ [Schema] [Schema] [Schema] [Schema] [Guardrail] [Guardrail] [Guardrail] [Guardrail] [Eval: 0.9] [Eval: 0.85] [Eval: 0.9] [Eval: 0.95] Core Abstractions # agent_eval/schemas.py from pydantic import BaseModel , Field from typing import Literal , Any class DecomposeInput ( BaseModel ): user_query : str context : dict = Field ( default_factory = dict ) class DecomposeOutput ( BaseModel ): sub_questions : list [ str ] = Field ( min_length = 1 , max_length = 5 ) requires_tools : bool reasoning : str class PlanInput ( BaseModel ): sub_questions : list [

A practical roadmap for students who want to build real AWS skills, create an impressive portfolio, and prepare for cloud engineering careers. Introduction Every year, thousands of students begin learning AWS. They watch training videos, collect certificates, complete online courses, and share digital badges on social media. Yet when internship interviews or entry-level cloud engineering opportunities arrive, many struggle to answer a simple question: "What have you actually built on AWS?" The cloud industry rewards practical experience, not passive learning. AWS itself focuses beginner learning on hands-on experience with foundational services such as Amazon S3, Amazon EC2, Amazon VPC, Amazon RDS, and cloud security because these services power most real-world cloud environments. If you're a student aiming for a career in Cloud Engineering, DevOps, Site Reliability Engineering (SRE), Solutions Architecture, or Platform Engineering, this article provides a practical roadmap that can help you become job-ready. Why Students Should Learn AWS Cloud computing has become the backbone of modern technology. Companies of every size use cloud platforms to: Host applications Store data Deploy AI workloads Build scalable systems Reduce infrastructure costs Improve reliability As a result, companies continue to hire professionals with cloud skills across software engineering, cybersecurity, DevOps, networking, and data engineering domains. AWS offers dedicated learning paths, hands-on labs, certification tracks, and career-focused programs specifically designed to help learners develop these skills. The key question is not: "Which AWS service should I memorize?" The better question is: "Can I design, deploy, secure, and troubleshoot cloud solutions?" The 5 AWS Projects Every Student Should Build Instead of completing another course, build these five projects. Project 1: Host a Static Website Using Amazon S3 What You'll Learn Cloud Storage Static Website Hosting Bucket Policies O

AI coding tools have gotten very good at one thing: generating code fast. What they haven't gotten good at is discipline. They don't know your architecture. They don't remember that you rejected a pattern last sprint. They don't know which parts of your context window are signal and which are noise. And they have no concept of a development workflow — no phases, no review gates, no verification steps. You describe what you want, they generate, and you hope the output fits. For small tasks this works fine. For anything that touches your real codebase at scale — refactors, new features with cross-cutting concerns, compliance-sensitive changes — the lack of workflow structure creates subtle, expensive problems that compound over time. We've been building Ortho to address two of these problems: workflow discipline (through ASES, a 6-phase AI development methodology) and context discipline (through a 9-component token optimization pipeline). This post explains both. The workflow problem with AI coding tools When a junior engineer joins your team, you don't just hand them a task and say "generate." There's a process: understand the codebase, plan the change, get the architecture reviewed, build it, test it, verify it, get it reviewed. The process exists because individual steps catch different categories of mistakes. AI coding tools collapse all of that into one step. You prompt. You get code. Done. The result isn't always bad code per file. The problem is architectural: the AI has no model of your layer boundaries, so it imports from layers it shouldn't touch. It has no memory of past decisions, so it re-proposes patterns you've already rejected. It has no verification step, so it confidently generates code that looks right but has subtle issues a reviewer would have caught in 30 seconds. The missing piece isn't a smarter model. It's a workflow. ASES: A 6-phase workflow for AI-assisted development ASES (v1.2) is the methodology built into Ortho's orchestration layer. It

A request starts at 10:00:00.900 and finishes 200 ms later. Your latency log says -800 ms . That sounds impossible until the machine corrects its clock between the two reads. Then ordinary code turns an adjustable wall clock into a broken stopwatch. The subtraction that works—until it does not This is probably hiding in one of your metrics helpers: async function timed ( operation ) { const startedAt = Date . now (); const result = await operation (); return { result , durationMs : Date . now () - startedAt , }; } Most of the time, this reports a sensible number. That is what makes it dangerous. Date.now() answers a calendar question: how many milliseconds have passed since the Unix epoch according to this machine? The answer must remain comparable with logs, users, and other computers, so synchronization software is allowed to correct it. A correction can change its rate, jump it forward, or move it backward. If either correction lands between the two calls, the subtraction measures the clock adjustment as if it were work. A timestamp is a coordinate. A duration is a distance. They need different instruments. Your computer already has two kinds of clock Operating systems expose several clocks, but two mental buckets cover most application code: Question Clock JavaScript example “When did this happen?” Wall clock Date.now() , new Date() “How long did this take?” Monotonic clock performance.now() A wall clock follows civil time. It has an epoch and can be serialized as an ISO timestamp. A monotonic clock starts at an arbitrary origin and promises that later readings will not be lower than earlier readings. wall: 1000 ── 1001 ── 0998 ── 0999 clock correction monotonic: 40 ───── 41 ───── 42 ───── 43 On Linux, CLOCK_REALTIME is settable wall time . CLOCK_MONOTONIC cannot be set and does not make discontinuous jumps backward, although gradual frequency adjustment can affect its rate. Linux even has CLOCK_BOOTTIME for a monotonic counter that includes suspended time. Thre

Open the network tab on almost any web app, trigger a failed request, and you'll usually find one of two things staring back at the user: a raw stack trace, or a message so generic it might as well say "something happened." Neither one helps. Both exist for the same reason they were written by developers, for developers, and never translated for the person actually using the product. The Error Message Nobody Designed Most UI elements go through some level of design scrutiny. Buttons get spacing decisions. Forms get validation states. But error messages? They're usually whatever string got thrown at the moment something broke, copy-pasted straight from a try/catch block into a toast notification. "Error 500: Internal Server Error." "Failed to fetch." "Unexpected token in JSON at position 4." These are diagnostic breadcrumbs for engineers debugging a system. To a user trying to submit a form or complete a purchase, they're just noise confirmation that something went wrong, with zero indication of what to do next. This is where good web app design services earn their keep not in the buttons and layouts everyone notices, but in the failure states nobody plans for until users start complaining. Why This Keeps Happening It's not that teams don't care. It's that error handling sits at the intersection of two disciplines that rarely talk to each other at the moment. Backend logic throws whatever exception the code produces. The front end just needs something to display so the app doesn't silently freeze. Nobody's job, at that moment, is to ask: "what should the user actually understand right now?" The result is a UI layer that's polished everywhere except the one place users encounter when things go wrong which, ironically, is exactly when clear communication matters most. What a Good Error Message Actually Does A well-designed error message does three things a raw exception never does. It tells the user what happened, in plain language not "Error: NetworkException," but "W

You ask an LLM for JSON. You get this: json { "name": "test", "valid": True, "items": [1, 2, markdown Three problems in one response: markdown fences the LLM wasn't supposed to add, True instead of true (Python literal), and the response was cut off mid-array. JSON.parse() throws on all three. A linter tells you what's wrong. But you're left fixing it manually. I kept hitting this wall so often that I built a dedicated repair pipeline: AI JSONMedic . Why existing tools don't cut it JSONLint / JSONFormatter — great for valid-ish JSON with one missing comma. Not built for LLM failure modes: they flag errors but don't repair them. jsonrepair (npm) — solid library, handles many cases. AI JSONMedic actually uses it as a last-resort fallback. But it doesn't tell you what it changed, and doesn't handle all the LLM-specific cases we needed. The 14 failure modes we target Through building this, we catalogued how LLMs specifically break JSON: Markdown fences — ` json wrapping the output Trailing commas — [1, 2, 3,] (the model "runs out" of items but adds one more comma) Python literals — True , False , None instead of true , false , null Single quotes — {'key': 'value'} instead of double quotes Smart quotes — "key" (curly quotes from copy-paste) Unclosed brackets — truncated at max_tokens mid-array or mid-object Unclosed strings — "value without closing Concatenated objects — {"a":1}{"b":2} when streaming produces multiple chunks NDJSON — newline-delimited JSON that needs wrapping Python-style comments — # this is a comment inside JSON JavaScript-style comments — // inline or /* block */ Escaped backslashes — \\n instead of \n Duplicate keys — same key appearing twice (ambiguous — we warn, not silently pick) BOM / encoding issues — UTF-8 BOM at start of response What makes the repair pipeline different Each pass targets one failure mode. The order matters — strip fences first, then normalize quotes, then fix commas, then close truncated structures. Each change is tracked. The

When you wire up a payload cms live preview , you’re not just plumbing a URL—you’re building a contract between the admin panel and your Next.js App Router. The goal is keystrokes-ago fidelity: editors click Preview, land on your front end, and see exactly what’s in the draft, even when the public site is cached to the hilt. Our implementation at techpotions settled on one preview route, one shared secret, and one shared URL builder. Here’s every decision that made it work. One /next/preview route for the entire site The admin panel’s preview button doesn’t need to know about your page structure. It calls a single /next/preview route with a secret query param and a slug search param that points to the document being previewed. // app/(payload)/next/preview/route.ts import { draftMode } from ' next/headers ' import { redirect } from ' next/navigation ' export async function GET ( request : Request ) { const { searchParams } = new URL ( request . url ) const secret = searchParams . get ( ' secret ' ) const slug = searchParams . get ( ' slug ' ) if ( secret !== process . env . PREVIEW_SECRET ) { return new Response ( ' Invalid token ' , { status : 401 }) } const draft = await draftMode () draft . enable () redirect ( slug ?? ' / ' ) } That’s the entire route. No collection-specific logic, no second-guessing which page type is involved. The redirect lands on the actual page, which reads draftMode().isEnabled and fetches accordingly. This is the pattern the Payload CMS preview documentation expects: a function that resolves to a string with additional URL parameters pointing to your app. The preview-URL builder lives in one shared lib Here’s where most implementations drift apart. The admin config, the preview route, and each page component all need to agree on how a preview URL is constructed. Store that logic in one place—a single getPreviewUrl utility imported everywhere—or you’ll be chasing 404s in production when someone renames a collection slug. // lib/getPreviewU

GitLab has introduced a new approach to Green DevOps, demonstrating how software engineering teams can measure the carbon emissions generated by their CI/CD pipelines. By Craig Risi

Building AI Agents That Don't Hallucinate: Structured Workflows, Guardrails, and Per-Step Evaluation How we replaced fragile prompt chains with typed schemas, validation gates, and evaluation at every step — 94% task success vs 60% baseline The Prompt Chain Trap January 2024. We built a "research agent" — 12 prompts chained together: Decompose question → 2. Search planning → 3. Execute searches → 4. Extract facts → 5. Synthesize → 6. Fact-check → 7. Format → ... It worked 60% of the time. The other 40%: Step 3 returned malformed JSON → Step 4 crashed Step 5 hallucinated citations → Step 6 missed it Step 7 output wrong format → Downstream consumer failed No visibility into which step failed Debugging meant reading 12 LLM calls' worth of logs. Adding a step broke three others. The Shift: Agents as Typed Workflows We moved from prompt chains to structured workflows with: Pydantic schemas for every step input/output Guardrails that validate and auto-retry Explicit state machine (not implicit chaining) Evaluation harness per step (not just end-to-end) ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │ Decompose │──▶│ Search │──▶│ Extract │──▶│ Synthesize │ │ Question │ │ Planning │ │ Facts │ │ Answer │ │ │ │ │ │ │ │ │ │ In: Query │ │ In: Plan │ │ In: Results │ │ In: Facts │ │ Out: SubQ[] │ │ Out: Steps │ │ Out: Fact[] │ │ Out: Answer │ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ │ │ │ │ ▼ ▼ ▼ ▼ [Schema] [Schema] [Schema] [Schema] [Guardrail] [Guardrail] [Guardrail] [Guardrail] [Eval: 0.9] [Eval: 0.85] [Eval: 0.9] [Eval: 0.95] Core Abstractions # agent_eval/schemas.py from pydantic import BaseModel , Field from typing import Literal , Any class DecomposeInput ( BaseModel ): user_query : str context : dict = Field ( default_factory = dict ) class DecomposeOutput ( BaseModel ): sub_questions : list [ str ] = Field ( min_length = 1 , max_length = 5 ) requires_tools : bool reasoning : str class PlanInput ( BaseModel ): sub_questions : list [

After spending countless hours designing, coding, debugging, and improving every little detail, I'm excited to share my personal portfolio with the developer community! 🌐 Live Website : https://didulagamage.pages.dev/ I'd Love Your Feedback ❤️ If you have a few minutes, I'd really appreciate it if you could visit my portfolio and share your thoughts. Your feedback helps me become a better developer. Thanks for reading! 🚀

Two posts ago I said I'd shipped code I never read. It's easy to write that as a general observation about the industry. It's less comfortable to go open your own repo and count. So I did. Here's what I found, and what I've actually done about it. The repo It's a desktop app I built for my own company. Roughly fourteen thousand lines of Python, a Tkinter UI, invoicing and documents and reports, the kind of internal tool nobody else will ever see. I had never written Python before I started it. I built it anyway, with a lot of AI help, learning as I went. That combination - no prior experience, heavy AI assistance, a real deadline because the business actually needed the thing - is basically a diff debt factory. I wasn't cutting corners on purpose. I just didn't have the knowledge to evaluate half of what I was merging, and it worked, so I moved on. What I actually found The clearest example took me months to notice. Five different parts of the app generate PDFs. Quotes, proformas, reports, petitions, heat treatment certificates. Every one of them was failing, in different ways, at different times, and I kept fixing them individually. Different error, different module, different patch. The actual cause was one thing: LibreOffice wasn't installed on the machine. Every one of those modules quietly depended on it to do the document conversion. Not one of them said so anywhere. I'd merged that dependency five separate times without ever registering that I'd taken it on. That's diff debt in its purest form. The code wasn't messy. It wasn't badly written. It just made an assumption I'd never read, and I paid interest on it five times over before I understood the principal. There were smaller ones too. A path handling bug that only showed up because my Windows username has a Turkish character in it - the code assumed ASCII and nobody, including me, had thought about it. Two Python versions installed side by side, quietly fighting. A stray quotation mark in my system PATH th

I spent a week using video editing Skills to build a video editing Agent. It feels amazing! It can automatically edit a 30-minute video in just 10 minutes. Video editing Agent demo: automatically editing a 30-minute video in 10 minutes. I often use CapCut to edit talking-head videos, but after using it for a long time, I found several problems. Problem 1: Smart talking-head editing does not understand meaning Because it cannot understand the meaning, it sometimes fails to identify repeated sections. If I speak continuously for 20 or 30 minutes, editing the video myself becomes exhausting. Problem 2: The subtitle quality is poor The automatically generated subtitles contain many incorrect words and typos. So I used the Skills feature in Claude Code to build a video editing Agent. The fundamental difference is simple: CapCut vs. Agent: a fixed tool vs. an adaptive assistant. The key difference is: CapCut = fixed tool + manual operation Agent = adaptive system + automatic learning I am not replacing CapCut with a better algorithm. I am replacing it with a system that can continuously improve itself. But that is not even the most impressive part. The most impressive part is this: the more I use it, the better it understands me, and the faster it becomes. Three Core Designs 1. Agent Logic It only takes four steps. Video editing Agent workflow: from the video file to the final video. 2. The Skills System At first, I put every function into one large Skill. I had to add instructions to distinguish between different tasks, which was very inconvenient. Now I have separated the five core video editing tasks into five independent Skills and placed them in the .claude/skills/ directory. This makes the structure clearer and the tasks easier to select. When I enter /v , Claude Code automatically lists the five available Skills. The list of five independent Skills. I select one, and the AI runs that Skill. Simple, right? A manual task that used to take 10 minutes now only requires