DEV Community: Anwar Laksir

[Boost]

Anwar Laksir — Sun, 09 Nov 2025 16:11:04 +0000

Introducing Beitha: The first AI browser automation extension

Anwar Laksir ・ Nov 9

Introducing Beitha: The first AI browser automation extension

Anwar Laksir — Sun, 09 Nov 2025 16:09:01 +0000

Say Bye Bye to Atlas from OpenAI and Comet from Perplexity: Introducing Beitha

Beitha brings AI directly into your browser to help you focus, organize, and achieve more every day. From managing tabs and drafting emails to planning your week, it’s your personal assistant that turns chaos into clarity instantly and securely.

Check it out: Here
Check it our on Chrome web store: Here

You Can Now Save Prompts On Cubent

Anwar Laksir — Mon, 01 Sep 2025 02:11:25 +0000

If you’ve ever worked with AI, you already know one truth: prompts matter. The right prompt can feel like a superpower — unlocking sharper answers, faster workflows, and more creative outcomes. But here’s the problem: finding that perfect prompt often takes trial, error, and plenty of tweaking. And once you finally get it right, it’s way too easy to lose track of it.

That’s why we’re excited to introduce our new feature: Save Prompts ✨

Why Prompts Deserve a Home

Prompts aren’t just throwaway instructions. For many users, they’re the secret sauce behind consistent results. Whether you’re writing code, generating content, brainstorming ideas, or researching — prompts are reusable assets. But until now, there hasn’t been an easy way to manage them.

What Save Prompts Lets You Do

With Save Prompts, you can:

Store your best prompts so you never lose them.

Organize them in a way that fits your workflow.

Reuse instantly without rewriting or digging through old chats.

Experiment with variations while keeping your original safe.

It’s like building your own personal prompt library, tailored to your exact needs.

Who It’s For

Developers fine-tuning coding queries.

Writers building repeatable content structures.

Teams standardizing communication or research workflows.

Anyone who’s tired of reinventing the wheel every time they start a new project.

Why It Matters

AI isn’t about one-off magic tricks — it’s about consistent performance. By giving prompts a permanent place in your toolkit, you’ll save time, reduce frustration, and make your interactions with AI more powerful.

Ready to Try It?

Next time you land on the perfect prompt, don’t let it slip away. Hit Save. Build your collection. And unlock the value of your own creativity on demand.

Benchmarking GPT-5 vs Claude 4 Sonnet on 200 Requests

Anwar Laksir — Tue, 26 Aug 2025 14:50:32 +0000

Authors & Volunteers: Dr. Mattia Salvarani (UNIMORE), Prof. Carlos Hernández (University of Cambridge), Dr. Aisha Rahman (University of Toronto), Prof. Luca Moretti (ETH Zürich)

Github Repo: Click Here

Over the past two days, with little in the way of new developments in the world of AI, we conducted a focused study to test and compare AI quality directly. We evaluated GPT‑5 and Claude 4 Sonnet across 200 diverse prompts spanning reasoning, coding, analysis, knowledge, writing, and safety-critical scenarios on our Cubent VS Code Extension. Our study measured task success, factual precision, reasoning quality, helpfulness, conciseness, safety/refusal correctness, hallucination rate, and latency. In accordance with real-world user experience constraints, we also reported p50/p90/p95 latency and cost-normalized quality.

Key findings:

Speed: Claude 4 Sonnet is consistently faster (median 5.1 s) than GPT‑5 (median 6.4 s); Sonnet also shows lower p95 latency.

Precision: On fact-heavy tasks, Sonnet is slightly more precise (93.2% vs. 91.4% factual precision) and exhibits a lower hallucination rate.

Overall Quality: GPT‑5 achieves higher task success overall (86% vs. 84%), particularly on multi-step reasoning and code generation/debugging.

Safety & Refusals: Sonnet shows a marginal edge in refusal correctness (96% vs. 94%), while both models maintain high safety compliance.

Domain Trends: Sonnet is faster and a touch more precise on editing, summarization, and short-form Q&A; GPT‑5 leads on complex reasoning, code synthesis, data analysis, and multilingual tasks.

All results include bootstrapped 95% confidence intervals and effect sizes. We release the prompt taxonomy, rubric, and annotation protocol for reproducibility.

Introducing Cubent - The Smartest AI Coding Agent

Anwar Laksir — Tue, 29 Jul 2025 01:50:02 +0000

We’re excited to introduce Cubent, an AI-powered development assistant built to make coding faster, easier, and more productive. Unlike traditional autocomplete tools or bloated platforms, Cubent is lightweight, agentic, and deeply integrated into the workflows of real developers using VS Code.

Why We Built Cubent

Most AI tools today fall into two extremes: either too basic to be useful or too expensive and opaque to trust. We built Cubent because we wanted something better — a tool that doesn’t just write code, but helps you think, debug, refactor, and build faster, with more control and transparency.

What Cubent Does

Cubent goes far beyond basic code suggestions. It can explain functions, fix bugs, generate unit tests, write new components, and reason across multiple files. It adapts to your context and lets you speak or type natural commands. It’s designed to be your real-time coding partner, not just an autocomplete engine.

Designed Around Developer Control

With Cubent, you’re always in control. You can bring your own API key (BYAK), or use our managed access to top models like GPT-4, Claude, and Gemini. We don’t track your code, and we don’t hide usage. You decide which models you use and when. The result is full transparency and much lower costs.

Agentic Workflows, Custom Commands

One of the things we’re most excited about is Cubent’s support for agent-like behaviors and custom tasks. You can ask it to scaffold an entire API, explain a legacy codebase, or walk through a debugging process step by step. You can also create your own commands tailored to your stack and style.

Built for Practical Productivity

We’re not interested in hype. We’re building for everyday developers who want something practical: fast context awareness, natural interaction, minimal friction. Cubent is meant to feel like a natural extension of your coding environment — not a tool you need to “learn” or fight against.

Pricing That Respects Developers

Most AI assistants cost $20–$30/month with vague usage rules. Cubent is $5/month. That’s it. You can try it for free for 7 days, no credit card required. We keep things simple, because developers deserve predictable pricing and freedom to choose how they work.

What’s Coming Next

This is just the beginning. We’re actively working on even more control over what context Cubent sees, better model switching, voice-to-text improvements, local LLM support, and new features for small teams. We’re moving fast and always listening to feedback.

Try Cubent Today

Cubent is live and available now on vscode. No waitlist. No complicated onboarding. Just install the extension, connect your API key, and start building with real AI support.