AI Bangladesh

  • Home
  • AI Bangladesh

AI Bangladesh Trusted AI insights, news, and practical guides in Bangla.

Stop asking, start commanding! 🤖✨​Whether you're coding, writing, analyzing data, or brainstorming, here’s a massive che...
27/07/2026

Stop asking, start commanding! 🤖✨

​Whether you're coding, writing, analyzing data, or brainstorming, here’s a massive cheat sheet of 90 commands to get the absolute best results out of Claude.

​📌 Bookmark/Save this post for the next time you get stuck prompt-writing!

​What’s your go-to command when using AI? Let us know below! 👇

​

Weekly AI Roundup: Kimi K3 Drops, Inkling Releases, & Google Delays Gemini 3.5 Pro- Moonshot AI Drops Kimi K3: Massive 2...
17/07/2026

Weekly AI Roundup: Kimi K3 Drops, Inkling Releases, & Google Delays Gemini 3.5 Pro

- Moonshot AI Drops Kimi K3: Massive 2.8T-Parameter Open MoE Model Rivaling Claude Opus 4.8 and GPT-5.5, Full Weights Coming Soon
- Thinking Machines (Mira Murati's Lab) Releases Inkling: 975B Open-Weight Multimodal MoE with Controllable Thinking and Strong General Capabilities
- Google Delays Gemini 3.5 Pro: Months Behind on Coding Goals, Sending Alphabet Shares Lower as Rivals Ship Agentic Models
- OpenAI Pushes into Hardware: ChatGPT Home Speaker in Development + $230 Codex Keyboard for Power Users, Plus GPT-Red Internal Red-Teaming Model
- PrismML Launches Bonsai 27B: Extreme-Quantized Qwen-Based Model That Runs Natively on Phones (iPhone 17 Pro)
- DeepSeek Eyes Huge $74B Valuation Raise and IPO: China’s Low-Cost Leader Accelerates Amid Open-Weights Boom
- Anthropic Expands Free Claude for Teachers + Teen Safety Features Across Vendors (OpenAI, Meta)
- AWS Adds Grok 4.3 to Bedrock and Boosts Agent Tools as Enterprise Agentic Adoption Surges

Open-weights flooding the frontier, hardware bets intensifying, and big labs racing on agents and safety. The pace is relentless — what a week! 🚀

OpenAI's first hardware isn't what anyone expected — it's a $230 AI keyboard âŒ¨ī¸đŸ¤–OpenAI has launched Codex Micro, its fir...
16/07/2026

OpenAI's first hardware isn't what anyone expected — it's a $230 AI keyboard âŒ¨ī¸đŸ¤–

OpenAI has launched Codex Micro, its first branded hardware product—a $230 programmable mini keyboard built with Work Louder for developers using the Codex AI coding agent. Rather than replacing your keyboard, it's designed to control and monitor multiple AI agents from your desk.

The device features 13 mechanical keys, a joystick, a rotary dial, and RGB status lights that show whether AI agents are thinking, finished, waiting for approval, or have hit an error. Users can also trigger voice input, accept or reject code changes, and adjust an agent's reasoning level with dedicated controls.

This is a limited-run release aimed at developers, while OpenAI's much-anticipated screen-free AI companion device with Jony Ive remains under development. The launch underscores OpenAI's growing focus on AI agents that work in parallel rather than traditional chat interfaces.

At first glance, a $230 AI keyboard sounds underwhelming. But the bigger story isn't the hardware—it's the workflow. OpenAI is betting that developers will soon be managing multiple AI agents simultaneously, and that those agents deserve physical controls just like music producers use mixers or gamers use macro pads.

If that vision catches on, the future of AI may involve dedicated hardware for orchestrating AI work, not just typing prompts into a chat window.

Sources:

The collaboration with Work Louder is separate from OpenAI’s main hardware project, rumored to be some kind of smart speaker.

āĻāĻ• āϜāύ⧇āϰ āĻŦ⧇āϤāύ āĻšāϞ⧋ ā§§ā§Ē.ā§Š āĻŦāĻŋāϞāĻŋāϝāĻŧāύ āĻĄāϞāĻžāϰ, āφāϰ āϤāĻžāϰ āĻŦ⧇āϤāύ⧇āϰ āϜāĻ¨ā§āϝ āϚāĻžāĻ•āϰāĻŋ āĻšāĻžāϰāĻžāϤ⧇ āĻšāϞ ā§Ž,ā§Ļā§Ļā§Ļ āĻŽāĻžāύ⧁āώāϕ⧇!āφāĻ°ā§āϟāĻŋāĻĢāĻŋāĻļāĻŋ⧟āĻžāϞ āχāĻ¨ā§āĻŸā§‡āϞāĻŋāĻœā§‡āĻ¨ā§āϏ āĻŦāĻž AI-āĻāϰ ...
16/07/2026

āĻāĻ• āϜāύ⧇āϰ āĻŦ⧇āϤāύ āĻšāϞ⧋ ā§§ā§Ē.ā§Š āĻŦāĻŋāϞāĻŋāϝāĻŧāύ āĻĄāϞāĻžāϰ, āφāϰ āϤāĻžāϰ āĻŦ⧇āϤāύ⧇āϰ āϜāĻ¨ā§āϝ āϚāĻžāĻ•āϰāĻŋ āĻšāĻžāϰāĻžāϤ⧇ āĻšāϞ ā§Ž,ā§Ļā§Ļā§Ļ āĻŽāĻžāύ⧁āώāϕ⧇!

āφāĻ°ā§āϟāĻŋāĻĢāĻŋāĻļāĻŋ⧟āĻžāϞ āχāĻ¨ā§āĻŸā§‡āϞāĻŋāĻœā§‡āĻ¨ā§āϏ āĻŦāĻž AI-āĻāϰ āĻāχ āϰ⧇āϏ⧇ āϟāĻŋāϕ⧇ āĻĨāĻžāĻ•āĻžāϰ āϜāĻ¨ā§āϝ āĻŦ⧜ āĻŦ⧜ āĻŸā§‡āĻ• āϕ⧋āĻŽā§āĻĒāĻžāύāĻŋāϗ⧁āϞ⧋ āϕ⧀ āĻĒāϰāĻŋāĻŽāĻžāĻŖ āĻŽāϰāĻŋ⧟āĻž, āϤāĻžāϰ āĻāĻ•āϟāĻž āĻŦ⧜ āĻĒā§āϰāĻŽāĻžāĻŖ āĻĻāĻŋāϞ āĻŽā§‡āϟāĻž (Meta)āĨ¤ āĻ•āĻŋāĻ¨ā§āϤ⧁ āĻāϰ āĻĒ⧇āĻ›āύ⧇āϰ āĻ—āĻ˛ā§āĻĒāϟāĻž āĻŦ⧇āĻļ āύāĻŋāĻ°ā§āĻŽāĻŽāĨ¤

āφāϞ⧇āĻ•āϜāĻžāĻ¨ā§āĻĄāĻžāϰ āĻ“ā§ŸāĻžāĻ‚ (Alexander Wang) MIT-āϤ⧇ āĻŽāĻžāĻ¤ā§āϰ āĻāĻ• āĻŦāĻ›āϰ āĻĒ⧜āĻžāϰ āĻĒāϰ āĻĄā§āϰāĻĒāφāωāϟ āĻšāύ āĨ¤ ⧍ā§Ļā§§ā§Ŧ āϏāĻžāϞ⧇ āĻĒā§āϰāϤāĻŋāĻˇā§āĻ āĻž āĻ•āϰ⧇āύ 'Scale AI' āĻāĻŦāĻ‚ āĻŽāĻžāĻ¤ā§āϰ ⧍ā§Ē āĻŦāĻ›āϰ āĻŦ⧟āϏ⧇āχ āĻšā§Ÿā§‡ āϝāĻžāύ āĻŦāĻŋāĻļā§āĻŦ⧇āϰ āϏāĻ°ā§āĻŦāĻ•āύāĻŋāĻˇā§āĻ  āϏ⧇āϞāĻĢ-āĻŽā§‡āĻĄ āĻŦāĻŋāϞāĻŋ⧟āĻ¨ā§‡ā§ŸāĻžāϰāĨ¤

āĻāĻĻāĻŋāϕ⧇ āĻŽā§‡āϟāĻž-āĻĒā§āϰāϧāĻžāύ āĻŽāĻžāĻ°ā§āĻ• āϜāĻžāĻ•āĻžāϰāĻŦāĻžāĻ°ā§āĻ— āĻĻ⧇āĻ–āĻ›āĻŋāϞ⧇āύ āϝ⧇, OpenAI āĻāĻŦāĻ‚ Google-āĻāϰ āϤ⧁āϞāύāĻžā§Ÿ āĻŽā§‡āϟāĻžāϰ āύāĻŋāϜāĻ¸ā§āĻŦ AI āĻŽāĻĄā§‡āϞāϗ⧁āϞ⧋ āĻŦ⧇āĻļ āĻĒāĻŋāĻ›āĻŋā§Ÿā§‡ āĻĒ⧜āϛ⧇āĨ¤ āϤāĻžāχ āĻ—āϤ āĻŦāĻ›āϰ (⧍ā§Ļ⧍ā§Ģ āϏāĻžāϞ⧇āϰ āϜ⧁āύ⧇) āϜāĻžāĻ•āĻžāϰāĻŦāĻžāĻ°ā§āĻ— āĻāĻ• āĻ…āĻ­āĻžāĻŦāĻ¨ā§€ā§Ÿ āϚāĻžāϞ āϚāĻžāϞāϞ⧇āύāĨ¤ āφāϞ⧇āĻ•āϜāĻžāĻ¨ā§āĻĄāĻžāϰ āĻ“ā§ŸāĻžāĻ‚āϕ⧇ āĻŽā§‡āϟāĻžāϰ āϏ⧁āĻĒāĻžāϰ-āχāĻ¨ā§āĻŸā§‡āϞāĻŋāĻœā§‡āĻ¨ā§āϏ āĻ˛ā§āϝāĻžāĻŦ⧇āϰ āĻĒā§āϰāϧāĻžāύ āĻšāĻŋāϏ⧇āĻŦ⧇ āύāĻŋā§Ÿā§‡ āφāϏāĻžāϰ āϜāĻ¨ā§āϝ āϤāĻŋāύāĻŋ āĻ–āϰāϚ āĻ•āϰāϞ⧇āύ āϰ⧇āĻ•āĻ°ā§āĻĄ ā§§ā§Ē.ā§Š āĻŦāĻŋāϞāĻŋāϝāĻŧāύ āĻĄāϞāĻžāϰ!

āĻ•āĻŋāĻ¨ā§āϤ⧁ āĻāχ āĻŦāĻŋāĻĒ⧁āϞ āϟāĻžāĻ•āĻž āϕ⧋āĻĨāĻž āĻĨ⧇āϕ⧇ āφāϏāϛ⧇?
āĻŽā§‡āϟāĻž āĻāχ āĻŦāĻ›āϰ AI-āĻāϰ āĻĒ⧇āĻ›āύ⧇ āĻĒā§āϰāĻžā§Ÿ ā§§ā§Ēā§Ģ āĻŦāĻŋāϞāĻŋāϝāĻŧāύ āĻĄāϞāĻžāϰ āĻ–āϰāϚ āĻ•āϰāĻžāϰ āĻĒāϰāĻŋāĻ•āĻ˛ā§āĻĒāύāĻž āĻ•āϰ⧇āϛ⧇, āϝāĻž āφāϗ⧇āϰ āĻŦāĻ›āϰ⧇āϰ āϤ⧁āϞāύāĻžā§Ÿ āĻĒā§āϰāĻžā§Ÿ āĻĻā§āĻŦāĻŋāϗ⧁āĻŖāĨ¤ āφāϰ āĻāχ āĻŦāĻŋāĻļāĻžāϞ āĻ–āϰāĻšā§‡āϰ āĻœā§‹āĻ—āĻžāύ āĻĻāĻŋāϤ⧇ āϜāĻžāĻ•āĻžāϰāĻŦāĻžāĻ°ā§āĻ— āĻŽā§‡āϟāĻžāϰ āĻĒā§āϰāĻžā§Ÿ ā§Ž,ā§Ļā§Ļā§Ļ āĻ•āĻ°ā§āĻŽā§€āϕ⧇ āĻ›āĻžāρāϟāĻžāχ āĻ•āϰ⧇āϛ⧇āύ āϝāĻž āϤāĻžāĻĻ⧇āϰ āĻŽā§‹āϟ āĻ“ā§ŸāĻžāĻ°ā§āĻ•āĻĢā§‹āĻ°ā§āϏ⧇āϰ āĻĒā§āϰāĻžā§Ÿ ā§§ā§Ļ%! āϜāĻžāĻ•āĻžāϰāĻŦāĻžāĻ°ā§āĻ— āύāĻŋāĻœā§‡āχ āĻ¸ā§āĻŦā§€āĻ•āĻžāϰ āĻ•āϰ⧇āϛ⧇āύ āϝ⧇, AI-āϤ⧇ āĻŦ⧜ āĻŦāĻŋāύāĻŋā§Ÿā§‹āϗ⧇āϰ āĻĢāĻžāĻ¨ā§āĻĄ āĻœā§‹āĻ—āĻžā§œ āĻ•āϰāϤ⧇āχ āĻāχ āĻ›āĻžāρāϟāĻžāχāĨ¤
āϏāĻšāϜ āĻ­āĻžāώāĻžā§Ÿ āĻšāĻŋāϏāĻžāĻŦāϟāĻž āĻĻāĻžāρ⧜āĻžā§Ÿ āĻāχāϰāĻ•āĻŽ:
ā§§ āϜāύ āĻŽāĻžāύ⧁āώāϕ⧇ āϕ⧋āĻŽā§āĻĒāĻžāύāĻŋāϤ⧇ āφāύāĻžāϰ āĻĒ⧇āĻ›āύ⧇ āĻ–āϰāϚ āĻšāϞ⧋ ā§§ā§Ē.ā§Š āĻŦāĻŋāϞāĻŋāϝāĻŧāύ āĻĄāϞāĻžāϰ āφāϰ āϏ⧇āχ āĻ–āϰāĻšā§‡āϰ āĻŽāĻžāĻļ⧁āϞ āĻĻāĻŋāϤ⧇ ā§Ž,ā§Ļā§Ļā§Ļ āϏāĻžāϧāĻžāϰāĻŖ āĻŽāĻžāύ⧁āώ āϤāĻžāĻĻ⧇āϰ āϚāĻžāĻ•āϰāĻŋ āĻšāĻžāϰāĻžāϞ⧇āύāĨ¤

āĻ­āĻŦāĻŋāĻˇā§āϝāϤ⧇āϰ āĻĒā§āϰāϝ⧁āĻ•ā§āϤāĻŋāϰ āϞ⧜āĻžāĻ‡ā§Ÿā§‡ āϟāĻŋāϕ⧇ āĻĨāĻžāĻ•āϤ⧇ āϕ⧋āĻŽā§āĻĒāĻžāύāĻŋāϗ⧁āϞ⧋āϰ āĻāχ āύ⧀āϤāĻŋāϕ⧇ āφāĻĒāύāĻŋ āϕ⧀āĻ­āĻžāĻŦ⧇ āĻĻ⧇āĻ–āϛ⧇āύ? āĻāϟāĻž āĻ•āĻŋ āĻŦā§āϝāĻŦāϏāĻžāϰ āĻ–āĻžāϤāĻŋāϰ⧇ āĻĒā§āĻ°ā§Ÿā§‹āϜāĻ¨ā§€ā§Ÿ āĻĒāĻĻāĻ•ā§āώ⧇āĻĒ, āύāĻžāĻ•āĻŋ āĻ•āĻ°ā§āĻĒā§‹āϰ⧇āϟ āĻĻ⧁āύāĻŋ⧟āĻžāϰ āϚāϰāĻŽ āύāĻŋāĻˇā§āϠ⧁āϰāϤāĻž?

05/03/2026

GPT-5.3 Instant rolls out — smoother, smarter, less cringe ChatGPT is here 🤖đŸ’Ŧ

OpenAI just released GPT-5.3 Instant as the new default model in ChatGPT — designed to fix tone issues, reduce unnecessary refusals, and give more natural, conversational answers.

The update significantly cuts hallucinations (e.g., ~26.8 % fewer with web search) and blends web results with internal reasoning for more accurate, contextualized replies.

GPT-5.3 Instant trims preachy disclaimers and defensive phrasing — based on user feedback that previous versions sometimes felt overly cautious or robotic.

This isn’t a flashy new capability update — it’s the one that feels better every day: smoother flow, clearer answers, stronger web synthesis, and a more human-like chat vibe.
If OpenAI can make interactions feel less awkward and more helpful without sacrificing accuracy, this rollout could reset expectations for everyday AI assistants across the ecosystem.

05/03/2026

Apple finally drops the $599 MacBook Neo — a budget Mac that could shake up the laptop market đŸ’ģđŸ”Ĩ

Apple just unveiled the MacBook Neo, its most affordable MacBook ever with a $599 starting price (and $499 for education buyers), now available for pre-order and shipping March 11.

It runs on the A18 Pro chip (from the iPhone 16 Pro line), offers up to 16 hours battery life, a 13â€ŗ Liquid Retina display, spatial audio speakers, 1080p we**am, and solid everyday performance.

Some premium features are trimmed to hit the price — no Thunderbolt, limited ports, and 8 GB unified RAM — but Apple says it’s 50 % faster than top Intel PCs on basic tasks and 3× faster on on-device AI work.

This could be a huge disruptor — Apple finally betting big on the budget laptop segment to take on Chromebooks and entry-level Windows machines without ditching the “Mac experience.”

If people embrace it, the MacBook Neo might open the Apple ecosystem to students and first-time buyers in a way priced Macs never have before — but whether the compromises (memory, ports) are worth it will likely spark lively debate.

05/03/2026

OpenAI amends its Pentagon AI deal after major backlash over surveillance and ethics đŸ˜ŦđŸ‡ē🇸

CEO **Sam Altman admitted the original Pentagon agreement “looked opportunistic and sloppy,” and OpenAI is now revising the contract to add clearer safeguards against misuse.

The updated wording will explicitly ban intentional domestic surveillance of U.S. persons and nationals, and bar intelligence agencies like the NSA from using its AI without further approvals.

This comes after intense public criticism — subscription cancellations, boycott campaigns like QuitGPT, and rival AI Claude surging in downloads as users protested the initial military deal.

OpenAI’s quick course-correction shows how sensitive the AI-military intersection has become — even a major defense contract can trigger consumer backlash and force a rethink of how AI is deployed in national security.

If the amended deal sticks and the safeguards hold up, it might set a precedent for future AI-government partnerships — but critics will be watching closely for loopholes and enforcement.

01/03/2026

India blocks Supabase access — developer apps go silent in a key market đŸ‡ŽđŸ‡ŗđŸšĢ

- India’s ISPs — including Jio, Airtel and ACT Fibernet — have been instructed to block DNS resolution for *.supabase.co domains, making Supabase backend services unreachable for many Indian users and apps.

- Supabase says its infrastructure is fully operational globally, but projects in the India region can’t connect, breaking auth, API calls, realtime updates and storage for affected developers.

- The restriction appears linked to a government blocking order under Section 69A of India’s IT Act — though officials haven’t publicly explained the reason yet.

This has hit developers hard — dozens of Indian startups, hobby apps, student projects and production services using Supabase suddenly find their core infrastructure inaccessible without workarounds like DNS changes or VPNs.

If this block remains in place, it could push developers toward alternative backends or force architecture changes just to stay online in one of the fastest-growing app markets in the world.

Address


Alerts

Be the first to know and let us send you an email when AI Bangladesh posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Shortcuts

  • Want your business to be the top-listed Engineering Company?

Share