AI News

Follow the latest AI and automation news

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Signit raises $15 million Series A led by Raed Ventures

Signit, a Saudi startup, has raised $15 million in a Series A funding round to expand its AI-powered contract management platform. The round was led by Raed Ventures, with participation from STV, Seedra Ventures, Takamol Ventures, and Suhail Ventures. Founded in 2021 by Mohamed El Abbouri, Signit offers digital signatures and contract lifecycle management solutions for government and enterprise clients. The funding will enhance AI capabilities, contract automation, and secure digital trust infrastructure.

Business

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Sufra AI secures $100K from Snoonu as Startup Factory launches

Sufra AI, a Qatar-based F&B tech startup, has secured a $100,000 pre-seed investment from Snoonu, marking the first deal under its Startup Factory initiative. Founded in 2025 by Ekaterina Demenkova and Jemal Velihanova, Sufra AI develops AI-powered smart menu systems for restaurants. The investment is part of Snoonu's new venture-building program, which aims to support early-stage founders with capital, mentorship, and operational support. This initiative aligns with Qatar National Vision 2030 and positions Snoonu as an active venture builder in the region.

Arab AI Scene

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Arabic.AI partners with HeyBreez to scale Arabic voice AI for enterprises

Arabic.AI, based in the UAE, has announced a strategic partnership with HeyBreez to deliver production-grade Arabic voice AI solutions for enterprises and governments. This collaboration combines Arabic language intelligence with real-time infrastructure for scalable voice AI deployment. Arabic.AI provides advanced NLP models covering Modern Standard Arabic and regional dialects, while HeyBreez contributes low-latency infrastructure supporting streaming and multi-agent orchestration. This partnership aims to address the needs of over 400 million Arabic speakers.

💡

INVIA raises $1.2 million to build AI-powered financial OS for SMEs in Egypt

What impact will INVIA have on the SME market in Egypt?

Models

DeepSeek Launches V4 API with 1 Million Token Context

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Models

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

DeepSeek Launches V4 API with 1 Million Token Context

On April 23, DeepSeek launched its new V4 API model lineup, featuring DeepSeek V4-Pro and DeepSeek V4-Flash. Both versions support a native context window of up to 1 million tokens, with V4-Flash optimized for low latency and cost efficiency in real-time applications. V4-Pro utilizes a mixture-of-experts (MoE) architecture for tasks like coding, mathematics, and complex reasoning. The API facilitates long-document processing, tool integration, and multi-step workflows, ensuring compatibility with major development frameworks. Legacy APIs will be discontinued on July 24, 2026.

Agents

XYZ Embodied AI Launches BotPack B Series Edge Computing Backpack

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Agents

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

XYZ Embodied AI Launches BotPack B Series Edge Computing Backpack

On April 23, XYZ Embodied AI unveiled the BotPack B Series backpack at the Hannover Messe in Germany. The backpack is designed to provide edge computing capabilities for robots, including quadrupeds and humanoids. The series features two models, B5 and B4, both powered by an NVIDIA chip. These systems enable robots to run AI models locally, reducing reliance on cloud computing and minimizing response times. The devices support high-speed data transfer via 10G Ethernet, 5G, and Wi-Fi 7, and include GPS modules for navigation. The backpack weighs under 2.5 kg and is compatible with platforms like Unitree Robotics' Unitree G1.

$DeepSeek V4 - almost on the frontier, a fraction of the price$

Models

DeepSeek V4 - almost on the frontier, a fraction of the price

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

$DeepSeek V4 - almost on the frontier, a fraction of the price$

Models

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

DeepSeek V4 - almost on the frontier, a fraction of the price

Chinese AI lab DeepSeek has released its new V4 models, including DeepSeek-V4-Pro and DeepSeek-V4-Flash. The Pro model features 1.6 trillion parameters with 49 billion active, while Flash has 284 billion parameters with 13 billion active. DeepSeek-V4-Pro is now the largest open weights model, surpassing Kimi K2.6 and GLM-5.1, and is more than double the size of DeepSeek V3.2. Both models are available under the MIT license, making them accessible for developers.

Business

Deep Seek Launches AI Model 'V4' in Collaboration with Huawei

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Business

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Deep Seek Launches AI Model 'V4' in Collaboration with Huawei

Chinese startup Deep Seek has launched its new AI model "V4," designed to operate efficiently on Huawei processors, reinforcing Beijing's ambitions for technological self-sufficiency. The professional version of "V4" outperforms all open-source models, surpassed only by Google's Gemini Pro 3.1, showcasing Deep Seek's advancements in the field. The launch of "V4" follows close collaboration with Huawei, which announced full support for its entire high-performance computing product line for Deep Seek's models.

Agents

An update on recent Claude Code quality reports

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Agents

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

An update on recent Claude Code quality reports

Recent reports indicate that Claude Code experienced a decline in result quality over the past two months due to real issues. The problems stemmed from three separate issues in the Claude Code harness, which directly impacted user experience. This highlights the complexity of building AI systems and the need for developers to address such bugs effectively. Understanding these challenges is crucial for improving AI performance and ensuring a better user experience in future iterations.

Business

Serving the For You feed

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Business

24 Apr 20267 Dhuʻl-Qiʻdah 1447 AH

Serving the For You feed

One of the key features of PlusSky is the ability for users to run their own customized feeds. Currently, around 72,000 users utilize the "For You Feed," managed by the user spacecowboy. The feed operates on a single Go server using SQLite on a high-performance PC, which has 16 cores, 96 GB of RAM, and 4 TB of storage. The system relies on user preferences to deliver personalized recommendations. Operating costs for the system are approximately $30 per month, with the capacity to handle one million daily active users if the most efficient algorithm is employed.