AI models

AppWizard
June 26, 2025
Android users will soon see enhancements to Gemini, Google's AI assistant, starting July 7th. Gemini will be able to manage device features and applications, allowing users to make phone calls, send WhatsApp messages, and manage utilities without enabling the Gemini Apps Activity setting. Users can disable this setting to prevent their interactions from being used for product development and AI model personalization. Google clarified that users maintain control over app connections and can disable them at any time. Gemini is set to replace Google Assistant on Android devices later this year. While turning off Apps Activity will prevent interactions from appearing in the activity log, Google will retain conversations for up to 72 hours for security purposes.
Winsage
June 25, 2025
Microsoft Corporation unveiled a small language model named Mu on June 23, designed for local operation on personal computers, particularly Copilot+ PCs. Mu is a 330 million parameter encoder-decoder language model optimized for deployment on Neural Processing Units (NPUs), allowing for quick responses with reduced power consumption and memory usage. It was pre-trained on a dataset of hundreds of billions of educational tokens and employs advanced fine-tuning techniques, including distillation and low-rank adaptation, along with transformer enhancements like Dual LayerNorm, Rotary Positional Embeddings, and Grouped-Query Attention. Users will access Mu's capabilities through the Windows operating system's Settings function, translating natural language inputs into system commands.
Winsage
June 24, 2025
Microsoft has introduced a small language model named Mu, designed to perform complex language tasks directly on devices like Copilot+ PCs, operating on a device's Neural Processing Unit. This model improves response times and reduces power and memory consumption compared to larger cloud-based AI models. Mu is developed using insights from Microsoft's Phi models and high-quality educational data, employing techniques such as distillation and low-rank adaptation for fine-tuning. Mu features an encoder–decoder architecture that separates input and output processes, enhancing speed and minimizing latency. It incorporates innovations like rotary positional embeddings, grouped-query attention, and dual LayerNorm. Microsoft also utilized quantization to reduce memory usage and improve speed without sacrificing accuracy, making Mu ideal for on-device applications. To enhance the AI agent in Windows Settings, Mu was fine-tuned with over 3.6 million examples, improving its ability to understand and manage system settings. The model addresses ambiguous commands and provides rapid assistance with precise multi-word commands, while maintaining a compact size that is one-tenth that of similar AI tools. Analysts on Wall Street have a positive outlook on MSFT stock, giving it a Strong Buy consensus rating based on 30 Buys and five Holds, with an average price target of 6.14 per share, indicating a potential upside of 6%.
Tech Optimizer
June 17, 2025
EnterpriseDB has introduced significant advancements to its EDB Postgres AI (EDB PG AI) platform, enabling secure and compliant deployment of AI solutions across Postgres environments. The platform integrates transactional, analytical, and AI workloads into a unified system, featuring automatic pipelines and built-in development tools for data automation. Key capabilities include low-code/no-code simplicity for rapid AI pipeline creation and comprehensive hybrid management for real-time insights across databases. Recent research indicates that only 13% of enterprises have successfully implemented agentic AI applications at scale, with early adopters achieving up to 227% higher ROI. Collaborations with Red Hat aim to provide organizations with solutions for successful AI outcomes. New features include enhanced data security, a purpose-built PG AI Analytics Engine for high-performance queries, and a universal data store for various data models. Performance advantages include six times better total cost of ownership (TCO) compared to SQL Server and up to 150 times faster NoSQL performance than MongoDB. A comparative analysis showed that EDB PG AI reduced overall complexity by 67% and maintenance costs by 38%.
Tech Optimizer
June 14, 2025
The integration of real-time data for AI applications is crucial, but traditional ETL processes can introduce latency and complexity. Open-source Postgres is a popular choice for developers, and Neon has developed an AI-driven approach to database creation. Databricks has launched Lakebase, a managed Postgres database, after acquiring Neon. Striim has expanded its Postgres offerings to facilitate high-throughput ingestion from Neon into Databricks for real-time analytics and enables rapid data delivery from legacy systems into Neon. This allows for seamless data flow for AI applications, including support for real-time ingestion, Change Data Capture (CDC), and event streaming from Apache Kafka. Striim's platform supports operational data replication from various traditional systems and upholds data governance with AI-driven PII detection. This expansion enhances partnerships with Databricks and supports real-time SQL Server data access.
Tech Optimizer
June 13, 2025
A new strain of Windows malware called "BrowserVenom" is exploiting interest in DeepSeek's AI models by targeting users through deceptive Google ads. These ads lead to a counterfeit website, "https[:]//deepseek-platform[.]com," where users are tricked into downloading a harmful file named “AILauncher1.21.exe.” This malware monitors and manipulates internet traffic, allowing attackers to intercept sensitive data. The operation is believed to involve Russian-speaking threat actors, and the malware has infected users in several countries, including Brazil, Cuba, Mexico, India, Nepal, South Africa, and Egypt. The fraudulent domain has been suspended, but the malware can evade many antivirus solutions. Users are advised to verify official domains when downloading software.
Tech Optimizer
June 12, 2025
The demand for real-time data in artificial intelligence is increasing, but integrating data from legacy systems poses challenges due to traditional ETL pipelines that can introduce latency. Open-source Postgres is favored for operational needs, and Neon offers a new approach to database creation using AI agents. Databricks has launched Lakebase, a managed Postgres database for AI applications, after acquiring Neon. Striim has expanded its Postgres offerings to enable high-throughput data ingestion from Neon into Databricks for real-time analytics and supports rapid data delivery from legacy systems into Neon. Striim's platform allows real-time replication of operational data from various traditional systems to Neon, real-time ingestion into Databricks, and enhances generative AI applications with inline data enrichment. Alok Pareek from Striim highlighted the importance of this expansion for Postgres-native teams to build real-time AI architectures. Striim also supports Databricks Delta Lake and SQL2Fabric-X for real-time SQL Server data access. Striim's platform processes over 100 billion daily events with sub-second latency, aiding proactive decision-making.
Winsage
June 11, 2025
Apple held its annual Worldwide Developers Conference (WWDC) in Cupertino, California, where it announced updates to its software ecosystem but did not introduce major AI announcements. The company plans to open its AI software to third-party developers, allowing access to on-device language models, which could lead to innovative software solutions and enhance user privacy. New features include live real-time translation in Phone, FaceTime, and Messages, improvements to the Visual Intelligence app, AI capabilities in the Reminders app, and a generative AI-powered fitness coach in the Apple Watch's Workout app. These updates aim to integrate AI into Apple's operating systems, positioning the company competitively in the AI landscape.
Search