Morning Brief

1. Apple confirms multiple new product launches in March
2. Perplexity introduces Perplexity Computer
3. Google launches Nano Banana 2
4. LM Studio releases remote connection solution LM Link
5. Flow receives a revamped upgrade
6. Minimax unveils MaxClaw

---

Apple confirms multiple new product launches in March

On February 26, Apple CEO Tim Cook confirmed via social media that Apple will begin releasing multiple new products starting next week. According to recent media speculation, Apple’s upcoming lineup may include an entry-level MacBook, iPad, Mac mini, and Studio Display. Source

---

Perplexity introduces Perplexity Computer

On February 26, AI search platform Perplexity officially unveiled Perplexity Computer — a new browser-based chat interface capable of integrating multiple AI models with autonomous execution abilities to automatically complete end-to-end workflows.

During use, users only need to describe their objective, and the system will automatically generate sub-agents to handle tasks such as web search, document generation, data processing, or API calls.

The system is powered by Opus 4.6 as its core model and works in coordination with Gemini, Grok, ChatGPT 5.2, the image generation model Nano Banana, and the video processing model Veo 3.1. Each task runs in an isolated and secure environment, equipped with a dedicated browser, file system, and tool interfaces.

Perplexity Computer is currently available only to Max subscribers and will be opened to Pro users at a later date. Source

---

Google launches Nano Banana 2

On February 27, Google announced the release of its next-generation image generation model, Nano Banana 2. The model retains its high-quality image generation capabilities, delivers improved text rendering performance, and comes at a lower generation cost. Nano Banana 2 is now available and will be used by default when enabling image generation within Gemini. Source

---

LM Studio releases remote connection solution LM Link

On February 26, LM Studio announced the launch of a new feature, LM Link, which allows users to securely connect to remote LM Studio instances and invoke models across devices. According to LM Studio, LM Link uses end-to-end encryption, supports loading local models, and enables continued usage in on-the-go or mobile scenarios. Supported remote endpoints for this connection include local devices, dedicated large-model hosts, and cloud-based virtual machines.

LM Studio also stated that this feature is released in collaboration with Tailscale, with LM Link relying on the latter’s networking capabilities to enable remote access and device interconnection. Source

---

Flow receives a revamped upgrade

On February 26, Google announced a major update to its AI creation tool Flow, expanding it from a video-focused generator into a “full AI creative studio,” designed to support story drafting, visualization, and iterative revision within a single workflow, while also optimizing the overall creative process.

Flow further strengthens its “image-and-video together” workflow approach by supporting grouped organization and editing of assets. Its editing method has also become more natural-language-driven, with a newly added lasso tool that allows users to select image regions and apply localized edits using text instructions. Direct annotation on images is also supported to assist with modifications.

In terms of video editing capabilities, Flow introduces more continuous editing and fine-tuning controls, including extending clip duration, adding or removing objects in videos, generating “what happens next” content, and controlling camera movement through panning and zooming. These features bring generation and editing more tightly into a unified workflow. Source

---

Minimax unveils MaxClaw

On February 26, the Minimax team introduced functional upgrades to MiniMax Agent Expert and unveiled a new AI assistant, MaxClaw. In Expert 2.0, Minimax further streamlines the creation of expert Agents. Users no longer need to configure Skills, SubAgents, MCPs, or prompt structures — simply describing task objectives or capability requirements in natural language allows the Agent to automatically handle SOP planning, tool orchestration, and capability configuration.

Alongside the Expert 2.0 upgrade, Minimax will roll out MaxClaw, a cloud-based AI assistant built on OpenClaw and directly integrated into the MiniMax Agent web interface. It enables users to deploy and run OpenClaw in the cloud without requiring their own servers or API keys.

In terms of details, MaxClaw provides systematic upgrades to OpenClaw’s existing Skills such as image understanding, video understanding, web extraction, and search, while also adding built-in tools for image generation, video generation, image search, and web deployment. All built-in tools require no third-party API integration and do not incur additional API costs. Source