Otter just launched three new AI agents, led by a voice-activated Meeting Agent that can verbally answer questions during calls. It taps into your company's historical meeting archive to retrieve answers, schedule follow-ups, and draft emails. The feature is currently rolling out on Zoom, with Teams and Google Meet support coming soon.
ARC Prize just unveiled ARC-AGI-2 and launched its 2025 competition to stress-test general fluid intelligence. Every test task was solved by multiple humans in two tries or less, yet pure LLMs score a clean 0% and public reasoning systems barely reach single digits. Why it matters: simply scaling up parameters won't close this reasoning gap; AGI needs fundamentally new architectures.
Anthropic just detailed the "think" tool, a technique giving Claude dedicated space to reason mid-execution during complex agentic workflows. Unlike extended thinking done upfront, it allows Claude to inspect tool outputs and navigate strict policies, driving a 54% relative improvement on τ-Bench airline tasks.
OpenAI just rolled out its compute-heavy o1-pro model to the API for select Tier 1–5 developers. It burns more compute to deliver sharper reasoning responses, while adding vision, function calling, and structured outputs. The catch is the eye-watering price: $150 per million input tokens and $600 per million output tokens.
Open-source tool Browser Use just blew up after users realized it powers the viral AI agent Manus. Co-creator Gregor Zunic says daily downloads jumped from around 5,000 on March 3 to 28,000 on March 10. By letting models click buttons and fill out forms, it is rapidly becoming the shovel of choice in the web agent gold rush.
Sakana AI just announced that a scientific paper fully generated by its upgraded The AI Scientist-v2 passed peer review at an ICLR workshop. Conducted with UBC and Oxford, it marks the first time a completely AI-authored paper cleared the exact same review process human researchers face. Full details of the v2 system will drop soon.
Microsoft just rolled out the Responses API and Computer-Using Agent (CUA) inside Azure AI Foundry. CUA lets AI autonomously navigate software interfaces, click buttons, and fill forms from natural language prompts, while the new API bundles tool calls into a single request. It turns AI from an advisor into an operator, though human oversight is still urged for sensitive workflows.
Harvey just launched agentic workflows in Assistant, designed to collaborate with legal professionals across transactional, litigation, and financial tasks. Instead of requiring detailed prompts, the agents proactively guide users and surface their thinking steps. In blind evaluations, Harvey matched lawyers in contract drafting and beat them in analyzing court transcripts. The feature rolls out broadly at the end of March.
Reka just open-sourced Flash 3, a 21B parameter reasoning model built completely from scratch. Despite its lightweight footprint, Reka claims competitive performance and is already using it as the core engine behind its new enterprise intelligence platform, Nexus.
Manus AI just partnered with Alibaba's Qwen team to integrate its viral general AI agent with Qwen's open-source models and domestic computing infrastructure. After facing massive hype and severe site crashes following its launch last week, Manus is leaning on Alibaba's resources to manage traffic surges and scale beyond its current invite-only rollout.
OpenAI just launched a dedicated suite of tools to help developers build AI agents, anchored by the brand-new Responses API alongside built-in web search, file search, and computer use. It also released an Agents SDK and tracing tools to orchestrate complex multi-step workflows. Why it matters: OpenAI is streamlining agentic apps into a single endpoint and plans to sunset the Assistants API by mid-2026.
Tencent Hunyuan just unveiled Hunyuan-TurboS, an ultra-large MoE model blending Transformer and Mamba architectures to ease long-context training and KV-cache overhead. The team claims it tops GPT-4o and DeepSeek-V3 in math and reasoning, all while cutting inference costs to one-seventh of its predecessor.
Highlight just launched a universal interface for MCP servers, aiming to bring Cursor-like interactions across every app on your computer. It features screen grounding that lets you mention services in-context and write with Tab, pitching itself as what Apple Intelligence and Copilot should have been.
ElevenLabs just launched its voice AI models on Google Cloud Marketplace. Enterprises can now easily pair ElevenLabs' lifelike speech generation with Google's fast Gemini 2.0 Flash model to build real-time conversational agents, localized dubbing, and marketing voiceovers within their existing cloud setup.
Apple just confirmed that its most anticipated Siri upgrades—the ones with personal context and cross-app actions—are delayed until "the coming year." Getting an AI assistant to dig through private emails and reliably execute real-world tasks proved harder than planned. Basically, that truly smart Siri likely won't arrive until iOS 19.
Convergence just introduced Template Hub, a centralized library packed with hundreds of ready-to-deploy agentic workflows created by its users. With one click, the Proxy agent gets straight to work, and creators can share setups and track run counts. It basically turns complex daily automations into shared templates anyone can run.
Google just expanded its medical AI system AMIE from one-off diagnoses to ongoing disease management across multi-visit consultations. Powered by a dual-agent Gemini setup grounded in clinical guidelines, AMIE matched or outperformed primary care doctors across 100 simulated patient cases. It marks a crucial step toward AI-assisted chronic care, though real-world clinical validation is still pending.
Butterfly Effect's new autonomous agent Manus just went viral, with invite codes fetching thousands of dollars after claims it outperforms OpenAI's tools. But beneath the hype and multi-model routing (including Claude and Qwen), early testers find it frequently crashes, hits endless loops, and fails basic tasks like ordering food or booking flights. The team says it's still in closed beta.
D-ID just partnered with Microsoft to bring its real-time AI avatars directly to Azure. Developers and enterprises can now integrate interactive digital humans into Microsoft Teams and other apps, offering lip-synced multi-language translation and sign language accessibility features.
OpenAI is reportedly preparing to launch specialized AI agents with monthly subscriptions ranging from $2,000 to $20,000 for tasks up to PhD-level research. The rollout features three tiers covering business, development, and research, with SoftBank committing $3 billion. OpenAI is betting enterprises will pay senior executive wages for automated expertise.
