LHC

This weekend we hosted the first ElevenCreative Sessions in Paris. We brought together hundreds of creators for a full day of demos, community talks, and a live creative challenge.

Fine-tuning your own musical signature, opening video generation to all developers, rotating a 2D object into 3D in Photoshop, or building an entire campaign by chatting with an AI agent. On the ecosystem front, Anthropic puts $100 million on the table to structure its partner network, Publicis acquires AdgeAI to predict creative performance, and Runway publishes a clear-eyed post on the risks of giving a face to an AI. In London, Google DeepMind names its new HQ after a legendary Go move. Full roundup below, Bonne heure ▼

🔹 Anthropic invests $100 million in the Claude Partner Network Anthropic launches the Claude Partner Network, a program for organizations helping enterprises adopt Claude. The initial $100 million commitment for 2026 will fund training, dedicated technical support, co-marketing, and joint business development. The partner-facing team will grow fivefold, with Applied AI engineers and technical architects available for client projects. The program includes a first technical certification (Claude Certified Architect, Foundations), a portal with access to Anthropic's internal resources, a directory of certified partners, and a starter kit for legacy code modernization. Claude remains the only frontier model available on all three major cloud providers: AWS, Google Cloud, and Microsoft. Among partners already engaged, Accenture is training 30,000 professionals on Claude, Cognizant is deploying the tool across its 350,000 associates, and Infosys has established a dedicated Anthropic center of excellence. Deloitte is also among the network's early supporters. Membership is free and open starting today. anthropic.com/news/claude-pa…

Paris, see you in 48h!

Morning: Coffee and Create (10:00 AM - 1:00 PM CET) Featuring a coffee bar, live demos from our team, short talks from local creators, open networking, free credits, and exclusive merch. Community partners: Voice AI Space, @StudioLHC and @nebsh83

We’re launching ElevenCreative Sessions, an in-person creative playground for AI creators. Our first session will be in Paris this Saturday, the 14th of March. Walk in. Create something. Leave inspired.

🔹 Claude integrates more deeply into Excel and PowerPoint with shared context across files Anthropic announced a significant update to its Claude add-ins for Excel and PowerPoint. The main change: both applications now share the full context of the same conversation. In practice, Claude can read data in a spreadsheet, build a financial model, then switch to PowerPoint to update a presentation, all without the user having to re-explain anything between steps. The other major addition involves "Skills," complete workflows turned into reusable one-click actions. Anthropic provides a finance-oriented starter kit: model auditing, LBO/DCF template building, comparable analyses, and data cleaning on the Excel side, along with competitive deck creation and presentation review on the PowerPoint side. Skills already configured in Claude (desktop or web) work directly within the add-ins. On the deployment side, the add-ins are now accessible through Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry, allowing enterprises to route traffic through their own LLM gateway. Claude also works natively in Agent Mode within Excel, alongside Microsoft 365 Copilot. The update is available in beta for all Mac and Windows users on paid plans. claude.com/blog/claude-ex…

🔹 Funding the Commons takes over San Francisco for a 36-hour festival dedicated to AI and public goods The 2026 edition of Funding the Commons will take place on March 14–15 at frontiertower in San Francisco (formerly Burning Man's HQ), strategically timed between GDC and NVIDIA GTC. The event, incubated by Protocol Labs, bills itself as a vertical festival spread across 16 floors, each operating semi-autonomously with its own budget and documented governance. The program blends conferences, co-design workshops, and hands-on experiments around a central theme: building coordination infrastructure between humans and AI in service of the commons. Topics covered include physical AI and robotics, agentic funding, model safety and evaluation, and sovereign infrastructure. Live public goods funding experiments will be conducted on-site, and an overnight hackathon will bring together up to 200 builders around these same themes. On the practical side, the venue houses a community biolab (genetic testing, disease research), a 4,000 sq ft maker lab equipped with lasers and 3D printers, AI agent workshops featuring motion capture and GPUs, as well as immersive art installations and live performances running until sunrise. Notable detail: 20% of profits will feed a community treasury co-managed by humans and AI. Among the speakers: Illia Polosukhin (co-author of the Transformer paper, NEAR Protocol), Tom Kalil (former White House science advisor), Brian Behlendorf (EFF, Mozilla), Leo Gao (EleutherAI), and Wendy Hanamura (Internet Archive). Over 1,000 attendees are expected, with support from Protocol Labs, ElevenLabs, NEAR Protocol, Filecoin Foundation, Solana, and Bittensor, among others.

🔹 Anthropic creates the Anthropic Institute to anticipate AI's societal impacts The Anthropic Institute is a new structure dedicated to studying the major challenges that powerful AI will pose to societies. Its ambition is twofold: publicly share what the company learns while building frontier models, and collaborate with researchers, communities, and industries to anticipate the transformations ahead. The Institute consolidates and expands three existing Anthropic teams: the Frontier Red Team (stress-testing the outermost limits of model capabilities), Societal Impacts (studying real-world AI use), and Economic Research (tracking impact on jobs and the economy). New efforts are already underway, including forecasting AI progress and understanding how powerful AI will interact with the legal system. Leadership is entrusted to Jack Clark , Anthropic co-founder, who takes on the title of Head of Public Benefit. Among the founding hires: Matthew Botvinick (formerly Google DeepMind, specializing in AI and law), Anton Korinek (economist, University of Virginia), and Zoë Hitzig (formerly OpenAI, socioeconomic impacts). In parallel, Anthropic is expanding its Public Policy team under Sarah Heck's leadership and will open a Washington, D.C. office this spring. The company states a clear conviction: AI progress will accelerate considerably over the next two years, and society needs to start preparing now.

🔹 Sword Health launches Pulse, an AI-powered solution for continuous cardiometabolic care Sword Health, a specialist in AI-augmented healthcare, is rolling out Pulse, a new program designed to provide integrated management of multiple cardiometabolic conditions: hypertension, prediabetes, type 2 diabetes, high cholesterol, and weight management, including support for patients on GLP-1 medications. A major challenge in the United States, where these conditions affect more than half of the adult population and account for over $400 billion in annual healthcare spending. The system relies on an AI/human tandem. On the AI side, an assistant called Phoenix provides personalized daily follow-up through persistent clinical memory, drawing on data from connected devices (blood pressure, glucose, weight, activity). On the human side, clinicians oversee patient journeys and step in as needed. The approach prioritizes lifestyle changes (nutrition, movement, sleep, stress management) before any medication escalation. Sword Health reports concrete outcomes: 91% of members report improved health status, a gross ROI of 2.5:1 for employers, and an outcome-based pricing model. Pulse integrates with Sword's existing ecosystem (pain, mental health, women's health) and targets employers, health plans, and brokers.

🔹 Runway creates Runway Labs, an internal incubator for next-generation AI video applications On March 11, Runway announced the launch of Runway Labs, an internal structure dedicated to exploring new use cases for its generative video models and General World Models. The company believes AI video has reached an inflection point and that its impact will extend far beyond filmmaking, into healthcare, education, gaming, advertising, sports, and real estate. The incubator will be led by Alejandro Matamala-Ortiz , Runway's co-founder and Chief Innovation Officer, with a team dedicated to prototyping experimental products. The stated goal: identify and develop applications that don't exist yet. Runway is actively hiring for this initiative: design engineers, ML and full-stack engineers, builders, and ex-founders are encouraged to apply through the company's website.

🔹 Higgsfield AI launches Soul Cast, an AI casting tool to create customizable virtual actors Higgsfield is expanding its Cinema Studio 2.0 platform with Soul Cast, a feature that lets users design fully customizable AI actors that remain visually consistent from scene to scene. The tool offers granular control over each character: film genre, era (dating back to the 1900s), narrative archetype (hero, rebel, sage…), identity (gender, origin, age), detailed physical appearance (build, hairstyle, scars, tattoos), and wardrobe. The main value lies in the character's visual persistence across successive generations, a recurring issue in AI video workflows. Once parameters are set, the actor remains identical from shot to shot, opening up practical applications for independent filmmakers, recurring content creators, advertising agencies, and e-learning. Higgsfield positions Soul Cast as an alternative to the constraints of traditional casting: no talent budget, no shooting logistics, and the ability to quickly generate variants for A/B testing or localized campaigns.

🔹 ElevenLabs unveils Flows, a node-based editor for orchestrating multimodal creative pipelines Flows is a new visual workspace integrated into the ElevenCreative platform. The concept relies on an infinite canvas where users connect nodes to build complete generation chains, from image to video, voice, music, and sound effects. The tool provides access to over 35 image and video models, combined with ElevenLabs' in-house technologies (text-to-speech, lip-sync, music, SFX). Each node can be re-run independently without regenerating the entire pipeline, enabling rapid variant testing at scale. ElevenLabs specifically highlights a concrete use case: performance marketing, where products, avatars, and voices can be swapped to produce ready-to-test creatives in batches. Flows also offers a library of preconfigured templates and integrates with Studio, ElevenCreative's timeline editor, for final adjustments. The feature is currently in alpha, available only to paid subscribers, and API access has been announced for the near future.

🔹 Perplexity AI launches "Personal Computer," a local, always-on AI OS Personal Computer is the latest addition to the Perplexity ecosystem. The idea: turn a Mac (such as a Mac mini) into an always-on AI assistant, acting as a digital proxy of its user around the clock. The system is built on Perplexity Computer, launched in late February, which orchestrates 19 AI models (including Opus 4.6, Gemini, Grok, and GPT-5.2) to break down complex goals into subtasks executed in parallel. Personal Computer takes the logic further by anchoring this orchestration directly on the local machine, with permanent access to the user's files, applications, and sessions. On the use case side, Perplexity highlights autonomous workflow management: sorting job applications, drafting emails, generating financial reports, and conducting market research through premium sources. The whole thing connects to over 400 services (Slack, Gmail, GitHub…) and can be controlled remotely from any device. On the security front, every sensitive action requires explicit approval, all operations are logged, and a kill switch can shut everything down instantly. Perplexity states it does not train its models on user data. The product is currently available via waitlist. As a reminder, access to Perplexity Computer requires the Max subscription, priced at $200/month.