• Forward Future Daily
  • Posts
  • šŸ§‘ā€šŸš€ Mistralā€™s Rise, Medical AI & Kryptos Code Mystery

šŸ§‘ā€šŸš€ Mistralā€™s Rise, Medical AI & Kryptos Code Mystery

Mistralā€™s AI rise, clinical AI for better care, Kryptos code mystery, ChatGPT growth, xAIā€™s Memphis expansion, Larry Pageā€™s new startup, and AI-driven military and propaganda news.

Good morning, itā€™s Monday. French startup Mistral is making big moves in the global AI race, AI systems are starting to think like a team of doctors, and some overconfident models believe theyā€™ve cracked Kryptos (spoiler: they havenā€™t).

Plus, in the first installment of our I Will Teach You to AI series, weā€™re diving into reasoning AIā€”how it works, when to use it, and when other models are a better fit.

Read on!

šŸ“Š MARKET PULSE

From AI Image Generation to Video Creation

OpenArt Hits $12M ARR

In the rapidly evolving landscape of AI-generated content, OpenArt has carved out a lucrative niche. Initially finding product-market fit with text-to-image AI models, the company has grown to a Sacra-estimated $12 million in annual recurring revenue (ARR) as of February 2025, marking a 1,100% year-over-year growth. ā†’ Read the full post here.

ā˜ļø POWERED BY LANGTRACE
Langtrace Inverted Logo

Langtrace is an AI-focused software development and consulting company that partners with enterprises to build AI products to propel business growth. Their products are open-source, SOC 2 Type II compliant, and power over 1 billion AI interactions monthly. Their observability and evaluations platform is trusted by thousands of developer teams ranging from startups to Fortune 500 enterprises, and their SDKs have been downloaded over 600K+ times.

Visit www.langtrace.ai to learn more about how Langtrace can help you build and transform your AI agents from prototypes to enterprise grade products.

šŸ—žļø ICYMI RECAP

Top Stories You Might Have Missed

šŸ§© AI Claims to Crack Kryptosā€”The Artist Says No Chance: AI-generated "solutions" to the Kryptos code are flooding Jim Sanbornā€™s inboxā€”but none are right. Sanborn insists AI is misleading would-be solvers, and heā€™s not giving away the answer anytime soon.

šŸ“ˆ ChatGPT Doubles Weekly Users in 6 Months: ChatGPT hit 400M weekly active users by Feb 2025, fueled by GPT-4o and new features like Advanced Voice Mode. Mobile use also surged, with 175M weekly users.

šŸ—ļø xAI Expands in Memphis with Massive Property Buy: Elon Muskā€™s xAI acquired a 1M sq. ft. site in Memphis to boost AI data center capacity. Plans include upgrading to 1M NVIDIA GPUs this year.

šŸš€ Larry Page Launches New AI Startup: Google co-founder Larry Page is reportedly behind a new AI startup, though details remain scarce. The venture signals growing competition in the AI race.

šŸ•µ Scale AI Under Investigation by U.S. Labor Department: The DOL is probing Scale AI over potential labor violations, including misclassifying workers as contractors. Scale AI denies wrongdoing and says it complies with labor laws.

šŸš› Palantir Delivers AI-Equipped Trucks to US Army: Palantir's $178M Titan contract includes mobile battle stations with AI-powered data systems. The rugged trucks double as high-tech workspaces for battlefield operations.

šŸ Apple Delays 'More Personalized Siri' Rollout: Appleā€™s upgraded Siri, promised with Apple Intelligence, is now delayed until 2026. Critics say Siri is falling behind rivals like ChatGPT.

šŸš© Russian Propaganda Reportedly Influencing AI Chatbots: A Moscow-based network, Pravda, allegedly planted false claims repeated by AI chatbots 33% of the time. SEO tricks may be helping the disinformation spread.

šŸ“… Google Adds Gemini Panel to Calendar: Googleā€™s AI-powered Gemini now helps manage schedules in Calendar, letting users add events and check details conversationally. Currently in early access via Workspace Labs.

šŸš€ MISTRAL

How Europe's AI Champion is Carving Its Own Path

Mistral, Europeā€™s biggest AI

The Recap: Mistral, a French AI startup founded in 2023, has emerged as Europe's leading AI company despite facing stiff competition from American and Chinese giants. While its models lag behind top-tier labs like OpenAI and Anthropic, Mistralā€™s open-source strategy and European identity are helping it stand out in an increasingly geopolitical AI landscape.

Highlights:

  • Mistral was valued at ā‚¬5.8 billion ($6.2 billion) in June 2024, making it Europeā€™s most valuable AI startup

  • It has released 16 models in 18 months, including specialized ones like Codestral (coding), Mathstral (math), and Pixtral (image recognition)

  • Mistralā€™s open-source models are free to use for non-commercial purposes, setting it apart from American rivalsā€™ closed models

  • Rising U.S.-EU tensions have made Mistralā€™s European roots more appealing to customers wary of American and Chinese dominance

  • In February, Mistral launched Le Chat, a fast ChatGPT-style assistant built on Cerebras chips, amassing 1 million downloads in 10 days

  • French companies like BNP Paribas and Orange have backed Mistral, with user reviews highlighting its European identity as a selling point

  • Despite a high valuation, Mistralā€™s recurring revenue is only ā‚¬30 millionā€”just 0.5% of its market worth

Forward Future Takeaways:
Mistralā€™s ability to thrive hinges on its open-source approach and growing European pride amid geopolitical tension. While its financials are underwhelming for now, the companyā€™s strategic positioning as a European alternative to U.S. and Chinese AI models could secure long-term success as AIā€™s political and commercial stakes continue to rise. ā†’ Read the full article here.

šŸ§‘ā€šŸ« I WILL TEACH YOU TO AI [PREMIUM]

Unlocking the Power of AI Reasoners

Reasoning AIs, or reasoners, are AI models designed to draw logical conclusions, recognize relationships, and make decisions based on structured logic. Unlike generative AI models (such as GPT-4o or Claude 3.5), which rely on pattern recognition and probabilities, reasoners follow a ā€œchain-of-thoughtā€ approach.

Inspired by Daniel Kahnemanā€™s System 2 thinking (slow, logical reasoning), reasoners analyze data step by step, ensuring logical consistency rather than responding based on probability.

šŸ©ŗ HEALTHCARE

Teaching AI to Think Like a Team of Doctors

Train clinical AI

The Recap: As the EUā€™s Artificial Intelligence Act takes effect, AI systems in clinical settings face intense scrutiny due to their high-risk nature. To gain trust and comply with regulations, AI models should mirror the reasoning process of medical teamsā€”focusing on shared, interpretable concepts rather than opaque statistical predictions.

Highlights:

  • The EU AI Act imposes strict oversight on high-risk AI, including medical applications, requiring transparency and human oversight.

  • Current AI decision-making lacks explainability, making it difficult for doctors to trust and integrate AI recommendations into patient care.

  • Medical teams use concept-based reasoning, discussing diagnoses and treatment plans through shared, high-level medical terms rather than raw data.

  • AI models that mimic multidisciplinary team discussionsā€”using interpretable concepts like "tumor stage" rather than abstract probability scoresā€”could improve trust and compliance.

  • Explainability methods like saliency mapping (highlighting key pixels in images) have limitations, as they may not align with human reasoning.

  • Concept Bottleneck Models (CBMs) offer a solution by training AI to justify decisions with clinically meaningful intermediate concepts, aiding transparency.

  • These models allow clinicians to correct AI mistakes in real-time, helping the system learn and improve through human feedback.

Forward Future Takeaways:
The EU AI Act signals a shift toward accountability and interpretability in clinical AI. By adopting concept-driven AI reasoning, healthcare systems can ensure that AI not only improves efficiency but also maintains trust and regulatory compliance. Future AI models will need to think less like machines and more like collaborative medical experts, fostering human-AI teamwork rather than replacing human judgment. ā†’ Read the full article here.

šŸ§‘ā€šŸš€ SPREAD THE WORD

Love Forward Future? Help us grow. Share this newsletter with a friendā€”just forward this email or tap the button below.

šŸ“½ļø VIDEO

QwQ: Tiny Thinking Model That Tops DeepSeek R1 (Open Source)

Alibaba's QWQ-32B is a new open-source AI model with 32B parameters, delivering DeepSeek R1-level performance while running on local machines. Trained with reinforcement learning for math and coding, it achieves fast reasoning at 450 tokens/second. Get the full scoop! šŸ‘‡

šŸ§° TOOLBOX

Smarter Document Parsing, AI Voice Agents, and Creative Art Generation

šŸ— OmniParser V2: Extract structured data from complex documents with AI, handling both text and images for better document parsing.

šŸ“ž EchoWin: Build AI voice agents to handle calls, chats, and Discord support 24/7, with multilingual and sentiment analysis capabilities.

šŸŽØ OpenArt AI: Generate and edit art with AI using text, images, and styles; create custom models, blend images, and upscale with ease.

šŸ—’ļø FEEDBACK

Help Us Get Better

What did you think of today's newsletter?

Login or Subscribe to participate in polls.

šŸ¤  THE DAILY BYTE

India's 'Drone Sisters' Are Breaking Traditions and Flying High

CONNECT

Stay in the Know

Follow us on X for quick daily updates and bite-sized content.
Subscribe to our YouTube channel for in-depth technical analysis.

Prefer using an RSS feed? Add Forward Future to your feed here.

Thanks for reading todayā€™s newsletter. See you next time!

The Forward Future Team
šŸ§‘ā€šŸš€ šŸ§‘ā€šŸš€ šŸ§‘ā€šŸš€ šŸ§‘ā€šŸš€ 

Reply

or to participate.