Santosh C.

One of the most transformative innovations powering today’s GenAI revolution is the advancement of foundation models (FMs) — large models pre-trained on Internet-scale data that demonstrate remarkable generalization across a wide range of tasks. FMs have redefined performance benchmarks in many fields including, natural language processing, computer vision, and computational biology, becoming the cornerstone of modern GenAI. Yet, progress on structured enterprise data such as tabular time series has lagged behind, where task-specific models optimized on manually extracted signals still dominate. Inspired by the success of FMs in other domains, we set out to explore whether FMs can also be built for finance and banking. After over three years of R&D, I’m excited to share that we’ve deployed our first structured-data FM into production at TD: TD AI Prism https://lnkd.in/gEAb7Scm. Trained on billions of customer events, Prism generalizes to many downstream predictive tasks, delivering higher accuracy and significantly faster time to market. Through this work, we're beginning to observe early signs of scaling laws, similar to those seen in LLMs, where more data and larger models lead to stronger generalization. As part of this effort, we also released a public FM trained on Internet tabular data, capable of making predictions on new datasets without any fine-tuning. Check it out here: https://lnkd.in/gmbWdAf7 As enterprises evolve with AI, we believe that foundation models for structured data will be a major force in redefining how predictive AI is deployed at scale, driving the next wave of innovation in enterprise applications. Layer 6 AI has multiple exciting initiatives in this space, so stay tuned for more updates!

361

15 Comments

GCCs Shift from Cost Centers to Innovation Engines Global Capability Centers are evolving beyond cost efficiency to drive enterprise-wide innovation. India’s GCCs are leading this shift with AI-driven transformation, tighter alignment to business goals, and increasing adoption of generative AI across functions. #GCC #AI

4

Know Your Model: Magistral by Mistral AI 1. Model Snapshot Magistral – Mistral AI’s first dedicated reasoning LLM, launched in June 2025. Comes in two versions: open‑weight Magistral Small and enterprise-focused Magistral Medium. 2. What It Is? A chain-of-thought reasoning model built to simulate multi-step, logical thinking, offering transparency into its reasoning process. Small is for local use, Medium targets businesses. 3. Under the Hood (Tech Highlights) Architecture & Size - Small: 24 B parameters, Apache‑2.0 open‑weight - Medium: ~45 B parameters, proprietary, available via Le Chat and API Training Reinforcement Learning from Verifiable Rewards (RLVR), supervised traces from Medium, no bootstrapping from other models Capabilities - Multilingual reasoning: supports English, French, Spanish, German, Italian, Arabic, Chinese, and many more - Up to 128K context window (recommend ~40K for quality) 4. What Makes It Unique - Built around explicit chain-of-thought reasoning with traceable, step-by-step outputs - Magistral Medium delivers up to 10× faster throughput in Le Chat “Flash Answers” mode - Open‑source Small variant encourages transparency, community adoption 5. Where You’ll See It - Magistral Small: available on Hugging Face and deployable locally (e.g., RTX 4090 GPU or 32GB Mac with quantization) - Magistral Medium: preview via Mistral’s Le Chat, API, Amazon SageMaker, with upcoming support on IBM WatsonX, Azure AI, Google Cloud Marketplace 6. Pro Tips - Best tasks: logic-heavy workflows—legal research, financial modelling, structured coding, strategic analysis. Great when transparency is essential. - Prompting: ask it to “show your reasoning step-by-step”. 7. Other similar models - Google’s Gemini 2.5 Pro - OpenAI’s o3 - Claude Sonnet 4 Sources: https://lnkd.in/e8aQAide https://lnkd.in/eaEDmA92 Next: exploring OpenAI’s GPT o3 – stay tuned! #KnowYourModel

7

1 Comment