AI BriefingAI21 Labs news
The latest AI and developer news from AI21 Labs, with short summaries.
Visit website →AI Handles All the Work
AI21 Labs' Ori Goshen presented Maestro at RAISE Summit 2025 to address LLM limitations.
The Foundation of Enterprise AI: Grounding
AI21 presented grounding as the key challenge for enterprise AI.
Karpathy's Leash
Karpathy emphasized that control matters more than full autonomy for AI agents.
Automating Product Descriptions
A planning-based AI pipeline was presented as a way to manage both product description quality and regulation together.
AI-based compliance monitoring
It presents an AI system that cross-checks regulatory changes against contracts and policies in real time to find compliance gaps.
Maestro Turning Knowledge into Practice
For AI agents to succeed, they must go beyond simple information retrieval and perform knowledge work that handles complex reasoning and context.
Maestro AI, a System for Enterprise Knowledge Work
AI21 has released Maestro for enterprise customers.
RAG Agent Solutions
RAG agents should be evaluated on accuracy, observability, adaptability, time-to-value, and enterprise readiness.
How Boards Should Design AI Strategy
Boards need to manage AI from a business, risk, and ROI perspective rather than a technology one.
RAG Evaluation, the Current Approach Is Wrong
RAG evaluation should look at real-world document relationships rather than just scores.
AI21 Joins NVIDIA Inception
AI21 has joined NVIDIA Inception to strengthen its enterprise AI deployment capabilities.
NVIDIA Maestro
AI21's Maestro combines with NVIDIA NIM to target the shift of enterprise AI into production.
Maestro AI Plan Orchestration
AI21 unveiled Maestro plan orchestration to automate data-intensive enterprise work.
AI Reasoning: Planning vs Prediction
The future of AI reasoning lies not in token prediction but in action planning, the authors argue.
Technical Overview of Maestro
AI21's Maestro plans and executes LLM tasks through requirement verification and iterative refinement.
Jamba 1.6 Released
AI21 released Jamba 1.6, an open model supporting 256K context.
Enough with Prompt and Pray
For enterprise AI, system design and product fit matter more than the LLM alone.
Accuracy at Scale
AI21 has presented the QLA concept to guarantee accuracy and reliability for enterprise AI.
Enterprise AI
Enterprise AI emphasized that trustworthy, controllable, and traceable systems matter more than benchmarks.
Is GenAI Living Up to the Hype? Insights from 3 Years in the Field
Based on 3 years of field experience, AI21 lays out GenAI's limitations and Jamba's solution.