Category Robots Podcast

Over the past decades, roboticists have introduced a wide range of systems that can effectively tackle some real-world problems. Most of these robots, however, often perform poorly on tasks that they were not trained on, particularly those that entail manipulating previously unseen objects or handling objects that were encountered before in new ways.

The shift from prototyping to having agents in production is the challenge for AI teams as we look toward 2026 and beyond. Building a cool prototype is easy: hook up an LLM, give it some tools, see if it looks like it’s working. The production system, now that’s hard. Brittle integrations. Governance nightmares. Infrastructure wasn’t built for the complexities and nuances of agents. 

For AI developers, the challenge has shifted from building an agent to orchestrating, governing, and scaling it in a production environment. DataRobot’s latest release introduces a robust suite of tools designed to streamline this lifecycle, offering granular control without sacrificing speed.

New capabilities accelerating AI agent production with DataRobot

New features in DataRobot 11.2 and 11.3 help you close the gap with dozens of updates spanning observability, developer experience, and infrastructure integrations.

Together, these updates focus on one goal: reducing the friction between building AI agents and running them reliably in production. 

The most impactful areas of these updates include:

• Standardized connectivity through MCP on DataRobot
• Secure agentic retrieval through Talk to My Docs (TTMDocs) 
• Streamlined agent build and deploy through CLI tooling
• Prompt version control through Prompt Management Studio
• Enterprise governance and observability through resource monitoring
• Multi-model access through the expanded LLM Gateway
• Expanded ecosystem integrations for enterprise agents

The sections that follow focus on these capabilities in detail, starting with standardized connectivity, which underpins every production-grade agent system.

MCP on DataRobot: standardizing agent connectivity

Agents break when tools change. Custom integrations become technical debt. The Model Context Protocol (MCP) is emerging as the standard to solve this, and we’re making it production-ready. 

We’ve added an MCP server template to the DataRobot community GitHub.

• What’s new: An MCP server template you can clone, test locally, and deploy directly to your DataRobot cluster. Your agents get reliable access to tools, prompts, and resources without reinventing the integration layer every time. Easily convert your predictive models as tools that are discoverable by agents.

• Why it matters: With our MCP template, we’re giving you the open standard with enterprise guardrails already built in. Test on your laptop in the morning, deploy to production by afternoon.

Talk to My Docs: Secure, agentic knowledge retrieval

Everyone is building RAG. Almost nobody is building RAG with RBAC, audit trails, and the ability to swap models without rewriting code. 

The “Talk to My Docs” application template brings natural language chat-style productivity across all your documents and is secured and governed for the enterprise.

• What’s new: A secure, governed chat interface that connects to Google Drive, Box, SharePoint, and local files. Unlike basic RAG, it handles complex formats from tables, spreadsheets, multi-doc synthesis while maintaining enterprise-grade access control.
  
• Why it matters: Your team needs ChatGPT-style productivity. Your security team needs proof that sensitive documents stay restricted. This does both, out of the box.

Agentic application starter template and CLI: Streamlined build and deployment

Getting an agent into production should not require days of scaffolding, wiring services together, or rebuilding containers for every small change. Setup friction slows experimentation and turns simple iterations into heavyweight engineering work.

To address this, DataRobot is introducing an agentic application starter template and CLI, both designed to reduce setup overhead across both code-first and low-code workflows.

• What’s new: An agentic application starter template and CLI that let developers configure agent components through a single interactive command. Out-of-the-box components include an MCP server, a FastAPI backend, and a React frontend. For teams that prefer a low-code approach, integration with NVIDIA’s NeMo Agent Toolkit enables agent logic and tools to be defined entirely through YAML. Runtime dependencies can now be added dynamically, eliminating the need to rebuild Docker images during iteration.

• Why it matters: By minimizing setup and rebuild friction, teams can iterate faster and move agents into production more reliably. Developers can focus on agent logic rather than infrastructure, while platform teams maintain consistent, production-ready deployment patterns.

Prompt management studio: DevOps for prompts

As prompts move from experiments to production assets, ad hoc editing quickly becomes a liability. Without versioning and traceability, teams struggle to reproduce results or safely iterate.

To address this, DataRobot introduces the Prompt Management Studio, bringing software-style discipline to prompt engineering.

• What’s new: A centralized registry that treats prompts as version-controlled assets. Teams can track changes, compare implementations, and revert to stable versions as prompts move through development and deployment.

• Why it matters: By applying DevOps practices to prompts, teams gain reproducibility and control, making it easier to transition from prototyping to production without introducing hidden risk.

Multi-tenant governance and resource monitoring: Operational control at scale

As AI agents scale across teams and workloads, visibility and control become non-negotiable. Without clear insight into resource usage and enforceable limits, performance bottlenecks and cost overruns quickly follow.

• What’s new: The enhanced Resource Monitoring tab provides detailed visibility into CPU and memory utilization, helping teams identify bottlenecks and manage trade-offs between performance and cost. In parallel, Multi-tenant AI Governance introduces token-based access with configurable rate limits to ensure fair resource consumption across users and agents.

• Why it matters: Developers gain clear insight into how agent workloads behave in production, while platform teams can enforce guardrails that prevent noisy neighbors and uncontrolled resource usage as systems scale.

Expanded LLM Gateway: Multi-model access without credential sprawl

As teams experiment with agent behavior and reasoning, access to multiple foundation models becomes essential. Managing separate credentials, rate limits, and integrations across providers quickly introduces operational overhead.

• What’s new: The expanded LLM Gateway adds support for Cerebras and Together AI alongside Anthropic, providing access to models such as Gemma, Mistral, Qwen, and others through a single, governed interface. All models are accessed using DataRobot-managed credentials, eliminating the need to manage individual API keys.

• Why it matters: Teams can evaluate and deploy agents across multiple model providers without increasing security risk or operational complexity. Platform teams maintain centralized control, while developers gain flexibility to choose the right model for each workload.

New supporting ecosystem integrations

Jira and Confluence connectors: To power your vector databases, DataRobot provides a cohesive ecosystem for building enterprise-ready, knowledge-aware agents.

NVIDIA NIM Integration: Deploy Llama 4, Nemotron, GPT-OSS, and 50+ GPU-optimized models without the MLOps complexity. Pre-built containers, production-ready from day one.

Milvus Vector Database: Direct integration with the leading open-source VDB, plus the ability to select distance metrics that actually matter for your classification and clustering tasks.

Azure Repos & Git Integration: Seamless version control for Codespaces development with Azure Repos or self-hosted Git providers. No manual authentication required. Your code stays centralized where your team already works.

Get hands-on with DataRobot’s Agentic AI 

If you’re already a customer, you can spin up the GenAI Test Drive in seconds. No new account. No sales call. Just 14 days of full access inside your existing SaaS environment to test these features with your actual data.  

Not a customer yet? Start a 14-day free trial and explore the full platform.

For more information, please visit our Version 11.2 and Version 11.3 release notes in the DataRobot docs.

The post DataRobot Q4 update: driving success across the full agentic AI lifecycle appeared first on DataRobot.

Operational excellence used to mean efficiency. Now, it means consistency. In a volatile manufacturing environment, the winners aren’t those with the best machines or biggest budgets — they’re the ones who can execute the same playbook flawlessly, every day, on every line.

Gemini 3 Flash offers frontier intelligence built for speed at a fraction of the cost.

Step inside the Soft Robotics Lab at ETH Zurich, and you find yourself in a space that is part children's nursery, part high-tech workshop and part cabinet of curiosities. The lab benches are strewn with foam blocks, stuffed animals—including a cuddly squid—and other colorful toys used to train robotic dexterity. Piled up on every surface are sensors, cables and measurement devices. Skeletal fingers, on show in display cases or attached to powerful robotic arms, seem to reach out to grab you from every corner.

Step inside the Soft Robotics Lab at ETH Zurich, and you find yourself in a space that is part children's nursery, part high-tech workshop and part cabinet of curiosities. The lab benches are strewn with foam blocks, stuffed animals—including a cuddly squid—and other colorful toys used to train robotic dexterity. Piled up on every surface are sensors, cables and measurement devices. Skeletal fingers, on show in display cases or attached to powerful robotic arms, seem to reach out to grab you from every corner.

KNF vacuum pumps for automation applications are designed for a long service life, with micro gas pumps used as cobot pumps achieving more than 20,000 hours. The latest generation of KNF brushless DC motors has an innovative bearing design that withstands high mechanical loads. This technical strength protects the vacuum pump's longevity, especially with fast switching cycles.

United Parcel Service Inc. will invest $120 million in 400 robots used to unload trucks, according to people familiar with the matter, revealing new details on the logistics giant's $9 billion automation plan that aims to boost profits by decreasing labor costs.

Computer-aided design (CAD) systems are tried-and-true tools used to design many of the physical objects we use each day. But CAD software requires extensive expertise to master, and many tools incorporate such a high level of detail they don't lend themselves to brainstorming or rapid prototyping.

By combining flexible vision technology with automated processing, manufacturers and system integrators can shorten deployment cycles, reduce reliance on fixtures, and achieve the adaptability needed for high-mix, high-precision production.

Open interpretability tools for language models are now available across the entire Gemma 3 family with the release of Gemma Scope 2.

A newly developed AI can predict which diseases specific genetic mutations are likely to cause, not just whether they are harmful. The breakthrough could speed up diagnoses and open new paths for personalized treatment.

Robots small enough to travel autonomously through the human body to repair damaged sites may seem the stuff of science fiction dreams. But this vision of surgery on a microscale is a step closer to reality, with news that researchers from the University of Pennsylvania and the University of Michigan have built a robot smaller than a millimeter that has an onboard computer and sensors.

Thirteen global teams push the boundaries of “no remote control” robotics, driving the evolution from teleoperated tools to autonomous agents

Generations in Dialogue: Bridging Perspectives in AI is a podcast from AAAI featuring thought-provoking discussions between AI experts, practitioners, and enthusiasts from different age groups and backgrounds. Each episode delves into how generational experiences shape views on AI, exploring the challenges, opportunities, and ethical considerations that come with the advancement of this transformative technology.

Human-robot interactions and social robotics with Professor Marynel Vázquez

In the fourth episode of this new series from AAAI, host Ella Lan chats to Professor Marynel Vázquez about what inspired her research direction, how her perspective on human-robot interactions has changed over time, robots navigating the social world, potential for using robots in education, modeling interactions as graphs, addressing misunderstandings with regards to robots in society, getting input from target users, the challenge of recognising when errors happen, making robots that adapt, and more.

About Professor Marynel Vázquez:

Marynel Vázquez is a computer scientist and roboticist whose research focuses on Human-Robot Interaction (HRI), particularly in multi-party settings. She studies social group dynamics—such as spatial behavior and social influence—in HRI, and develops perception and decision-making algorithms that enable autonomous, socially aware robot behavior. A central theme in her work is modeling interactions as graphs, allowing robots to reason about individuals, relationships, and groups simultaneously. Her interdisciplinary approach combines computer science, behavioral science, and design, and she enjoys building new robotic systems and research infrastructure to bring theoretical ideas into real-world practice.

About the host

Ella Lan, a member of the AAAI Student Committee, is the host of “Generations in Dialogue: Bridging Perspectives in AI.” She is passionate about bringing together voices across career stages to explore the evolving landscape of artificial intelligence. Ella is a student at Stanford University tentatively studying Computer Science and Psychology, and she enjoys creating spaces where technical innovation intersects with ethical reflection, human values, and societal impact. Her interests span education, healthcare, and AI ethics, with a focus on building inclusive, interdisciplinary conversations that shape the future of responsible AI.