The Ravit Show

AI models are easy. Making them useful is hard. At NVIDIA GTC, I had a blast chatting with Roberto Barroso-Luque from Fireworks AI at the Amazon Web Services (AWS) booth. Fireworks helps teams fine-tune open-source models and run them in production. Built on AWS, it gives you scale, speed, and reliability without managing infrastructure. Use cases are already real. Coding assistants. Customer support that resolves issues You can access it through AWS. Marketplace or start with an API. From models to real applications. #data #ai #AWSPartner #NVIDIAGTC #fireworks #api #models #theravitshow

Everyone is talking about AI. But what actually powers it? At NVIDIA GTC, I spoke with Travis White from Zilliz at the Amazon Web Services (AWS) booth. Exciting conversation!!!! We talked about vector databases. The layer that helps AI search through images, videos, and text, not just tables. Zilliz offers Milvus (open source) and a managed version on AWS. That means you can scale, stay secure, and avoid managing infrastructure. Real use cases are already here. Autonomous vehicles Drug discovery, and a few more You can also find Zilliz on AWS Marketplace and get started quickly. Simple idea. AI is not just models. It is the data layer underneath. Learn more from the conversation!!!! #data #ai #awspartner #nvidiagtc #theravitshow

Everyone is building RAG. But the real question is what comes next. At NVIDIA GTC, I spoke with Jay Wilder from deepset, makers of Haystack, and the focus was clear. Moving from RAG to agents. Deepset, the team behind Haystack, is helping developers build flexible AI systems where you can choose your models, data, and guardrails. What stood out was their demo. They showed how RAG pipelines can evolve into agent-based systems by integrating across ecosystems like AWS, NVIDIA, and tools like Weights & Biases. This is where things get real. In banking, workflows that took weeks are now being done in hours. In pharma, teams are testing and iterating on complex data pipelines much faster. And all of this is built with flexibility in mind. Not locked into one stack. With Amazon Web Services (AWS) as a long-time partner and availability through AWS Marketplace, it is becoming easier for teams to get started and scale. This is the shift I am seeing. From static pipelines to adaptive AI systems. #data #ai #awspartner #NVIDIAGTC #deepset #api #haystack #theravitshow

One thing I keep thinking about from hashtag#NVIDIAGTC. We are moving too fast in building AI, but not fast enough in validating it. I spoke with David Arakelyan from Deepchecks, and the conversation really came down to trust. Once you move beyond demos, the question is no longer what your AI can do, but whether you can rely on it in production. Deepchecks is focused on this exact problem. Their LLM evaluation and monitoring solution helps teams test and stress their systems before they go live. What stood out was their new feature, Know Your Agent, which lets you generate a full report on your AI system with just an API key and understand where it can break. They are also one of the few solutions integrated directly into AWS SageMaker as a partner app and available on AWS Marketplace, making it easier for teams to bring this layer into their workflows. What do you think about it? #data #ai #awspartner #NVIDIAGTC #deepcheck #api #evals #theravitshow

Everyone at NVIDIA GTC is talking about AI. But my conversation with Vinay Sridhar from Snowflake came down to one thing. AI is only as good as the data behind it. Snowflake is building Cortex, an AI layer that runs directly on governed enterprise data. So instead of moving data to AI, teams bring AI to their data. We also touched on how tools like Cortex Code and text-to-SQL are changing how fast teams can build. And behind all of this is Amazon Web Services (AWS), powering the infrastructure and scale. This is the shift. From models to data-driven AI systems. Learn more from the conversation!!!! #data #ai #AWSPartner #NVIDIAGTC #snowflake #NVIDIAGTC #api #models #theravitshow

AI agents are everywhere at NVIDIA GTC. But here is what stood out in my conversation with Rudy Chetty from Amazon Web Services (AWS) on The Ravit Show. AWS is leaning heavily into this with Marketplace. A growing hub where you can search, purchase, and deploy AI solutions in minutes. From foundation models to vector databases to monitoring tools, everything is starting to come together in one place. We also talked about the role of partners. Because scaling AI is not a one-company job. It takes an ecosystem. AWS provides the infrastructure. More from the AWS Marketplace Kiosk at GTC. #data #ai #awspartner #nvidiagtc #nvidiagtc2026 #api #awsmarketplace #theravitshow

Data quality is still one of the most underestimated problems in AI. Everyone talks about models. Very few talk about whether the data behind those models can actually be trusted. At Gartner D&A, I sat down with Nick Oldham, COO at Equifax, and the conversation was very real. Equifax operates at the center of the global financial ecosystem. That means data integrity is not just a technical problem. It is a business risk. A regulatory risk. A trust problem. What stood out was this shift: - They are not treating data quality as a one-time cleanup anymore. - They are building systems to monitor, detect, and fix issues continuously. That is where platforms like Anomalo come in. Instead of reacting to broken dashboards or failed pipelines, the goal is simple Catch issues before they impact the business. And this becomes even more critical as AI enters the picture. Because AI systems do not fail loudly. They fail quietly when the data is wrong. Nick also shared how their journey was not just about tools. The harder part was organizational. - Aligning teams on what “good data” actually means - Moving from siloed ownership to shared accountability - Bridging the gap between data strategy and real execution Looking ahead, with investments in Google Cloud and tools like Dataplex, the direction is clear More automation More observability Less manual firefighting If you are building AI without solving for data integrity first, you are building on weak foundations. If this topic interests you, Nick will be going deeper into this at an upcoming session: The Road to Self-Driving Data, April 2 go for it https://engage.anomalo.com/the-road-to-self-driving-data This is one of those conversations that is less about hype and more about what it actually takes to make data work at scale. #data #ai #anomalo #dataquality #theravitshow

We are surrounded by the biggest Data and AI leaders at Gartner. Yet one question keeps coming up. Why are so many AI initiatives still failing? I asked Rich Hoyland, President, Global Field Operations, Anomalo to answer that directly on The Ravit Show at Gartner Orlando. His take was simple. AI does not fail because of ambition. It fails because of bad data. We spoke about: -- Why agentic AI raises the stakes for data quality -- Why speed without trust is a risk for executives -- What data quality failure actually looks like inside large enterprises -- Why traditional rule-based approaches are not enough anymore -- And where Anomalo sees the future of data management going One part of the conversation stood out. When AI moves from insight to action, data quality stops being a reporting issue. It becomes a business risk issue. Old approaches were built for dashboards. Now we are feeding data into agents that make decisions. That changes everything. Rich also shared how their long term vision goes beyond just catching bad data to providing enterprises with an agentic “self-driving data” system. It is about building continuous trust in data across the enterprise from ingestion to decision so AI agents can operate and scale safely. If you care about AI that actually works in production, this is one to watch. #data #ai #anomalo #dataquality #theravitshow

At Gartner D&A Day 1, I sat down with Antonio, CEO and Co-Founder of BEM, to talk about a problem many enterprises quietly struggle with. Messy data. While everyone is excited about agents, Antonio made one thing clear. Agents do not work well with unstructured, inconsistent inputs. That is where hallucinations and failures begin. BEM focuses on turning messy inputs, from PDFs and contracts to voice and video, into clean, structured outputs that enterprises can actually trust. We discussed: * Why so many AI pilots fail before reaching production * How BEM acts as the foundation layer before agents * Why regulated industries like healthcare and finance need production-grade accuracy * How some teams deploy in minutes by starting with one painful workflow The message was simple. If you want agents to work, fix the data first. #data #ai #bem #gartnerda #theravitshow

“Your data is fine. Your AI isn’t good enough.” That is the bold statement behind TextQL, and it immediately caught my attention here at Gartner. I sat down with Ethan Ding, Co-Founder, CEO & Head of Product, TextQL, to unpack what he means by that and why they are challenging many assumptions around BI and analytics. Most enterprises have spent years building ETL pipelines, cleaning data, and preparing dashboards. The belief has been that AI will only work once data is perfectly structured. Ethan disagrees. He believes the real limitation has been the AI systems themselves. We talked about: -- What enterprises are misunderstanding today about AI and data quality -- Why traditional BI tools like Tableau or Power BI were built for a different era -- How TextQL enables AI analytics even when data is messy or not fully ETL’d -- Why they believe seat-based pricing for dashboards is broken -- How their approach focuses on trust and verification so enterprises can validate AI-generated answers One idea stood out during the conversation. Executives do not just want answers. They want conviction that the answer is correct. That is where their “Query to Conviction” concept comes in. AI does not just generate an answer. It shows the reasoning, the data path, and the verification behind it. For CIOs walking the Gartner floor, Ethan had a simple suggestion. Do not ask vendors how good their AI looks. Ask them how their AI proves it is right. #data #ai #textql #gartnerda #theravitshow

Most AI analytics platforms assume one thing. Your data lives in one place. Kapil Chhabra, Co-Founder and CPO at WisdomAI on The Ravit Show challenged that assumption immediately. Enterprises are distributed. Their data is fragmented across clouds, warehouses, operational systems, and business units. Forcing everything into a single layer before AI can work is slow and expensive. That is why WisdomAI introduced Federated Agentic Intelligence. Instead of centralizing first and analyzing later, the system works across distributed sources. It assembles context at runtime. We spent time on what they call the Enterprise Context Layer. Without context, AI gives generic answers. With context, AI understands how metrics connect, what definitions mean, and how governance rules apply. Kapil was clear that federation is not a feature. It is a design principle for the modern enterprise. We also talked about what is next on their roadmap and what capabilities they are most excited about over the coming year. The focus is less on flashy features and more on depth, reliability, and scale. If your organization is struggling to move from AI experiments to AI that executives actually trust, this conversation goes deep into architecture decisions that matter. #data #ai #gartnerda #wisdomai #theravitshow

We have been building dashboards for 20 years. Now everyone is adding AI on top of them. But what if the real issue is the stack itself? That is where my conversation with Soham Mazumdar, Co-Founder and CEO, WisdomAI went at Gartner D&A on The Ravit Show!!!! WisdomAI calls what they are building “Agentic Analytics.” Not a chatbot on top of BI. Not a copilot that still depends on humans to interpret everything. We talked about what is fundamentally broken in today’s analytics world: - Dashboards answer questions you already thought of - Executives need answers to questions they did not know to ask Soham shared how enterprises are moving from static reporting to agents that reason across metrics, detect issues, and explain why something happened. The trust problem came up quickly. Most AI analytics tools look impressive in a demo. Very few hold up under real enterprise scrutiny. We also discussed a real customer story with Cisco and what changed after deploying WisdomAI. The shift was not just faster answers. It was decision confidence. Looking ahead, Soham believes analytics teams will not disappear. They will evolve into designers and supervisors of intelligent systems that operate continuously across the business. For enterprise leaders rethinking the future of BI, this was a forward-looking and very practical discussion. #data #ai #gartnerda #wisdomai #theravitshow

#IBMPartner Governance, responsible AI implementation and delivering measurable value —these are top of mind for Jordan Byrd, AI/ML Ops Product Marketing Lead at @IBM. AI adoption feels different this year—faster with more framework. Watch our conversation from Gartner D&A where we caught up to discuss what’s really changing inside enterprises and what that means for the next phase of AI. If you are building AI at scale inside an enterprise, this one will resonate. Learn more at https://ibm.biz/Bdp2gG #Data #AI #IBM #GartnerDa #API #AIOPS #Governance #TheRavitShow

Everyone says AI is the priority. Yet many projects are quietly failing. At Gartner D&A, I asked Christopher Moore, Global Sr. Director, AI & Platform at Alteryx, to be direct about why. His answer was not about models. It was about execution. Too many AI initiatives are disconnected from real business workflows. They look good in a lab. They struggle in operations. We then got into a bigger tension inside enterprises. Business teams understand the problem best. But they rarely build the AI solutions themselves. Why? Because the tooling has been too technical. Too fragmented. Too dependent on centralized teams. Christopher explained how Alteryx is trying to close that gap. Not by lowering standards, but by enabling governed, production-grade AI where business users already work. We also talked about what MCP server and Agentspace unlock for long time Alteryx users. In simple terms, it is about moving from isolated workflows to orchestrated AI systems. From analytics automation to agent-enabled automation. And then we addressed the elephant in the room. Alteryx was once labeled shadow IT. That perception has shifted. In a world where AI governance is critical, the focus is now on controlled enablement. Visibility. Auditability. Guardrails built in. The message was clear. Empowering business users does not mean losing governance. It means designing platforms that balance speed with control. If you are navigating the tension between innovation and oversight, this is a conversation you will want to watch. #data #ai #gartnerda #alteryx #theravitshow

Synthetic data is everywhere in AI conversations!!!! But what does it actually solve? I had an amazing conversation with Michael Eckhoff on The Ravit Show at Gartner he brought this down to reality. We spoke about when synthetic data makes more sense than masking or subsetting production data. It shines when: • Compliance makes moving production data into lower environments a bottleneck • Teams need data that simply does not exist • Rare edge cases are missing from real datasets Synthetic data lets teams generate fit-for-purpose datasets on demand without copying real customer records across environments. We also tackled the big concern. Is synthetic realistic enough? Realistic does not mean copied. It means the relationships hold. The distributions look right. The system behaves the same way. And you prove it. You compare statistical properties. You validate patterns. You ensure no record is traceable to a real individual. Finally, where does synthetic fit in AI and GenAI? It removes the compliance friction. It helps balance datasets. It enables experimentation without exposing sensitive information. For AI teams trying to move fast and stay compliant, this is a serious lever. #data #ai #gartner #k2view #theravitshow

I had a blast at Gartner last week, here's my discussion with Hod Rotem from K2view on The Ravit Show, diving into one of the most important topics right now. What does AI-ready data architecture actually look like when it is running in production? We will break down: * How real-time, entity-level data gets assembled across dozens of systems * What it takes to support thousands of AI agents working in parallel * Why architecture, not just models, determines whether AI actually works If you are thinking about agentic AI beyond demos, this will be a practical and direct conversation. #data #ai #gartnerda #K2View #theravitshow

Mainframes. Synthetic data. AI-ready foundations. This was one of the most practical conversations we had at Gartner on The Ravit Show. I sat down with Ronen Schwartz, CEO at K2view and Michael Curry, President, Data Modernization, Rocket Software to talk about their partnership and why it matters right now. Here is the reality. A lot of enterprise data still lives in mainframes and core systems of record. At the same time, teams are racing to automate development, generate code with AI, and move faster than ever. That creates a real gap. We discussed: -- Why customers building data products, test data management, and synthetic data pushed K2View and Rocket Software to collaborate -- How modernization of legacy systems creates opportunities to generate and manage test data at scale -- Why synthetic data is critical when you cannot simply move production data into lower environments -- How teams can now generate code from a product story and also generate the data needed to test it -- Why governance is the layer leaders must get right before scaling AI One point stood out. The technology leap toward AI is not the hardest part. Getting the data foundation, quality, and governance right is. If AI agents are going to act on enterprise data, that data must be trusted, protected, and consistent across systems of record. Their advice to leaders was simple. - Build AI-ready data environments. - Partner with vendors who are deep in what they do. - Carry your governance investments forward into your agentic AI strategy. If you are modernizing mainframes, thinking about synthetic data, or preparing your enterprise for AI in production, this one is worth watching. #data #ai #rocketsoftware #gartnerda #k2view #api #mainframes #enterprise #theravitshow

Five years advising CDAOs at Gartner D&A. Now in the field helping enterprises actually implement AI and governance. That shift gives Austin Kronz, Head of AI & Data Strategy, Atlan, a rare lens. And this conversation was honest. We talked about the gap between what we say on stage at big events and what really happens inside companies once everyone flies home. Here is what we unpacked: • What surprised him most moving from analyst to operator • The real signals coming out of this summit around AI governance, metadata, and context for AI agents • The controlled experiments around context layers at companies like Workday and Fox, and what actually drove up to 5x improvement in AI accuracy • Where Fortune 500 teams get stuck when they say “we need AI governance” • The patterns he sees in companies like Cargill and PPG that succeed with context at scale One theme kept coming up.The winners are not the ones talking the most about AI. They are the ones operationalizing context, ownership, and governance in very practical ways. Whether you attended Gartner D&A or not, this one is worth watching. #data #ai #gartnerda #atlan #theravitshow

Your AI has a context problem. It’s not the model. It’s context!!!! That was the mic drop from Prukalpa, Co-Founder & Co-CEO, Atlan on The Ravit Show when we kicked off this conversation. And honestly, it set the tone for everything that followed. We spoke about why so many AI projects stall. Not because the model is weak. Not because the team is not smart. But because the data lacks shared meaning. No common definitions. No clear ownership. No business context. Here is what we unpacked: • What a “context layer” actually means in simple terms • Why this idea is suddenly everywhere at Gartner this year • Where the context layer fits in the modern data stack • What Atlan is building to make context usable, not theoretical • The one demo every data leader should see before leaving Orlando One big takeaway: If your AI does not understand your business context, it will confidently give you the wrong answer. If you are at Gartner this week, stop by Booth 313. See how context is being turned into something real, usable, and operational. #data #ai #gartnerda #atlan #theravitshow

"On-prem is the new cloud.” That statement is not just a headline. It reflects what many enterprises are quietly experiencing. I sat down with David Dichmann, VP Product Marketing and Evangelism, Cloudera, to unpack what is really driving this shift and how Cloudera’s Cloud Anywhere vision fits into the bigger picture. Here is what we discussed: -- Why rising cloud costs, data gravity, and regulatory pressure are pushing companies to rethink all-in cloud strategies -- What Cloud Anywhere actually means beyond marketing -- How enterprises can run advanced AI use cases without forcing massive data movement into one environment -- Why security, governance, and Private AI are central to this resurgence of on-prem -- The biggest roadblocks teams face when deploying AI across hybrid and multi-cloud environments -- How portability and consistency reduce friction for data and AI teams -- How Cloudera continues to lean into its open-source roots while evolving its platform One theme stood out. AI does not require you to centralize everything into one cloud. It requires control, flexibility, and a consistent experience wherever your data lives. For enterprises balancing cost, compliance, and AI ambition, this conversation goes beyond trends. It is about architecture decisions that will shape the next few years. #data #ai #cloudera #gartnerda #theravitshow"On-prem is the new cloud.” That statement is not just a headline. It reflects what many enterprises are quietly experiencing. I sat down with David Dichmann, VP Product Marketing and Evangelism, Cloudera, to unpack what is really driving this shift and how Cloudera’s Cloud Anywhere vision fits into the bigger picture. Here is what we discussed: -- Why rising cloud costs, data gravity, and regulatory pressure are pushing companies to rethink all-in cloud strategies -- What Cloud Anywhere actually means beyond marketing -- How enterprises can run advanced AI use cases without forcing massive data movement into one environment -- Why security, governance, and Private AI are central to this resurgence of on-prem -- The biggest roadblocks teams face when deploying AI across hybrid and multi-cloud environments -- How portability and consistency reduce friction for data and AI teams -- How Cloudera continues to lean into its open-source roots while evolving its platform One theme stood out. AI does not require you to centralize everything into one cloud. It requires control, flexibility, and a consistent experience wherever your data lives. For enterprises balancing cost, compliance, and AI ambition, this conversation goes beyond trends. It is about architecture decisions that will shape the next few years. #data #ai #cloudera #gartnerda #theravitshow