Audrey Rusch

“The Ontology is the substrate through which humans and AI work together to create operational value — rooted in your business, compliant with your norms, and measured against the outcomes you care about.“ — Palantir Chief Architect, Akshay Krishnaswamy 🎤 ⬇️

571

12 Comments

Cross-functional work is where good ideas go to die in many companies. It’s where you see the most Tension, Politics, Slowed-down decisions, and “This isn’t my job” behavior. So I sat down with Nick Palomba, who has led incredibly complex cross-functional projects at Microsoft, to ask him one question: How do you get different teams to actually move in the same direction? Here are 3 pieces of advice that stood out: 1️⃣ If nobody has time, maybe it’s not important If every team is “too busy” to help, it might mean: - The problem isn’t clearly defined - The impact isn’t compelling enough - Or the timing simply isn’t right As a leader, you either sharpen the “why”, or you stop pushing a project that isn’t truly strategic. 2️⃣ The job of the leader is not to know everything, but to give clarity People don’t need an all-knowing hero. They need: - Clear ownership - Clear priorities - Clear next steps When cross-functional work stalls, it’s almost always a clarity problem, not a competence problem. 3️⃣ Don’t assume. Ask, ask, ask. Most cross-functional conflicts come from: - Hidden constraints - Misaligned incentives - Different definitions of success You only uncover those by asking naïve, direct questions, especially when you think you already know the answer. 🎧 Listen to the full conversation with Nick Palomba (Microsoft) and learn how top leaders make cross-functional work actually work, not just on PowerPoint. 👉 https://lnkd.in/dH62WpAE

176

10 Comments

No A/B test? No problem. Learn how Regression Discontinuity Design (RDD) can uncover causal effects in observational data by exploiting sharp cutoffs. Dive into the theory and a real-world e-commerce example in Alejandro Alvarez Pérez's newest article.

21

2 Comments

💡 The enterprise GenAI hype is loud... but real impact remains rare. A latest AI in the Enterprise 2025 report debunks some persistent myths and brings much-needed clarity on what’s really happening in the field: 1️⃣ Mass layoffs? Limited and highly sector-specific. GenAI’s impact is mostly focused on content creation, support, and analytics. No executive consensus on long-term hiring. 2️⃣ Enterprise adoption? Lots of experimentation, but only 5% of projects make it to production. Leaders want ROI... but struggle to scale. 3️⃣ Biggest blocker? Not the model quality, but poor integration into existing workflows. Most tools don’t adapt to how people actually work. 4️⃣ Custom-built tools? Extremely hard to scale. Without deep workflow integration and continuous learning, internal tools often fail and users switch back to general-purpose tools. 5️⃣ Core business transformation? Not yet. In 7 of 9 sectors, GenAI hasn’t meaningfully reshaped operations. 🧭 What should companies do now? ✅ Focus on a few high-value, operational use cases 🧰 Leverage general-purpose tools for productivity, but continue to protect sensitive data 📈 Scale with workflow-native, learning-enabled tools not broad demos or flashy features 📊 And don’t overlook less flashy investments: predictive AI, business intelligence, and solid data infrastructure often deliver more immediate and reliable value We’ve seen this cycle before with big data 10 years ago. Let’s not fall into the same trap with GenAI and move to another AI Winter. 💬 How are you moving from GenAI experiments to real impact in your organization?

78

13 Comments

If your datasets are poorly labeled, you’re inviting a host of issues like noise, bias, and inconsistencies that can seriously hurt your model's accuracy and reliability when it’s out in the wild. That's why data labeling has evolved from being just a preliminary step into a vital part of engineering and operations in the AI lifecycle.

7

Palantir Technologies recently released a new Foundry update that lets you build directly on your source tables, removing the need for manual syncing. As Ontologize’s Ben Thomas explains, most teams used to have to move data into Foundry before they could start modeling. With this update, you can create ontology object types backed by virtual or iceberg tables, so your data stays where it already lives, whether that’s in Snowflake, S3, or another external system. It’s a small change that removes hours of setup and makes it easier to model directly on your source systems. Try it out, and let us know what you think!  

114

7 Comments

What if all your documents could become instantly searchable through AI? With Palantir Foundry + AIP, it is possible to design and structure a use case demonstrating how: - Complex PDFs (text, images, charts) - Can be transformed into structured datasets - Organized within a business Ontology - And queried using AI (GPT-4.1 or any other model) Simply ask a question and receive an answer grounded in all available documents. This type of solution can be deployed across multiple industries: 🏥 Healthcare: medical research, protocols, clinical reports ⚖️ Legal: contracts, compliance, regulatory documentation 🏭 Industry: technical documentation, maintenance, procedures 🏦 Finance: audit reports, risk analysis, regulatory files 🏢 Large enterprises: internal knowledge bases and document management At Neyo, we design and implement these types of solutions to transform your data into a real decision-making asset. #Palantir #Foundry #AI #Data #Ontology #Neyo

6

What Boards Often Misread as AI Progress I’ve seen boards feel good about AI progress when what they really had was activity. Across manufacturing and CPG environments I’ve worked in, the warning signs were consistent. Plenty of pilots. Plenty of demos. Very little clarity. Progress stalled when: • Ownership was implied • Value was discussed but not measured • Risk was assumed to belong to someone else AI progress is not motion. It is decisions getting better over time. When boards focus on outcomes, ownership, and durability, momentum follows. When they don’t, activity fills the gap. #AIGovernance #BoardLeadership #ExecutiveOversight #DigitalStrategy

2

We're facing more and more questions about the use of synthetic data in market research. I'd recommend this blog post, which really resonated with me. So much of the conversation is "how closely can this replicate human data?" But there are so many other considerations researchers should be thinking about. https://lnkd.in/eSQfSTPj

31

3 Comments

"The next wave of AI success is going to be completely dependent on structured data... Data is the mature, logical, boring middle, but if you want an honest assessment of where all the smart AI is going, look to those mature, logical, boring leaders pushing palatable hypergrowth." https://lnkd.in/gprqvzpz

9

1 Comment

AstroBee generates ontologies...but what is an ontology, and if they're important, why don't you already have one? I've written a short piece on Kimball star schemas, ontologies, and how each relate to Bill Inmon's original approach to data warehousing in the early 90s. Dig in to find an overview of a fascinating technological debate and the re-emergence of an approach to data modeling that's been sleeping for 20 years.

18

3 Comments

Most AI strategies wouldn’t last ten minutes in a cave....I know.....I lead people through them. In my day job, I work in data and AI. In my spare time, I take people underground. They have more in common than you think. Caves reveal: who panics, who leads and who people trust. AI reveals the same: It can optimise. It can predict. It can’t: decide what matters, carry risk, or calm humans. That’s still our job! New Depth and Data series: Going underground article, releases on Monday: “I Lead People Through Caves. That’s Why I Don’t Trust Your AI Strategy.” #DataStrategy #Leadership #DecisionMaking #HumanCenteredAI #DigitalTransformation #TechLeadership

30

5 Comments

Dr. Ramla Jarrar, an MMM expert, shares an important perspective every marketer using MMM should read. The customer journey is complex, and while MMM is a powerful tool for understanding how spend impacts that journey, its limitations are often overlooked and nuances brushed over once results are packaged in polished decks. The tension I often see: marketers need one strong KPI to track large marketing investments, but the complexity of measurement means no single framework provides that holy grail metric. Measurement methodologies are sophisticated and technically robust, yet each has constraints. Triangulating across multiple frameworks—MMM, incrementality testing, attribution, brand tracking—provides better insights, though interpretation remains as much art as science. This requires partnership. Marketers need to understand what each methodology can and cannot deliver. Analytics partners, whether internal teams or external consultancies, have the responsibility to explain assumptions, acknowledge limitations, and help interpret results responsibly. When both sides engage with this complexity honestly, we make better decisions. A question for marketers: where do you find the biggest gaps between marketing performance measurement and the questions you need answered? #MarketingMixModeling #MMM #MarketingAnalytics #MarketingMeasurement #MarketingROI #DataDrivenMarketing #MarketingStrategy

1

Too many industrial AI projects fail because they treat data like a byproduct, not infrastructure. What Radix is doing on top of Cognite Data Fusion flips that mindset. Ontology isn’t a luxury, it’s a requirement for scaling intelligence that works in the real world.

18

Anthropic launches a beta release of "Claude For Excel." Worksheet design, complex error tracing, formula proposals, what-if testing, and external data integration are just a few of the capabilities included. While only in beta testing at this point, Anthropic points the way forward for using AI to enhance human-driven creativity and problem solving in today's fast-paced finance environments. https://lnkd.in/ecGAhp4x

1

https://lnkd.in/gMiv7ZQB Special thanks and sincere appreciation to Vijay Mehrotra and Kara Tucker for publishing a highly favorable review of Why Data Science Projects Fail: The Harsh Realities of Implementing AI and Analytics, Without the Hype, the new book that I co-authored with Evan Shellshear, and my new book The Art of Data Science: A Practitioner's Guide in INFORMS Analytics online magazine. Both of these books provide invaluable insights delivered in the context of real-world case studies in the form of lessons learned, through both failure and success, best practices, and practical tools to increase Analytics, Data Science, and AI project success rates. As the AI hype train risks de-railing due to a dearth of wide spread, significant, tangible business value and economic impact delivery, now is the time to understand the fundamental elements of risk that so often jeopardize these projects and initiatives. Contrary to the instincts of many leaders and practitioners, the primary challenges lie not with the AI technology or mathematics--- but rather "human factors" and "business factors" are the most salient determinants of success and failure. Change management, a focus on the underlying business problems & processes, prioritizing and selecting projects based on the most relevant business problems and opportunities and their commensurate business value potential, overcoming data issues, committed leadership, a culture of getting and doing better, and delivering models from the "sandbox" into production systems, all play a meaningful and vital part in achieving success, and sidestepping failure.

18

It’s only Monday after last week’s MinneAnalytics' #DataTech2025 conference, and I’m already applying concepts from the discussions into my work - proof of how relevant and cutting-edge the topics were. Here's a quick rundown of some of them: MCP-Model Context Protocol: a standardized communication framework designed to enable AI models to interact seamlessly with external tools, APIs, and data sources. Its primary focus is on tool integration, interoperability, and enhancing AI reasoning capabilities by providing real-time access to external data and services. Agent-to-Agent (A2A): an open standard designed to enable seamless communication and collaboration between AI agents. In a world where agents are built using diverse frameworks and by different vendors, A2A provides a common language, breaking down silos and fostering interoperability. AI Agent Development Kit (ADK): a flexible and modular framework for developing and deploying AI agents. It is optimized for Gemini and the Google ecosystem but is model-agnostic and built for compatibility with other frameworks. Developers can use ADK to ground agent and application responses to different data connectors. Another tool that has been a highlight for me lately is Glean and I'm looking forward to attending #GleanGo 2025 tomorrow to learn more on how their work AI platform can put AI to work for enterprises. Join if you’re interested in this space, it's virtual and free! GleanGo 2025 Registration link: https://lnkd.in/gDPSvxtX

11

1 Comment