Jatin Solanki’s Post

Your AI will fail because the foundations aren't there   But what does that even mean?   Here is what you need:   💎 𝐓𝐡𝐞 𝐔𝐧𝐝𝐞𝐫𝐥𝐲𝐢𝐧𝐠 𝐃𝐢𝐫𝐞𝐜𝐭𝐢𝐨𝐧 1) What do you want to accomplish with your data? What value can it create for your business team? 2) What are the use cases to drive that value? How can AI enable that? 3) How do we action those AI use cases? What is the effort and steps required? 4) What is your organisation's current data maturity when it comes to those steps? 🔑 𝐓𝐡𝐞 𝐏𝐫𝐨𝐜𝐞𝐬𝐬𝐞𝐬 & 𝐆𝐨𝐯𝐞𝐫𝐧𝐚𝐧𝐜𝐞 1) Do we have adequate data quality and ways of working between business and data teams to action these AI use cases? Think data governance and operating model 2) How will these AI use cases interface with business and data teams? 3) Do people understand the value that these use cases will create? 4) Who needs to be involved to realise these use cases? Need to think beyond the data science teams here and how it impacts current business processes 5) Are business users bought into the direction? 💻 𝐓𝐡𝐞 𝐓𝐞𝐜𝐡𝐧𝐨𝐥𝐨𝐠𝐲 & 𝐄𝐧𝐠𝐢𝐧𝐞𝐞𝐫𝐢𝐧𝐠 1) Is our tech stack simple enough to be understood by all relevant stakeholders across the business? 2) Do our tools integrate properly from source to consumption? 3) Does our data flow through the platform properly without loss of quality? 4) Do we have the technical and engineering expertise to build AI tools on top of our current architecture without sacrificing quality and resources in necessary areas? 5) Is the ROI there to hire the necessary AI roles (Data Acience, MLOps, Data Engineers) to action these use cases? Have we really quantified the value of AI well enough to understand this? For the sake of the data world, ask yourself these questions before doing AI. You will thank me later Any builds? And if you want more info on all these things, check out my newsletter which has articles on all of this types of things (and will get to the rest later). Link is in the comments! #ai #datastrategy #datafoundations #DylanDecodes

Your AI will fail because the foundations aren't there   But what does that even mean?   Here is what you need:   💎 𝐓𝐡𝐞 𝐔𝐧𝐝𝐞𝐫𝐥𝐲𝐢𝐧𝐠 𝐃𝐢𝐫𝐞𝐜𝐭𝐢𝐨𝐧 1) What do you want to accomplish with your data? What value can it create for your business team? 2) What are the use cases to drive that value? How can AI enable that? 3) How do we action those AI use cases? What is the effort and steps required? 4) What is your organisation's current data maturity when it comes to those steps? 🔑 𝐓𝐡𝐞 𝐏𝐫𝐨𝐜𝐞𝐬𝐬𝐞𝐬 & 𝐆𝐨𝐯𝐞𝐫𝐧𝐚𝐧𝐜𝐞 1) Do we have adequate data quality and ways of working between business and data teams to action these AI use cases? Think data governance and operating model 2) How will these AI use cases interface with business and data teams? 3) Do people understand the value that these use cases will create? 4) Who needs to be involved to realise these use cases? Need to think beyond the data science teams here and how it impacts current business processes 5) Are business users bought into the direction? 💻 𝐓𝐡𝐞 𝐓𝐞𝐜𝐡𝐧𝐨𝐥𝐨𝐠𝐲 & 𝐄𝐧𝐠𝐢𝐧𝐞𝐞𝐫𝐢𝐧𝐠 1) Is our tech stack simple enough to be understood by all relevant stakeholders across the business? 2) Do our tools integrate properly from source to consumption? 3) Does our data flow through the platform properly without loss of quality? 4) Do we have the technical and engineering expertise to build AI tools on top of our current architecture without sacrificing quality and resources in necessary areas? 5) Is the ROI there to hire the necessary AI roles (Data Acience, MLOps, Data Engineers) to action these use cases? Have we really quantified the value of AI well enough to understand this? For the sake of the data world, ask yourself these questions before doing AI. You will thank me later Any builds? And don't say build an AI POC 😉 Follow along for daily data advice and memes by hitting the 🔔 on my profile and commenting away #ai #datastrategy #datafoundations #DylanDecodes

💡 Another brilliant post by Andrea Gioia💡 The picture is worth 1,000 words. 🔥 It's all about the data model 🔥 I have been preaching this for *literally* two decades. Over a year and a half ago, I made the connection between conceptual and logical data models and our ability to mature beyond glorified LLMs/GenAI auto-complete. Very few domain subject matter experts can create conceptual and logical data models. This is not a skill that one casually picks up throughout the course of one's career. It's a specific expertise that requires years of real-world application of obscure data modeling standards to actual use cases to become proficient. (And to call it out explicitly so there is NO ambiguity, we are NOT talking about physical data models, the expertise of information technologists and engineers.) Conceptual and logical data models are THE models that represent domain-specific knowledge. They are the blueprints of the nouns, verbs, and relationships that drive the creation of ALL computable business rules and analytics. For more information on our current ability to model these "nouns" in healthcare, see my recent newsletter here: https://lnkd.in/gCFREUcn #UML #ConceptualDataModes #SemanticInteroperability #DataHarmonization #SemanticDataQuality

Accurate depiction of the challenges of conceptual data modeling

Context is paramount. Building a knowledge model is information architecture. This starts with the people, their workflow, and how that translates to rows and columns. This is how you GenAI. MDClone data engineering with clinical expertise (basically rock stars) combined with our client rockstars enable health systems to build foundations for data literacy, democratization, and value pipeline so that copilots can reduce their tendencies and biases to hallucinate (: #selfservice #ADAMSPlatform #copilot integration #genAI

💫 We must start managing knowledge as a first-class citizen within our information architectures. 👉 In an era where data is the lifeblood of innovation, building modular, distributed, composable, and sustainable socio-technological architectures is essential. Managing data alone is not enough; we must also manage the domain-specific knowledge required to interpret, use, and integrate it effectively to support business strategy objectives. 🗓 This paradigm shift, which integrates advanced knowledge frameworks into data management practices, is the main topic of the talk that Jacopo Aliprandi and I will be giving at the upcoming Big Data LDN Here are some spoilers: 1️⃣ To fully unlock the potential of data products, we need to link them to a knowledge model, making them not only interoperable (for efficiency) but also composable (for efficacy). Distributed approaches to data management, like data mesh, work only if one plus one is greater than two. 2️⃣ To fully unlock the potential of GenAI (i.e., Statistical AI), we need to complement it with human-refined knowledge models (Symbolic AI). GenAI and knowledge models complement each other perfectly. While one is fast but error-prone, the other is slow but reliable. That's exactly the way our brain works as explained in the great book Thinking, Fast and Slow 3️⃣ Last but probably the most important one, treating knowledge as a first-class citizen is not a technical challenge, it's a sociotechnical challenge. Only people can translate their mental models into shared knowledge models. Technology can help, but people and processes come first. 👀 Hope to see you there! #TheDataJoy #dataproducts #knowledgegraphs #genAI

On the importance of marrying generative AI with symbolic systems, such as knowledge graphs with business domain models, taxonomies, ontologies.

The Future of a Data-driven World—Where Are We Heading? Somewhere between new opportunities and unimaginable growth, powered by rich data and fast technology! Yes, that's exactly where we'll be placed in the days ahead. Data-driven technologies are set to play an increasingly central role. Innovations in AI, real-time analytics, and digital twins are shaping a future where data drives decision-making and productivity. Here’s a glimpse into this data-backed future through five key points: 1. AI and Real-time Analytics as Game Changers AI models are advancing towards real-time applications. By 2025, technologies like reinforcement learning and data streaming will allow industries to generate dynamic insights. Finance and logistics already leverage digital twins for real-time scenario testing and instant adaptability. 2. Data as a Product Organizations now treat data as a product, with dedicated teams, governance structures, and continuous development cycles. This approach maximizes data’s value while elevating the role of CIOs in strategizing and executing data initiatives. 3. Flexible Architectures and Emerging Database Types Adopting diverse databases like graph, NoSQL, and time-series improves management of unstructured data. This accelerates AI development and digital tools like customer 360 platforms while enabling intricate relationship mapping via knowledge graphs. 4. Data Privacy and Ethical AI Reliance on data raises privacy and ethical challenges. Future legislation will likely standardize data privacy practices. Companies prioritizing ethical AI and transparent governance will gain trust and lead the way. 5. Empowering Data Literacy A data-driven future demands improved data literacy. Organizations are investing in training to make insights accessible across teams, fostering a data-empowered workforce. To wrap up, the data-driven future promises unprecedented opportunities for innovation and efficiency across industries, but with new responsibilities for ethical and data-literate practices. Data professionals must start working towards these transformations while fostering transparency and user empowerment. Let’s engage in meaningful discussions as to how we drive that transformation. Come and join us in this data revolution! #Data #EthicalAi #AI #dataprivacy

Sometimes, a delayed flight is just what you need to catch up on some quality reading! ✈️📖 I just read McKinsey & Company’s "Path to the Data- and AI-Driven Enterprise of 2030", authored by Dr. Asin Tavakoli, Holger Harreis, Kayvaun Rowshankish, and Michael Bogobowicz. It's a must-read for anyone interested in where AI and data are taking us. Three key takeaways that really stood out: Data everywhere: By 2030, data will be embedded in every process and decision. It’s not just about access - it’s about automation and real-time action. Databricks fits perfectly here with its Unified Data Platform, helping businesses turn raw data into actionable insights with machine learning at scale. Unlocking 'alpha' with AI: The competitive edge will come from customising AI models using proprietary data. Enter #MosaicML - now part of Databricks - which accelerates the development of custom models and LLMs, allowing companies to unlock new levels of productivity and personalisation. Capability pathways: To avoid 'pilot purgatory' and scale AI, it’s all about combining the right tech elements. Databricks’ Lakehouse architecture brings together data engineering, data science, and AI, providing a solid foundation for enterprises to deploy AI across multiple use cases efficiently. Oh, and here’s something fun: Dr. Asin Tavakoli shares a surname with Arsalan Tavakoli, one of Databricks' co-founders! I wonder if expertise in AI and data runs in the Tavakoli name family? 🤔 Let’s say some of the nitty-gritty details about which technology is used and what might have gotten mixed up a bit (love to see MLFlow) 😏. But overall, every forward-looking business leader should pay attention to McKinsey's vision of a fully data-integrated enterprise. . Curious to hear—how are you preparing for the AI-driven future? #ai #data #generativeai #leadership #futureofwork #mckinsey #innovation #ai2030 #digitaltransformation #flightdelaysarentallbad #tavakoliconnection #databricks #mosaicml #databricks

Building Your Path to AI Starts with a Strong Data Strategy – Here’s Why: A solid data strategy is essential for creating impactful Machine Learning (ML) and Artificial Intelligence (AI) models that drive your business forward. Here are some key components to consider when developing a strategy: 🎯 Define Business Goals: Start with clear business objectives. A focused data strategy aligns seamlessly with these goals, paving the way to achieve them with clarity and purpose. 🔐 Establish Data Governance: Implement policies for data quality, privacy, and security. This safeguards your data, ensures regulatory compliance, and builds trust by protecting customer information. 🌐 Design Your Data Blueprint: Structure your data storage and streamline data flow. A well-planned data architecture enables effective decision-making and supports scalable growth. 👌 Ensure Data Accuracy: Reliable, consistent, and complete data is the backbone of effective decision-making. Quality data drives quality outcomes. 🛠️ Select the Right Tools: Choose the best technology and tools to support your data strategy, from storage solutions to analytics platforms. This builds a roadmap that’s actionable and future-proof. A thoughtfully crafted data strategy aligns your data initiatives with business goals, ensuring every effort adds value and drives impact. Ready to discuss how we can help design a data strategy roadmap tailored to your needs? Book a FREE consultation with us today! #AI #ML #DataStrategy #DataScience #DataArchitecture #DataInfrastructure #Analitifi