How Unstructured Data Drives Workplace Innovation

AI’s ability to unlock insights from unstructured data is a massive breakthrough for businesses. I have been beating this drum for a while now. But the real magic? It happens when you combine structured and unstructured data. Here’s why. AI made it possible to ask questions of structured data, like company records, contact records and deal status, and get answers back in natural language. That was a breakthrough. Now, it is possible to ask evergreen questions of unstructured data, like emails, calls, video conferences, transcripts of meetings, and get real-time insights, also in natural language. That is another breakthrough. An even bigger one. But businesses don’t just need breakthroughs. They need results. And to get them, they need insights from both structured and unstructured data—working together. Let’s make it real with an example. Picture a sales leader getting a live feed of every time a competitor is mentioned in sales calls. Even better? AI identifies the salesperson who’s best at handling those objections. That’s unstructured data in action to deliver insights. But there are deeper questions they want to answer, like: Is there a competitor we consistently lose to? Is a new competitor suddenly appearing in deals in specific regions? To answer those questions, they need structured data. They need to cross-check their list of competitors with closed-lost and closed-won reports and pipeline trends by region. Now, they don’t just see what’s happening—they know which competitors to worry about and what messaging works best against them. That’s not just a useful insight—it’s a game-changing one. A smart sales leader won’t stop at knowing which competitor is a threat. They’ll turn that insight into action—launching targeted email campaigns, updating sales playbooks, and creating competitive content. But here’s the catch: AI-powered insights are only valuable if they’re accurate, governed, and respects permissions. AI has opened up a world of new possibilities. The question then becomes: How can businesses turn those possibilities into results? It is by unifying structured and unstructured data with the right context and governance to drive faster action. That's the key to unlocking AI's potential to help businesses grow! And that gets us excited everyday!

For decades, organisations have managed their data in two separate worlds. On one side is structured data - numbers, categories, and neatly organised information - stored safely in databases and easily processed by machines. On the other side is unstructured data - the rich, nuanced content buried in emails, chat logs, documents, images, and social media comments - largely out of reach for computers. 🔵 LLMs Changed The Game: LLMs can now sift through mountains of text to uncover insights and connections, understanding sentiment, context, and relationships in ways that were previously impossible. Suddenly, unstructured data can be treated as if it were structured. But traditional tabular databases are too rigid to handle the complex, nuanced relationships revealed in this data. 🔵 Knowledge Graphs Structure Complex Data: This is where knowledge graphs come in. They offer a more flexible and expressive way to structure data, capable of modelling complex networks of information. With knowledge graphs, you can transform unstructured text into triples - subject > predicate > object - and these triples together form a graph that connects your data in a meaningful, machine-readable way. 🔵 Bridging Structured and Unstructured Worlds:  But extracting insights isn’t enough. The real power lies in weaving those insights back into your core business systems. You don’t want to discard the well-structured data you’ve carefully curated in databases over the years. The opportunity is in linking the two together - integrating structured data points with insights mined from unstructured content. You can treat your tabular data as a graph as well, mapping the rows and columns into triples. This is what we knowledge graph folk have been doing for years. 🔵 The Power of URLs: Imagine every client, product, or asset in your organisation having a unique URL identifier - like a web address, but for an entity in your data. Whether they appear in a database, an email, or a customer support chat, every reference points back to the same URL, giving you a single source of truth across all systems. Even better, if you want to link two entities together, you can simply use their URLs - subject URL > predicate > object URL - it’s as straightforward as adding a hyperlink to a webpage! 🔵 This Is a Strategic Shift in Thinking: This isn’t just about tidying up your data infrastructure. It’s about making a strategic shift to unlock new capabilities. Patterns emerge. Redundancies disappear. Decision-making becomes faster, more precise, and better informed. you are ready for the Age of AI. ⭕ What is a Triple: https://lnkd.in/e-hr5eQK ⭕ What is a Knowledge Graph: https://lnkd.in/eG8DhxVn

Introducing Docs2KG: A New Era in Knowledge Graph Construction from Unstructured Data ... Did you know that 80% of enterprise data resides in unstructured formats? This makes it incredibly challenging to extract meaningful information and gain insights ... 🤔 Addressing the Challenge of Unstructured Data A recent research paper introduces Docs2KG, a novel framework for constructing unified knowledge graphs from heterogeneous and unstructured data sources like emails, web pages, PDFs, and Excel files. The key innovations include: 1. Flexible and dynamic knowledge graph construction that adapts to various document structures and content types, unlike existing approaches limited to specific domains or schemas. 2. A dual-path data processing strategy combining deep learning document layout analysis and markdown parsing to maximize coverage of different document formats. 3. Integration of multimodal data (text, tables, images) into a unified knowledge graph representation with structural and semantic relationships. 4. Facilitation of real-world applications like reducing outdated knowledge in language models and enabling retrieval-augmented generation. 5. Open-source availability encouraging further research and development. 💪 Strengths: - Addresses the crucial challenge of extracting insights from the vast amounts of unstructured enterprise data residing in data lakes. - Offers flexibility and extensibility to handle diverse document types across industries. - Leverages advanced AI/ML techniques for document understanding and information extraction. - Unified knowledge graph representation enhances data integration, querying, and exploration capabilities. - Open-source nature promotes collaboration and accelerates innovation. 👉 Potential Limitations: - Performance may vary based on the complexity and quality of input documents. - Integrating information across highly heterogeneous sources could be challenging. - Maintenance and updating of the knowledge graph as new data arrives needs to be addressed. 👉 Opportunities: - Enhance enterprise knowledge management and decision-making processes. - Enable new AI applications by providing structured, integrated data to train language models. - Extend the framework to support additional document types or modalities. - Explore domain-specific customizations or industry-focused solutions. 👉 Risks: - Adoption may be hindered if the system cannot handle proprietary or highly specialized document formats. - Data privacy and security concerns need to be carefully addressed, especially for sensitive information. - Reliance on external open-source libraries and models could introduce vulnerabilities or dependencies.

Being data-driven is often viewed as mastering measurement and optimization—but don't leave discovery and innovation on the table! When it comes to data, an organization's first impulse is to chase certainty, relying on dashboards, precision KPIs, and refined datasets. This is an important efficiency boost, but it's important to keep in mind that breakthroughs and new business models rarely result from meticulous planning. They emerge when someone recognizes an unusual pattern or an overlooked anomaly. This accidental brilliance is precisely what modern data-driven organizations must foster in addition to their hunt for efficiency. When it comes to their use of data, most companies aren't structured for serendipity. They operate in cycles of predictability, continuously refining data to meet expectations. While this optimization generates immediate efficiency gains, it often follows the economic principle of diminishing returns—each incremental improvement costs a bit more and delivers a bit less. Genuine data-driven innovation requires spaces for "curated chaos": environments intentionally designed to surface unexpected findings. Perhaps paradoxically, this demands a high level of data maturity—robust capabilities that create a stable foundation from which exploration can safely occur. Innovation and a data-driven mindset build on the same foundation. Both require intellectual bravery, eye-to-eye interaction across hierarchies, and patience to detect subtle signals. Curated chaos isn't a call to abandon rigor; it's creating spaces where overlooked connections can naturally emerge. It means deploying analytics not merely for measurements and predictions, but as exploratory instruments—provoking questions and challenging assumptions. The most innovative data-driven companies embody such structured curiosity. They balance analytical discipline with openness to surprise. They reward thoughtful questioning as vigorously as decisive answers and recognize that breakthroughs often appear quietly within noise. While optimization often provides the comfort of predictability and quantifiable returns, discovery operates on a different economic model where small investments in exploration can yield disproportionate value. While your competitors perfect their dashboards, consider what they might be missing—the next crucial insight might not be hiding in the cleanest dataset, but in the anomalies you've initially aimed to get rid of. Don’t just optimize with your data—explore it!

The first iterations of enterprise software made billions organizing data into neat boxes. But "real" business doesn't happen in spreadsheet cells. Hear me out: Look at Salesforce—a $35B business built on a simple premise: sales reps manually log their activities into predefined fields. But here's what actually happens: A rep spends two hours in a technical deep dive with a prospect. They uncover critical requirements, timing needs, and competitive dynamics. Then in the CRM they check a box marked “Technical review completed” 🤦🏻♀️ What we know about enterprise software: → Critical business data lives in unstructured conversations → Sales activities happen in emails, calls, and meetings first → Rich customer interactions get reduced to basic data points → Systems can't generate accurate predictions without complete context → Most valuable business intelligence never makes it into our tools As a result, 80% of valuable business data remained trapped in unstructured formats (trapped in emails, documents, and conversations rather than organized in databases). The step-change improvement comes from systems processing unstructured data at scale—software that captures every customer interaction, understands context, and initiates action without manual input. Technical founders building these systems stand to capture massive enterprise value. Salesforce built a $35B business on digital organization, but Service-as-Software powered by Systems of Agents fundamentally changes the game. Companies that can tap into the $4.6T workforce spend market will make today's software giants look small in comparison.

Many enterprise businesses still follow archaic, rigid data strategies. They waste countless hours manually correlating siloed sets, losing all meaning in between. Yes, dealing with unstructured data poses immense challenges. But in an age where even your newest competitor can quickly commoditize that data to your disadvantage, sticking your head in the sand is no longer an option. So, how do you make sense of the unstructured data beast without getting consumed? Here are four must-have steps: 1. Aggressively collect structured and unstructured data from all corners of your organization. This raw material fuels the next phase. 2. Organize the aggregation using flexible storage like data lakes and hubs. Avoid caging information prematurely—allow insights to emerge organically. 3. Analyze through text mining, natural language processing and machine learning algorithms. Derive connections and patterns even your best data scientists would miss. 4. Decide on actions based on consumptive facts, not opinions or hunches. Ground every strategic choice in 360-degree intelligence distilled from once-scattered data pools. Stop dreading data. Start understanding it. The business that can efficiently liberate insights from limitless sources of information will write the next chapter of industry disruption. Will you lead the charge?

The AI companies that crack the ‘context’ problem will win. For decades people have been saying ‘data is gold’. But they were missing half the picture. Data without context is useless. And context, well it lives everywhere - in Slack messages, Notion documents, and Zoom calls. We as humans are able to process, consolidate, and reason through all this knowledge in our heads, and analyze data within the context of all of this information, but historically this was not possible for machines. That's why data analysts were impossible to replace even with the best ML models. But then came the LLMs. And suddenly, everything changed. LLMs unlocked the ability for machines to imitate our ability to understand text (via semantic search), which means it can now extract and make use of context from those Slack messages, Notion documents, and Zoom transcripts. Crazy stuff. Now there's a race for AI companies to figure out how to use all of this contextual data effectively, so they can finally deliver the promise of replacing human employees with their AI products. But it's easier said than done. This valuable contextual data is unstructured by nature (doesn't fit in rows and columns) and lives across hundreds of different apps and file types, making it much harder to work with than the traditional structured data that we’re all used to. Not only that, the process of storing and retrieving this unstructured contextual data (RAG) introduces a lot of complexities around permissions and access - a topic for another post. We've learned a lot about these unstructured data challenges over the past year as more AI companies have started using Paragon, so I decided to centralize some of these learnings in case you’re on that same path. Give it a read below!

𝐒𝐨𝐦𝐞 𝐨𝐟 𝐭𝐡𝐞 𝐦𝐨𝐬𝐭 𝐯𝐚𝐥𝐮𝐚𝐛𝐥𝐞 𝐢𝐧𝐬𝐢𝐠𝐡𝐭𝐬 𝐢𝐧 𝐚 𝐜𝐨𝐦𝐩𝐚𝐧𝐲 𝐝𝐨𝐧’𝐭 𝐜𝐨𝐦𝐞 𝐟𝐫𝐨𝐦 𝐝𝐚𝐬𝐡𝐛𝐨𝐚𝐫𝐝𝐬. 𝐓𝐡𝐞𝐲 𝐜𝐨𝐦𝐞 𝐟𝐫𝐨𝐦 𝐜𝐨𝐧𝐯𝐞𝐫𝐬𝐚𝐭𝐢𝐨𝐧𝐬. Yet, most of these go unheard.   Dashboards are great at showing what happened. But if you want to understand why, you need to listen.   Calls, chats, and support tickets carry early signals—about customer sentiment, product gaps, and friction points. Yet too often, those signals get buried in transcripts or dashboards that only tell part of the story.   I recently explored how brands are increasingly leveraging AI to unlock powerful customer signals—and it’s not just the tech giants leading the way. Ulta Beauty is a standout example, driving 95% of its sales from returning customers through AI-powered personalization, tailoring experiences that keep shoppers coming back read more. Meanwhile, IndiGo Airlines has scaled its customer support dramatically, with an AI bot managing over 42 million messages across channels, ensuring fast, consistent service at scale read more.   What’s the common thread? These brands aren’t using AI to replace people—they’re using it to listen better, respond faster, and act smarter.   AI helps us hear more, faster. But meaning comes from how we respond. If we’re not learning from what customers say, we’re missing the point.   If you’re thinking about how to make unstructured data work harder for you, happy to share what I’ve seen. Sometimes, a short conversation about conversations can change everything.   #CustomerExperience #AI #Leadership #ContinuousImprovement #CXStrategy #AugmentedAI

The most valuable dataset in capital and commodities markets isn’t in your data warehouse. ⮑ It’s what's in your chat. If you strip markets down to their core, they’re not defined by screens, matching engines, or protocols. They’re defined by communication - the ability of buyers and sellers to exchange information, intentions and ideas. Yet in today’s highly electronified environment, the richest form of communication remains the least used: the daily stream of chat between clients, sales and traders. Every day, hundreds of millions of words move across platforms like Bloomberg IB, Symphony, ICE Chat, LSEG Messenger, WhatsApp, and internal messaging systems like Teams and Slack. If you sit on a trading floor, you know that many of the earliest and most meaningful signals - interest, hesitation, sentiment, conviction - show up in conversation well before they appear in a price or print. But while we’ve digitised nearly every other part of the market, this conversational flow sits largely unstructured, uncaptured, and underutilised. It isn’t that firms don’t recognise its value - it’s that historically the technology simply didn’t exist to process messy, unstructured shorthand and jargon heavy, conversational dialogue at scale. That’s now changing. We’re entering a moment where unstructured chat data can be captured, analysed and understood with far greater accuracy than most people realise. Once you start seeing conversations as data - as well as workflow - the implications for market structure, liquidity discovery and trading strategy are significant. This shift may be far bigger than people expect...

I spoke with 4 data leaders this week who all asked a version of the same question: “𝗪𝗵𝗮𝘁 𝗱𝗼 𝗱𝗮𝘁𝗮 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝘀 𝗹𝗼𝗼𝗸 𝗹𝗶𝗸𝗲 𝘄𝗵𝗲𝗻 𝗱𝗲𝗮𝗹𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝘂𝗻𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲𝗱 𝗱𝗮𝘁𝗮?” In the structured world, data products are clean, governed tables — consumed mostly by AI/ML practitioners. Unstructured data is different:   1. It can be represented in any number of structured forms.   2. Its quality dimensions are less mature.   3. It’s consumed daily by almost everyone in an organization. --- That’s why we’re shifting the conversation from 𝗗𝗮𝘁𝗮 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝘀 → 𝗞𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝘀. A knowledge product could range from: 🤖 A reusable set of high-quality files that sit in a vector DB, pre-enriched with semantic metadata for AI engineers, 📁 Or simply enriching SharePoint with business metadata so an analyst can quickly find their desired document. In other words: 𝘬𝘯𝘰𝘸𝘭𝘦𝘥𝘨𝘦 𝘦𝘹𝘪𝘴𝘵𝘴 𝘦𝘷𝘦𝘳𝘺𝘸𝘩𝘦𝘳𝘦. To build a true foundation for unstructured data, companies need systematic methods of enrichment and filtering that connect into their core data lakes, catalogs, and AI platforms. With Agentic workflows, consistent metadata standards across systems will be key — making domain taxonomies (and related structures like ontologies) the core data assets needing governance. Four dimensions will likely play into the creation of knowledge products: ✅  𝗦𝗲𝗻𝘀𝗶𝘁𝗶𝘃𝗲 𝗰𝗼𝗻𝘁𝗲𝗻𝘁 – Can I use this file? ✅  𝗗𝗮𝘁𝗮 𝗾𝘂𝗮𝗹𝗶𝘁𝘆 – Should I use this file? ✅  𝗕𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗺𝗲𝘁𝗮𝗱𝗮𝘁𝗮 – Helping humans find the right info. ✅  𝗦𝗲𝗺𝗮𝗻𝘁𝗶𝗰 𝗺𝗲𝘁𝗮𝗱𝗮𝘁𝗮 – Helping agents find the right info. The process isn’t linear, but the funnel below can be a helpful visualization. Curious — how are you defining data products for unstructured data? #unstructureddata #dataproducts #automatemetadata