Peter W.

What if the biggest challenge in leveraging AI and advanced analytics isn't the technology itself, but something far more mundane? Reading Nabeel S. Qureshi’s reflections on his time at Palantir, I was struck by how closely his experiences mirrored challenges I've encountered in the legal tech world. One aspect that stood out was how the biggest hurdle wasn't building the applications themselves, but gaining access to and securely structuring the data needed for those applications. Data integration is notoriously tough. It's not just about wrangling PDFs, countless Excel files (we all know how pervasive those are), or other disparate formats that resist easy analysis. The real obstacle often lies in organizational dynamics. Teams become gatekeepers of critical data sources, justifying their roles by controlling access and providing analyses. This can turn data access into a political and logistical maze. What Palantir realized early on was that these struggles weren't just about red tape—they were rooted in genuine data security concerns. To address this, they baked robust security controls directly into the data integration layer of their platform. Role-based access controls, row-level policies, security markings, audit trails—you name it. By doing so, they not only eased the data access issues but also enhanced overall data security. Implementing their platform made companies' data more secure, not less. This resonates with challenges we face in the legal industry. Law firms are sitting on a goldmine of valuable information within our document management systems like iManage and NetDocs. Yet accessing this data securely and efficiently is often more complicated than we'd like to admit. Organizational silos and stringent security requirements can hinder our ability to leverage this data effectively. This isn't just a legal problem, though—it's a challenge faced by nearly every large organization looking to harness the power of their data. Perhaps it's time we take a page from Palantir's playbook. By integrating robust security measures into our data platforms from the ground up, we can alleviate access concerns while maintaining the highest standards of data protection. It's not the most glamorous part of technology, but it's foundational. Without accessible and well-structured data, advanced analytics and AI-driven insights remain out of reach. How is your organization tackling these data integration challenges? Are there lessons we can learn from other industries?

23

3 Comments

#PET #OWNERSHIP? Reminder of proxy variables Looking at map of pet dog ownership tells… … easily mistaken story. The clear geographical delineation in graph (matching the cold war Iron curtain line) can easily sway you into cultural root cause. 🤦🏻‍♂️ And that would be a great example of the correlation fellacy. Because 25 years after the Iron curtain is down is 2-3 generations of the dogs, long enough time for the political factors to wear off, if they ever had effect. 📖 The real learning of this graph is that there might be hidden factor that is not obvious and still it has profound effect on the actual “target variable”. 🐕 In our dog situation the underlying hidden root cause for the pet ownership is the share of agriculture on total labor market 1.5-2 generations back. Because if you had a farm, you were more likely to get used to living with animals or even needed some (shepherd dogs anybody?). The 2 generations lag is important, because you don’t make pet purchase decisions until you live in your own household and that is roughly, when you start your family (hence when you pass on the generation baton). 🔗 This phenomena is also known as #proxy #parameter and has significant impact also (in positive means) in #MachineLearning. Because proxy parameters are strong predictors, but are insights that markets tend to oversee. So if you know them, proxy parameters give you edge. 💡 Now you know the phenomenon I have a #puzzle for you: It is proven by data in many insurance companies (I worked on one of such studies) that #Zodiac sign is very strong predictor of the #car #accident rate. Yes, really true! Can you guess what it is proxy parameter of?

3

I’m headed to Airflow Summit this week! Want to learn more about the state of the art of Airflow data pipeline development and how data engineers everywhere are building better data products with Airflow. If you’re going, want to catch up in SF, or just discuss the joys of healthy and stable data products, let’s meet up. Ps > the image is from a Dalle prompt to “draw me some healthy pipelines”. Everything looks right on the surface but the reality is broken pipelines, leaks, knobs that don’t seem to do anything, and pipes that run on forever. Data engineers: your jobs are safe lol

25

4 Comments

📖 Weekend Reading 📖 This week we had profiles on two highly influential figures in tech: Palantir Technologies CEO Alex Karp, and Anduril Industries founder Palmer Luckey. They are in some ways very different individuals, but in others they are radically aligned. 1️⃣ The first piece, on Karp, comes courtesy of Maureen Dowd in The New York Times. Karp appears to have an odd background to be leading a defense-oriented technology company. But what allows the German philosophy doctor to steer Palantir is an extremely principled and unapologetic geopolitical stance: “we have a consistently pro-Western view that the West has a superior way of living and organizing itself.” An interesting observation: “I think we’re in an age when nuclear deterrent is actually less effective because the West is very unlikely to use anything like a nuclear bomb, whereas our adversaries might…Where you have technological parity but moral disparity, the actual disparity is much greater than people think.” Interestingly, while Karp is vocal in his criticisms of some of the Democrats’ policies, he is firmly in the Biden / Harris camp (as a voter and donor) and often butts heads against Palantir’s chairman Peter Thiel when it comes to politics. https://lnkd.in/dMz7BWiV 2️⃣ Next up, Jeremy Stern (Tablet Magazine) profiles Palmer Luckey, another eccentric, highly principled, endlessly fascinating individual. Luckey, who created the Oculus Rift at 19 in a camper trailer, favors anime and collecting decommissioned nuclear missile bases over Karp’s tai chi. But like Karp, he is steadfast in his belief that the West must continue to out-innovate its geopolitical foes (Iran, Russia, China). He also has similar views on rebuilding deterrence, an idea that is core to the company’s mission: “Anduril was founded on the idea the US government had to be reformed so that it could buy deterrence again—and that the only way to deter wars is to be so clearly capable of winning them that no rival power tries to harm core US interests through the use of force.” Anduril is also backed by Peter Thiel, and here, while pursuing what appears like the same objective, Luckey finds himself on the opposite end of the political spectrum as Karp, having famously supported Donald Trump since his first real electoral bid in 2016. I'd recommend this article even just to get a more holistic view of what happened during that time, when Luckey was ousted from Facebook. https://lnkd.in/dpRpnGRF #palantir #ai #llm #defensetech #americandynamism #ukraine #anduril #peterthiel #vr #oculus #fintech #payments #upi

8

2 Comments

Databricks disclosed that they are expecting to cross $2.4B in revenue run-rate growing ~65% Y/Y by H1’25 (Jan FYE). This was alongside top-tier metrics, including 80%+ GMs, 140%+ NDR, and a multi-product story where ~50% of customers use 6+ products (incl. their Data Warehouse product which generated $400M in revenue up 4.0x Y/Y). Given their accelerated growth, Databricks is among the few scaled software companies, alongside hyperscalers, capitalizing on the AI boom. Equally noteworthy is their market share gain against competitors. Considering their growth trajectory, we explored various scenarios to determine when Databricks will surpass Snowflake in scale. The question is not if, but when, and we believe it is sooner than most anticipate.

1,010

26 Comments

Interesting that on the MIMIC IV readmission models, the LLMs get closer to the traditional ML models. Perhaps because LLMs are not great at representing numeric values (especially at the model sizes tested) that are informative of the prediction for acute outcomes, whereas readmission relies more on diagnoses and other more linguistically-coded inputs. (Though that doesn't explain the poor performance on MIMIC-III readmission)

3

You know there's some momentum happening if Marty Cagan decides to write about a product topic you've been at work trying to enable for years. Love to see it, and BIG shout out to Stacey Langer for providing her expertise in this article! 🧘‍♀️ Here are a few other pros that came to mind for the product model: 1. The government product model makes product (public) value and risk management possible. Product risks—such as value, desirability, viability, and feasibility—are not risks that traditional project management models can effectively manage. Government agencies interact with the public through the services they offer, but government digital experiences are not always inherently valuable. It’s what these digital services do for users that create individual and public value. Building tech alone ≠ value. 2. Evidence-driven decision-making uses discovery, iterative delivery, and testing with actual end users, where you continuously learn. This enables many things, mainly building with your users while validating the solution direction. It can also enable an incremental funding option specifically for products/services (traditionally called ‘projects’) that are especially complex, high priority, and/or contain many unknowns. 3. Driving technology from a place of vision and strategy measured in outcomes that can align and even influence policy changes. 🤸‍♀️ Some realities in trying to adopt a product model in gov: 1. The word ‘product’ often can be confusing, especially when it doesn’t fully represent what is being built and managed. A lot, if not most, of the technology work happening in government often involves solving service design challenges. So, in gov products, you focus on solving for the end-to-end experience, which can mean processes and non-tech things. 2. Conway’s Law and government systems replicate the organizational structures in agencies, which are often siloed and isolated. This is often represented as 'legacy tech.' Much of the work in 'product' in government is modernizing legacy systems, which requires the same principles as the product model but often a different emphasis on its varied approaches. Not all products or digital services start from the same place; many are not greenfield. 3. Traditional ATO structures can stop the product operating model in its tracks. To overcome this, it is important to integrate the ATO into the product model, utilizing a ‘continuous ATO’ operating model. But see point 2 above. At the end of the day, the product model isn’t a process you learn to implement and then just implement. It’s not just ‘Agile’ or Scrum or DevOps. Those methods can (or sometimes can’t) enable the product model. It's a mentality and discipline for continuously using new inputs to deliver value throughout a product's LIFE. It's a cultural change as much as an operating change, and there is no 'single way' to do it. However, operating from product thinking principles can help us start from the right place.

23

3 Comments

Excellent episode this week from Me, Myself, and AI: “Why Only 10% of Companies Succeed With AI“, with Professor Sam Ransbotham at Boston College Take aways: 1. 🏠Infrastructure (Good Data, a way to connect to it or easily scrape it) is Key 2. 🌓Getting a baseline for your users’ hopes or fears around Gen AI can help you tailor your approach around adoption 3. 🎓 Thanks to more varied tech adoption or lack there of Tech Education is leveraging AI to “meet learners where they are” to help or challenge them more depending their need. I see this as less emphasis around Competition between students too, and more emphasis around deeply learning. Direct Quotes Below: ———————————— “And the first few [for those 10% successful companies] are things that you might, I think we would expect, right? They're got to get their technology house in order. You can't have something like artificial intelligence, complicated machine learning models. If you're basically working on an outdated copy of Excel that is run on a dated PC, right? So there's a certain sort of infrastructural element to that. And also there's talent. You have to have somebody use these tools here. Now, so what we found was that 10 percent, to get to be one of those 10 percents, you got to have some of those basic building blocks in place. And we think of those as talent, infrastructure, and strategy.” “We did a study a few years ago, and we asked people, hey, what do you think about artificial intelligence? Do you hope that it's going to do some of your tasks, or do you fear that it's going to do some of your tasks? 73% of the people said they hoped it would do some of their tasks. 33% said fear. And I think that's where we are. Now, I'm not saying that those numbers are not going to change as we get more general or more knowledge-oriented tools. Maybe those numbers change. But right now, we've got a lot of people doing stuff that they don't want to be doing. And probably not adding value.” “What I mean by that is that we have some people who've gotten super curious about technology and come in and are just amazing and on top of things already, and I'm not sure what I can teach them. On the other hand, we also have some people who seem to be getting further and further. What it means is that that bell curve that we think about teaching to is changing. A bell curve is really nice in class if it's tall, because what that means is it's narrow. Because then I in class can talk about the same topic, and I'm going to bore two or three, and I'm going to lose two or three. But I'm going to hit that sweet spot in the middle. But when that thing spreads out, then I'm more in trouble, because I can still cover the same sliver, but then my tails get bigger, and I lose more and more people either through boredom or not being prepared. I think that's actually what I'm most excited about with artificial intelligence right now because I think it can help with that.”

10

2 Comments

So much fun talking about the power of LLMs with Amazon Bedrock and Anthropic Claude. In summary, the ever-evolving landscape of GenAI, it's becoming clear that there won't be a single LLM (Large Language Model) to rule them all. The rise of application-specific LLMs heralds a new era, where various models will carve out niches to excel in. This diversity opens up a wide playing field for innovation and tailored solutions. When it comes to deploying AI models, striking the right balance in evaluation is paramount. We must weigh factors like cost, speed, and precision to ensure optimal performance and resource utilization. Finding this equilibrium ensures that our GenAI solutions are not just efficient but also effective in real-world scenarios. Anthropic Claude emerges as a frontrunner in the realm of LLMs, distinguished by its exceptional capabilities and groundbreaking advancements. Boasting unparalleled scale, efficiency, and adaptability, Anthropic Claude represents a pinnacle of GenAI innovation. Its ability to seamlessly integrate with Amazon Bedrock amplifies its prowess, providing users with an unparalleled ecosystem for knowledge integration and generation. With state-of-the-art architecture and a commitment to pushing the boundaries of what's possible, Anthropic Claude stands as a testament to the transformative potential of GenAI in shaping the future of technology and beyond. Prompt engineering holds the key to maximizing the potential of applications. Crafting prompts that consider persona, context, tags, and positioning can significantly enhance the model's ability to comprehend and generate relevant responses. This strategic approach empowers AI systems to better serve their intended purposes. In the realm of GenAI architectures, a robust knowledge base is essential. Amazon Bedrock stands out as a key component for architectures like RAG (Retrieval-Augmented Generation), providing a solid foundation to fuel intelligent decision-making and content generation. As we march towards AI integration, bridging the gap with legacy applications is crucial. Agents equipped with open APIs offer a seamless way to facilitate communication and interoperability, ensuring a smooth transition into the AI-driven future. Let's continue to explore, innovate, and collaborate as we navigate the exciting frontier of GenAI together! #aws #LanguageModels #Innovation #TechTrends #AIIntegration #webdevcon #claude #bedrock

68

🚫🐍 How do you hurt your own open-source focused company? Learn from Anaconda, Inc.! The TLDR is that they now demand $600/user/year from ALL ENTITIES with 200+ employees; migrate ASAP to MAMBA which is a drop-in replacement, and remove default channels because they demand money for usage of those too. Non-profit, research, hospitals, universities are all in violation of Anaconda new license. Read more in the link below 👇 https://lnkd.in/ggBJHzK2 😱 I have unfortunately direct knowledge of large research groups and organizations executing mass migrations OUT of conda & miniconda; some recommend MAMBA, some recommend MINIFORGE. 🔥 The TRUST IS SHATTERED: even if Anaconda were to do a complete 180 now, they proved already to be too erratic of a company to be even remotely considered for ANY current or future projects. Mamba and Micromamba are the new DEFAULT. 🤷‍♂️ This is particularly painful to see for me personally because: 1️⃣ We maintain PathML among other things, which is currently using miniconda <= PathML has almost 40,000 downloads today and we are now planning the largest release to-date, v3, entirely *without* miniconda. I am a supporter of open-source and this bait & switch rubbed us the wrong way for sure. 2️⃣ I have personally used anaconda and recommended it to so many people. Now receiving & sending these mass "Please uninstall and replace" emails makes me realize how precarious anaconda position has become. 3️⃣ I liked their products very much. As recently as 2 weeks ago I had recommended another Anaconda product to a colleague who was about to teach a class to hundreds of students. I told them already to pick something else because teaching anaconda means teaching a tool students will not have access to when they join the workforce. And we do not teach bad habits. 🫥 Heads-up to Hugging Face 🤗 and all other companies who do & support open-source: don't do the bait & switch. In exchange for a little revenue-bump that will last a year or two at most, you are annihilating the user community that made you relevant and by doing so you are effectively migrating yourself out of that community, burning all bridges on your way out, and turning many users into active detractors. ⌚ In a recent statement anaconda said they will take until the end of the year to decide how to clarify their license for researchers. The lack of urgency further reinforces everything above ☝️ FYI ⚖️ This is the part of the license that changed "2.1 Organizational Use." at 👉 https://lnkd.in/gp_V4BjS

12

8 Comments

Just wrapped up Shreyas Doshi’s week-long Product Sense course. I highly recommend taking it if you’re in the business of making software products successful. I learned how to think deeply and apply principles that actually drive product success. As an engineer myself, any engineer who works on user facing products would learn a ton from this course. https://lnkd.in/eV2DpDpC

13

1 Comment

💡 Insights from Today's VC Meeting 💡 Had an insightful meeting with a VC today, and it really made me reflect on the core of what we're building at Med. Conversations like these push us to refine our vision and ensure we're on the right track. During the meeting, I was asked some critical questions that every founder should be prepared to answer. So, I decided to share some of the questions that VC could ask during meetings: 1. Who specifically is the initial customer? 2. What is your solution and how does it work? 3. How do you know this problem exists? 4. How do you find customers and how do you reach them? 5. How do you make money? 6. Why do you believe that now is the right time? 7. How do you define your target market and customer segments? 8. Who are the main competitors in your field? 9. What similar solutions exist in other markets? 10. How do you plan to defend your product against competitors? What is your moat? 11. What has been built and shipped? 12. What traction have you achieved so far and what are your key metrics? 13. Can you share some of the learnings since you started? 14. Who is on the team, and what are their unique skills and experiences? 15. How did you meet your co-founder? 16. What are the shares of co-founders? Are they full-time or not? 17. What are your superpowers? 18. How much is the company raising, what are the terms, and who else is involved? 19. How many months of runway do you have? How many months will you extend? 20. Where is the company located and domiciled? 21. What is the monthly burn rate? 22. What does the cap table look like before this investment round and after? 23. What are the key business and product milestones over the next 12-24 months? 24. Why do you do this? 25. What keeps you awake at night? These questions are not just for VCs but are crucial for any founder to continually ask themselves. They help clarify the vision, strategy, and execution plan, ensuring every step is deliberate and impactful. If you're gearing up for your next VC meeting or just refining your pitch, consider these questions as part of your prep. They will help you see your business from multiple angles and be ready for whatever comes your way. #startups #entrepreneurship #VC #healthtech #medtech #fundraising #founders #investment

1

1. IMO, agent patterns are one of the most interesting ways to extend LLMs. memary is based on leveraging a graph database (neo4j) for long-term memory (to get around the limited memory inherent in even the largest context windows in LLMs) and augmenting it with other tools like web search and Computer Vision. https://lnkd.in/gfFcBqJx Looks interesting. 2. More about how consolidation effects the entire healthcare industry: https://lnkd.in/gXJKqNJB The Change hack/ransomware attack not only affected claims, service, refills, etc. but now CMS & NCQA are delaying the entire HEDIS reporting process (used for Medicare/Medicaid quality ratings) due to the impact. 3. NARA bans ChatGPT use agency-wide. https://lnkd.in/gW8T94JU Federal Government agencies see this as a risk, including their information leaking and being used to train models (which may then output the wrong information or show PII/PHI other proprietary information). 4. Oh great, now I can't even be angry anymore :) https://lnkd.in/gcrN9mjx

1

1 Comment

👉🏼 RDguru: A Conversational Intelligent Agent for Rare Diseases 🤓 Jian Yang 👇🏻 https://lnkd.in/eSCm_MkQ 🔍 Focus on data insights: - 📈 RDguru outperforms ChatGPT in generating authoritative descriptions, highlighting its enhanced accuracy in knowledge Q&A tasks. - 🔍 A novel multi-source fusion diagnostic model shows a significant 5.47% improvement in diagnostic recall compared to PheLR. - 🧬 The use of advanced algorithms facilitates more effective automated phenotype annotation and differential diagnosis in rare diseases. 💡 Main outcomes and implications: - 🏥 RDguru demonstrates the potential to significantly improve clinical decision-making processes in rare disease diagnostics. - 💬 The conversational agent can enhance patient interactions through evidence-based Q&A, leading to better-informed medical consultations. - ⚙️ Integration of multi-source diagnostic strategies could pave the way for the development of more robust AI tools in healthcare settings. 📚 Field significance: - 🚀 This study positions RDguru as a promising solution for addressing the challenges of diagnosing rare diseases, particularly given the limitations of traditional methods. - 🌍 The findings may inspire further research into the implementation of LLMs in clinical environments, promoting wider acceptance and use of AI technologies in healthcare. - 🔒 Ensuring accuracy and reliability in AI outputs is crucial for safeguarding patient safety and enhancing trust in AI applications within medicine. 🗄️: [#RareDiseases #AIAssistedDiagnosis #ClinicalDecisionSupport #MachineLearning #HealthTech #NaturalLanguageProcessing #DiagnosticAccuracy]

1

I'm so excited about the problems "traditional" software companies are tackling right now. LLMs let software teams move faster, make sense of larger domains, and built much more complex tech than we were able to even five years ago. I believe the impacts in healthtech, govtech, and fintech will be large. Simultaneously, I'm amazed at all the ways one can lose money investing in AI companies - especially the foundational models. There's no part of this market that isn't trending toward commoditization. This has been true for a while but the release of llama3.1 makes it especially clear: Most workflows can retrain on this open-weight 405B model with no change to their budgets (AWS Bedrock charges *at max* the same as OpenAI) and then find ways to negotiate/optimize that cost way down while not sending their data away to a 3rd party. Even the B2C story is commoditized — what AI wrappers have believable ROI if I can run this one-liner and the hard part is accomplished for a fraction of a cent?: aws bedrock-runtime converse --model-id meta.llama3-1-405b-instruct-v1:0 --messages '[{"role":"user","content":[{"text":"Write me a post enumerating how investments in LLMs may yield no competitive advantage."}]}]' As I look past the flashy parts of AI I see: * Building one-off tools against novel APIs is now (and forever) instant * Analyzing requirements for most tasks and understanding new domains is an order of magnitude faster * GitHub Copilot yearns to write React for us * LLMs give the same boost as pair programming at zero cost I haven't been this excited to build complex products in a long time.

36

1 Comment

I loved seeing the response to the post on data stewards and how key they are in biotech. Can everyone be a data steward? It’s undeniable that having very organized, data-minded individuals on a team sets a company up for success. But how do you scale this? By making it a part of the process. When teams use tools and develop processes that curate data from the very beginning, everyone becomes a data steward. That way, data stewards aren’t a separate role - everyone is responsible. The trick here is boiling it down to the essentials and picking the right tools. That way, the lift on an individual-level isn’t that noticeable, but the gains on the team-level are powerfully compounding. That’s why we’ve spent so much time at Kaleidoscope thinking about the right ways to collect, track, and validate data. All while hiding 99% of that complexity from the end users — to allow teams to curate data at every step of the process, from experiment planning through final analysis, without feeling like they’re doing hard, distracting work. To learn more, head here: https://lnkd.in/eyrUp9Pj And if you or your team have found good ways to empower more data stewardship from within, let me know in the comments!

29

1 Comment

Wrote a post on how we're thinking about semantic modeling at Sigma, specifically focused on a GUI-based model, flexibility, integrations, and (of course) leveraging them for AI. Curious what folks think. Do these points resonate? Am I missing something important? Want to chat more about the topic in general? Let me know! https://lnkd.in/gnqtxnY3

9

Inspired by a side slack chat with my rockstar coworker Rohan Goel earlier today: "The future winners of the data management & governance space or any other software category, will be AI first companies" is too easy. IMHO the winners will be defined the way they've always been defined - by unlocking results and value. HOWEVER GenAI will become (& is fast becoming) non-negotiable infrastructure. It will be what electricity became eventually, what the internet became eventually. In the 90s you could say the winners would be internet-first companies and you'd have been right. But what's even more right is, the future turned out to be companies that used the internet for what is unlocked - distribution. That's the lens Jeff Bezos took. When asked, he said he wants to build the largest book store in the world, in a way that could only uniquely exist on the internet. You didnt need bulky physical book shelves, you needed to design a whole new virtal book world! Similarly, GenAI eliminates a lot of friction of building and doing. It *just does things* for us (given adequate data and instructions ofc). The future is reserved for companies that will ditch pre-conceived notions and unabashedly reimagine every element of friction in the path to value. Not to mention vastly reduced cycle times, building and iterating faster. GenAI is a builder's utopia if people can dream big enough. The future belongs to those who can take GenAI as assumed and build against that. It's a beautiful time to build for value!

42

🌟 Exciting News! 🌟 🚀 Thrilled to announce that our paper, "Beyond One-Size-Fits-All: Adapting Counterfactual Explanations to User Objectives," authored by myself, Jason Liartis, and guided by Professor Giorgos Stamou, has been accepted for presentation at the #HCXAI (Human-Centered Explainable AI) workshop of the #ACM #CHI2024 conference! 📝✨ 🔍 Ever wondered the intricacies of AI transparency and user-centric decision-making? 💭 Our research delves into Counterfactual Explanations (CFEs) to address just that! From healthcare to finance, understanding diverse user objectives is 🔑 to effective AI collaboration. But are current explanations hitting the mark? 🤔 🎯 Join us as we unravel the adaptability of CFEs to user goals and domains! 💡Our position paper offers insights into how the desired properties of CFEs may vary and conflict depending on user objectives, setting the ground for a more nuanced examination and utilization of CFEs! 📚 Dive into our position paper to explore this critical topic: https://lnkd.in/de9aS2zg 🔍 Curious to learn more? Check out #HCXAI #CHI2024 for updates! 🔗 Workshop site: https://lnkd.in/d6S5Admp 🔗 Conference site: https://chi2024.acm.org/ 🌐 Artificial Intelligence and Learning Systems Laboratory 🏛 National Technical University of Athens Spread the word and dive into the future of AI explanations with us! 🚀✨

74