Robert Hernandez’s Post

"Isn't AI just ChatGPT, those artsy image makers, and some clever folks crunching numbers in basements?" 🤔 Not quite. Enterprise #AI is more layered than that, and for good reason: ✔️ Business Layer: No fancy tech without a clear purpose. Value creation, KPIs, governance. Simple as that. Your AI must solve real problems and deliver measurable results. ✔️ Solution Layer: Where business meets tech. Use case prioritisation, architecture decisions, integration points. Get this wrong, and you'll build impressive demos that fail in production. ✔️ AI/ML Layer: The engine room. Model development, testing, monitoring. Yes, you might use GPT or stable diffusion - but also regression models, clustering, or custom neural nets. Horses for courses. ✔️ Data Layer: Quality data, clean pipelines, proper governance. Without these, you're building on quicksand. And no, synthetic data isn't always the answer. ✔️ Platform Layer: Infrastructure, security, scalability. Boring? Perhaps. Essential? Absolutely. Enterprise AI needs industrial-grade foundations. This isn't about complexity, for complexity's sake. Each layer addresses specific challenges we've learned the hard way. ⚙️ What's your take? Which layer poses the biggest challenge in your organisation? #EnterpriseAI #AIArchitecture #DigitalTransformation

Had an amazing time in London with one of my friends doing incredible work in AI. The biggest business problem in the world is that 97% of organizations know that they need to go full speed ahead when it comes to AI… and yet only 14% actually have the capacity to make that happen. That's where our team comes in. The majority of investment in AI is currently going into consumption of #GenerativeAI (ie ChatGPT or CoPilot) but when it comes to productization delivery, many times organizations simply do not have the clarity of strategic prioritization, the data science team and resources to deliver. It’s about impact. Keep it simple: 1. Set your strategy. 2. Get your data ready. 3. Go get a quick win. Then rinse and repeat — and level up. ⬆️ Our clients are often astounded what they can achieve in 80 days. We’re happy to help… and if you’d like to learn more about how we can help your organization with an AI strategy, you can message me directly. #artificialintelligence #GenAI #AIStrategy #AI is Vaital.

The AI landscape has evolved significantly since the launch of ChatGPT in 2022, and with that, we’ve seen a noticeable shift in how AI is developed and shared. Before, the spirit of open-source innovation drove AI research. Researchers freely exchanged ideas through platforms like arXiv. The community collectively advanced technologies like transformer architectures and deep learning frameworks. But today, as companies race to capitalize on AI, transparency is taking a backseat. The commercial focus has led to AI models being locked behind intellectual property protections, making them less accessible to the broader community. This lack of openness raises questions about fairness, accountability, and the true intentions behind these technologies. Transparency and democratization are crucial, especially in fields like healthcare, where AI’s decisions can have life-altering implications. When we open up AI architecture and training data, like what we’ve seen with projects like BLOOM and Stable Diffusion, we invite a broader range of voices into the conversation. This reduces monopolistic control and ensures that AI tools meet the diverse needs of all communities. Ultimately, AI is a powerful tool that, if appropriately managed, can drive tremendous positive change. But to do so, we need more open-source initiatives and fewer secrets hidden behind corporate walls. How are you ensuring transparency and compliance in your AI projects? Let’s discuss how we can collectively push for a more open and fair AI landscape. #GenerativeAI #AI #AIInnovation #DigitalTransformation #EnterpriseAI #TechLeadership

Everyone talks about AI. In fact, we also use #AI extensively in #VTDocs. But before delving into the nuances of AI and related technologies, it's important to understand the business context and the challenges it addresses. Let’s consider an example: When reviewing multiple documents, business analysts, project managers, and other professionals typically examine them for essential requirements: They identify dependencies within a document—perhaps paragraph three relates to paragraph 52 on page 19. These interdependencies occur both within individual documents and across multiple ones. Examples include standard operating procedures with specific policies, terms and conditions, or supplier agreements. How does their process look like today? The analyst opens a document and skims through it from start to finish, noting any critical points or elements of interest. They use the search function (Ctrl + F) to scan for terms like "warranty" or "indemnifications" in the contract. This method is time-consuming and ad-hoc. While experienced professionals might navigate quickly, those less experienced or new to the business need much more time and might overlook crucial details. VT Docs and its AI models offer a different approach: When uploading multiple documents into VT Docs, users instantly see common themes across documents – powered by AI. Imagine a matrix where each column represents a document, and on the left, thematic frequencies are highlighted. This visualization allows users to spot patterns quickly. For instance, if references to "indemnification" or variations of that appear across several documents, then VT Docs shows them automatically and provides an aggregated view. VT Docs utilizes an AI model specially designed for business and technical communications. So, there's no "hallucination" or inaccuracy— the AI doesn't use a generative model (like ChatGPT) but relies on a model that can trace back to the precise text, ensuring 100% accuracy. This precision is non-negotiable and pivotal. Another consideration, especially concerning generative AI, is security and data integrity. VT Docs and its AI model run completely isolated in the most secure locked-down environments. This ensures no data leaks or unauthorized learning from your data. AI has the potential to transform industries, but only if it maintains reliability and security and seamlessly integrates into existing workflows. At VisibleThread, we aim to pioneer solutions that embody these principles, ensuring not just innovation but secure and dependable results. #documentreview #artificialintelligence #enterprise

Why Adopting Gen AI is So Difficult https://lnkd.in/exgez_7v HBR reports many of the leaders they've interviewed report struggles with Generative AI. This is due to three primary concerns: 1️⃣ Most businesses haven't fully adopted predictive AI and traditional Machine Learning models. Lack of familiarity and capability to use data and models to drive value transfers over when they attempt to do similar things with the cutting edge Large Language Models. 2️⃣The complexity and expense of LLMs is greater than what could be achieved with simpler models. Despite the magic of ChatGPT, AI doesn't solve all problems auomatically. 3️⃣Massive uncertainty around the security, long-term costs, regulation, and impact on the job force. Despite these challenges, organizations are pressing forward by bringing in the needed expertise, rationalizing their analytics processes, choosing high-impact use cases, and building resilient AI systems.

Tony Blair's thinktank sought insights from ChatGPT on the potential impact of AI on public sector jobs. The analysis has drawn criticism for its perceived unreliability. The discussion around AI's influence on employment ranges from optimistic views that it will create new jobs to concerns about businesses replacing entire workforces with AI tools. The report suggests that over 40% of public-sector tasks could be partially automated by AI-based software and hardware. It also highlights the need for government investment in AI technology, data system upgrades, workforce training, and covering redundancy costs. The estimated annual costs for these initiatives could reach £4bn per year over the parliamentary term under an ambitious rollout scheme.

To successfully use AI, companies can take a two-step approach: 1️⃣ Productivity Tools: Start with simple AI tools, like ChatGPT or Microsoft Copilot, to help employees get familiar with AI. These tools make daily tasks easier and help build comfort with AI. 2️⃣ Impact-Driven AI Solutions: Once teams are comfortable, move to powerful AI solutions that can reshape business processes and boost profits on a larger scale. Read more: https://lnkd.in/d7GU34_T

They do RAG and what they do may ultimately become more important than ChatGPT. Cohere differentiates itself by offering models that can be deployed in various environments, bringing models to data rather than requiring data to be brought exclusively to a specific platform like Azure. Optimized for enterprise generative AI, search and discovery, and advanced retrieval the models are designed to augment and elevate the global workforce, so businesses can thrive and stay competitive in the AI era. How they differ from consumer focused AI? Command models are used by companies to build production-ready, scalable and efficient AI-powered applications. The combination of Rerank and Embed delivers more reliable and up-to-date responses for most application, grounding retrieval-augmented generation (RAG) in the data. Their Generative AI and Advanced Retrieval models work seamlessly together for applications requiring RAG. It is easily possible to connect and retrieve information from documents and enterprise data sources to build more powerful AI applications. I will monitor the progress carefully and if there is an IPO.....Needless to say this is not a European company, there are no offices inside the EU bloc.

💡 Prompt engineering is the process of carefully crafting inputs, or "prompts," that are provided to AI models like ChatGPT to generate useful, relevant, and accurate responses. The goal of prompt engineering is to guide the AI toward producing the most desired output by framing the question or instruction in a clear and precise way. 📌 1st Level: Basic questions, clear and specific. 📌  2nd Level: Contextual prompts for more detailed answers. 📌  3rd Level: Advanced, multi-step instructions for complex tasks. Key to Using AI at 1st, 2nd, and 3rd Levels: 1st Level: Basic Prompting At this level, you're giving simple instructions or asking straightforward questions to get basic responses from the AI. The key here is clarity. 2nd Level: Contextual Prompting Here, you provide additional context or detail to help the AI understand the nuances of the task or question. The AI can now respond with more complexity, drawing on patterns and concepts beyond just surface facts. The key here is contextualization. 3rd Level: Advanced Prompting At this stage, you're using highly structured prompts or a series of chained prompts to guide the AI through complex tasks. The key here is optimization—leveraging the AI’s full capabilities to handle multi-step tasks, generate creative solutions, or perform intricate analysis.

Building AI features is high risk, high reward. AI holds immense promise and potential, but user retention and satisfaction can suffer when that promise isn’t met. At Sprig, we’re all about reducing risk in AI product development by creating data proof of concepts (POC) before writing any code. Here's the approach that we've been refining over the past two years: #1: Start with ChatGPT! Upload test data sets and try various prompts to see how accurate and useful the results are. 💬 #2: Move on to testing prompts and test data sets directly with the GPT-4 API. ChatGPT performs better than the GPT-4 API because its prompts and outputs are fine-tuned. #3: Test your prompts with a wide range of customer use cases and data sets—small vs large, low-quality vs high-quality, etc. Continue to evaluate AI performance. #4: Build an MVP of the feature within your product, feature flagged to internal employees. Meet with customers in 1:1 interviews, have them try the feature, and see if the AI capabilities meet their expectations. #5: Complete the product development process and launch to customers! 🚀 By taking these steps, we de-risk AI feature development and ensure every AI feature meets or exceeds customer expectations! 🌟 How has your experience been developing with AI? Share in the comments below. 👇