Open Innovation Models

🔒 The Slow Death of the Closed EV Charging Network 📉 The old model of closed, locked-in EV charging networks is dying. Slowly, stubbornly...but inevitably. For years, some operators have clung to a walled-garden approach: proprietary hardware, restricted software, and a closed ecosystem designed to trap customers. It worked for a while. But times have changed. 🔌 Drivers demand choice. Nobody wants five different apps just to charge their car. Open networks, interoperability, and seamless roaming are becoming the norm. ⚡ Hardware should be flexible. Why limit site owners to one brand of chargers when open protocols (OCPP) allow for freedom of choice? Charging should be about reliability and experience, not vendor lock-in. 📉 The market is moving on. Governments and industry bodies are pushing for open standards. Networks that resist will slowly fade into irrelevance, overshadowed by those that embrace open ecosystems. The shift is clear: openness wins. The networks that get this will thrive. The ones that don’t? They’ll be footnotes in EV history. The real question is, how long will the holdouts keep resisting before they fade into irrelevance? #evcharging #opennetworks #evrevolution

As of yesterday, operators of public EV charging infrastructure in the European Union must share both static (where are chargers, what plugs, power levels and vehicle types are supported?) and dynamic (are chargers operational, occupied, what's the price?) data as open data. For the time being, these data will be made available in various forms at national access points, but will eventually converge into single national and European (2027) APIs - meaning it'll be much easier to include this information in whatever system you want to use it for. (The same requirements apply to hydrogen refuelling stations - and the few remaining H2 drivers will be pleased to know that refuelling stations must now report when they're running low, so you know when it's best to fill up to avoid being stranded for several weeks. However, given the small number of H2 cars, it might be better to organise an old-fashioned phone ring rather than building apps. A welcome data field not yet included would be the end of the initial subsidy period of an H2 station, so that drivers know when they are going offline indefinitely). A very important step in improving the driver experience, driving the charging business and also very useful for grid and charging network planning. Recent research shows the benefits of making this information freely available, and if the availability is reliable and trusted by (future) EV drivers, the EV fleet will grow faster: "If universal real-time data is accompanied by improved charger uptime and driver confidence in the accuracy of the real-time data, we predict that the EV share of new vehicle sales would grow by 8.0 percentage points in 2030, expanding the EV fleet by 13.2%, and reducing 2030 carbon emissions by 22.5 mmt, versus baseline projections for 2030."

American Enterprise Institute and The Brookings Institution just released our policy paper on how states can lead the EV transition with data transparency. The solution is simple and cheap.  States can require highway fast chargers share real-time data in an open format. Real-time status and price transparency let drivers reliably locate a working, affordable plug, slashing charging anxiety—the top cited barrier to EV purchases. The report includes turnkey legislation that states can adopt to encourage EV adoption while costing them virtually nothing. https://lnkd.in/eiDi57JP  #EV #charging

The administration released its AI Action Plan today - making a strong endorsement of open source AI and open weights (page 4). I encourage you to read it and understand the critical role open source will play in the future of our country's AI strategy. Don't discount the role that efforts like OpenFL will have in bringing federated learning to government and helping them partner with industry to build the best models available. "Open-source and open-weight AI models are made freely available by developers for anyone in the world to download and modify. Models distributed this way have unique value for innovation because startups can use them flexibly without being dependent on a closed model provider. They also benefit commercial and government adoption of AI because many businesses and governments have sensitive data that they cannot send to closed model vendors. And they are essential for academic research, which often relies on access to the weights and training data of a model to perform scientifically rigorous experiments. We need to ensure America has leading open models founded on American values. Opensource and open-weight models could become global standards in some areas of business and in academic research worldwide. For that reason, they also have geostrategic value. While the decision of whether and how to release an open or closed model is fundamentally up to the developer, the Federal government should create a supportive environment for open models." https://lnkd.in/eHDvT8r3

Many EV charging solutions are developing branded experiences in today's market. While innovation is valuable, research suggests most customers prioritize simplicity and convenience above all—similar to what they expect from gas stations. What customers consistently value: ✅ Seamless access without multiple apps ✅ Widespread charger availability ✅ Transparent, competitive pricing For the EV industry to reach its full potential, we recommend focusing on these customer priorities: → Supporting open standards for universal vehicle compatibility → Expanding availability at familiar locations like gas stations and parking lots → Displaying clear pricing information directly on chargers → Developing unified payment systems as intuitive as paying for gas Brands can certainly add value to the charging experience, but addressing these fundamental customer needs will likely drive the strongest adoption and industry growth. The most successful charging networks will be those that balance brand innovation with the convenience customers expect.

DeepSeek: What does it mean for those of us focused on the apps layer? Open weights models like DeepSeek are revolutionizing the AI landscape, offering a counterbalance to excessive profit-taking by dominant providers. Embracing open weights ensures independence from a single provider, fostering innovation and diversity in the AI space. DeepSeek's impact underscores the importance of maintaining a model-agnostic architecture, enabling seamless integration of various AI models into existing systems. This flexibility allows for efficiency gains and future-proofing strategies, regardless of the specific model chosen. The ripple effect of DeepSeek's advancements is driving down the prices of AI tokens industry-wide. Competitors are compelled to lower prices to remain competitive, either through immediate efficiency improvements or long-term strategic pricing adjustments. This pricing pressure fosters a continuous race to offer the best value at the lowest cost. Building software with a forward-thinking approach is key in leveraging advancements like DeepSeek effectively. Anticipating the evolution of AI technology allows for swift integration of improvements, while rigid structures may face challenges in adapting to changing market dynamics. DeepSeek's impact highlights the evolving economic landscape of AI, where the cost of AI services aligns with infrastructure expenses, leaving room for innovative applications to create and capture value. The future lies in leveraging AI across industries, business functions, and diverse use cases to drive growth and profitability. These insights offer a glimpse into the transformative potential of DeepSeek and the evolving AI ecosystem. Exciting times ahead! #AI #Innovation #DeepSeek #AvisoAI #LLM #Agenticframework

𝗔 𝗦𝗲𝗶𝘀𝗺𝗶𝗰 𝗦𝗵𝗶𝗳𝘁 𝗶𝗻 𝗔𝗜? 𝗪𝗵𝗮𝘁 "𝗢𝗽𝗲𝗻 𝗪𝗲𝗶𝗴𝗵𝘁𝘀" 𝗖𝗼𝘂𝗹𝗱 𝗠𝗲𝗮𝗻 𝗳𝗼𝗿 𝗛𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲 𝗜𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝗼𝗻 There's an interesting movement happening in the AI landscape that could fundamentally change how we develop and deploy powerful tools, particularly in sensitive fields like healthcare. We're seeing hints of a potential pivot from major players like Meta and OpenAI towards releasing models with "open weights." 𝗪𝗵𝗮𝘁'𝘀 𝗱𝗿𝗶𝘃𝗶𝗻𝗴 𝘁𝗵𝗶𝘀? The competitive environment is heating up. Open-source alternatives are gaining significant traction, demonstrating a massive appetite among developers and organizations for models they can customize and control. This pressure seems to be nudging even the giants towards greater openness. So, 𝘄𝗵𝗮𝘁 𝗮𝗿𝗲 "𝗼𝗽𝗲𝗻 𝘄𝗲𝗶𝗴𝗵𝘁𝘀"? It's 𝙣𝙤𝙩 full 𝗼𝗽𝗲𝗻–𝘀𝗼𝘂𝗿𝗰𝗲. Think of it as a strategic middle ground: access to the trained model parameters (the "brain" of the AI) without revealing the proprietary training data or underlying code. This allows for significant customization while protecting core intellectual property. 𝗪𝗵𝘆 𝘁𝗵𝗶𝘀 𝗺𝗮𝘁𝘁𝗲𝗿𝘀 𝗽𝗿𝗼𝗳𝗼𝘂𝗻𝗱𝗹𝘆 𝗳𝗼𝗿 𝗵𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲:  • 𝗧𝗮𝗶𝗹𝗼𝗿𝗲𝗱 𝗦𝗼𝗹𝘂𝘁𝗶𝗼𝗻𝘀: Imagine AI models fine-tuned to understand the specific nuances of clinical language, different medical specialties, or unique patient populations. Open weights could unlock the ability to adapt powerful foundation models for precise healthcare needs, moving beyond generic capabilities.  • 𝗘𝗻𝗵𝗮𝗻𝗰𝗲𝗱 𝗣𝗿𝗶𝘃𝗮𝗰𝘆 & 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆: The potential to deploy these customized models within an organization's own secure infrastructure (on-premise or private cloud) is critical for healthcare, ensuring sensitive patient data remains protected and compliant.  • 𝗔𝗰𝗰𝗲𝗹𝗲𝗿𝗮𝘁𝗲𝗱 𝗜𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝗼𝗻 & 𝗥𝗲𝗱𝘂𝗰𝗲𝗱 𝗖𝗼𝘀𝘁𝘀: Building state-of-the-art AI from scratch requires immense resources. Accessing pre-trained weights provides a powerful starting point, drastically lowering the barrier to entry and allowing R&D efforts to focus on specialization, potentially saving millions and speeding up development cycles. This isn't just about technical access; it's about enabling a new wave of specialized intelligence. 𝘛𝘩𝘦 𝘦𝘳𝘢 𝘰𝘧 𝘰𝘯𝘦-𝘴𝘪𝘻𝘦-𝘧𝘪𝘵𝘴-𝘢𝘭𝘭 𝘈𝘐 𝘮𝘢𝘺 𝘣𝘦 𝘳𝘦𝘤𝘦𝘥𝘪𝘯𝘨, paving the way for highly specific applications fine-tuned for diagnostics, drug discovery, clinical operations, patient engagement, and more. 𝗪𝗵𝗮𝘁'𝘀 𝗻𝗲𝘅𝘁? If this trend solidifies, expect an explosion of tailored AI models built upon these more accessible foundations. We could be on the cusp of a significant acceleration in healthcare AI development, driven by more adaptable and accessible tools. The real revolution might not just be general AI, but the widespread proliferation of highly specialized AI intelligence across every facet of healthcare.

Don't try and get cute It's not about the "perfect" solution or "perfect" partnership So tempting to over-engineer the solution that we want So easy to shift the focus to our own innovation So easy to forget that our "solution" is just a piece of the puzzle to the customer's actual pain points and goals So easy to loose focus the actual customer's unique needs When in doubt, just stay focused on the actual needs Partnerships Co-Innovation is a huge advantage but, we need the customer in the room during development and, we DO NOT need a formal partnership to get started Here's a better way: 1. Get some customer success stories 2. Ask questions about the other tech & services involved in that success 3. Ask more questions about a. why the problem mattered b. why the solution worked c. how everything got stitched together 4. Ask for intros to the customer's trusted people at those other vendors 5. Look at your other customers to see if you see similar needs 6. Meet and Build a plan with the other vendors on how to replicate the greater solution 7. Kickoff with an intro between your customers a. the customer with the success b. the customer with the need 8. Bring the other vendors into the conversation 9. Validate that the success can be replicated 10. Formalize the partnership Stop re-inventing the wheel Stop over-complicating it Stop making it about you! Anyone in the company with a customer relationship can get this started Are you waiting for permission? Or, are you going to make an impact? #partnerships #innovation #sales #CustomerSuccess

Everyone wants open-weight AI models right now, but not everyone understands the risks that come with them. Transparency, reproducibility, community oversight - these are real advantages, and the desire for more openness in the ecosystem makes complete sense! But there’s another side of the equation that gets far less attention. Once the weights are out, control disappears. There’s no monitoring, no enforcement, no visibility into how the model gets fine-tuned, merged, or repurposed... and the derivatives spread in a way nobody can trace or evaluate. The benefits definitely stay, but the ability to intervene is effectively gone... A new paper from MIT, Stanford, and UK AISI breaks this down into 16 unsolved technical problems that define the real risk surface of open-weight AI: from data curation and tamper-resistant training to adversarial evaluations, staged deployment, and provenance. Stopping openness isn’t realistic, and ignoring the risks isn’t either. There’s no simple way to make open-weight AI both fully open and fully secure. But there is a path to making it safer: better data curation, stronger tamper-resistant training, real adversarial evaluations, and reliable provenance. The goal is building the technical foundations that let both progress and security exist at the same time!