Innovation Challenges in Tech

New product initiatives within large companies often fail to achieve their potential because they have too much rather than too little. They have too much: 1) Headcount You are now under pressure to come up with something for all these people to do. Especially in cultures where “engineers must always be coding” and a PM is seen as failing if engineers are even briefly “blocked on requirements.” 2) Democratic decision making Creative ideas get killed (or watered down) by groups — yet this is the default in most big companies, even those that claim to use RAPID or similar frameworks. 3) Optics requirements You must now manufacture metrics and milestones to show straight-line progress and demonstrate certainty — during what is, by its very nature, an uncertain journey. 4) Involvement of the “core” product group To appease the leaders of the company’s cash cow, you make compromises that weaken your product. These leaders have the most power within the company and some may even try to confuse the CEO or quietly sabotage your initiative. 5) Reliance on the company’s distribution Due to the mirage of distribution, you won’t be incentivized to deeply understand your customer like a real startup would. Your initial traction is misleading — you get a usage spike, but: (a) those users are scattered across segments, not your core segment (have you even identified that core segment?) (b) what’s given will be taken away — that homepage slot for your new product will disappear next quarter due to VP jealousy or shifting OKRs (with some hand-wavy “metrics neutral” excuse). So if you are leading a new initiative within a larger company and your CEO/CxO asks you what you need to succeed, do not default to the answer that everyone in this situation gives: “I need more resources”. Instead, consider asking for less — less reporting, less certainty, less consensus-driven decision making, less meddling, and less pressure to build out a “full team” & great operations early on. If your CEO is competent, they’ll respect it. (clearly, this entire post is only for the intrepid product leaders who want to make winning products, it is not for everyone 🙂)

We've spoken with 30 companies who developed RAG-based chatbots on PDF documents. Every single one has failed: Core issues: 1) In vector space, "non-dairy products" is often closer to "milk" than "meat," this is a fundamental flaw of vector embedding search because they're very lossy. 2) Splitting documents into smaller chunks disrupts coherence, breaking cross-references and context. 3) Adopting new RAG architectures, re-embedding chunks with new models, and rerankers requires continuous, costly data (re)engineering efforts. 4) No Support for Aggregations – Vector search struggles with queries requiring aggregation (e.g., max, min, total), making it unreliable for analytical use cases. As a result, companies band-aid their chatbots by writing complex heuristics to patch these failures. Ironically, many end up going back to rule-based chatbots. Our advice is simple - Do You Even Need RAG? LLM models are dirt cheap now and quite comparable to embedding models. If your documents are small: just load them directly into the LLM context. If your documents are large: Enrich with rich metadata and query the right documents and pages based on the metadata. Chatting on documents must be redesigned.

The web succeeded because it solved a coordination problem: millions of independent actors needed to link documents without central control. Open standards - HTTP, URIs, HTML - made this possible by providing a shared protocol layer that no single party owned. AI agents face an analogous coordination problem. In multi-agent systems, agents built by different parties must exchange not just data but meaning: what a "customer", "approved", or "delivery date" actually denotes. Natural language alone cannot solve this. LLMs can interpret natural language flexibly, but flexibility is precisely the problem when agents must act reliably on shared information. Ambiguity that agents can resolve through context becomes a source of failure when machines transact autonomously at speed and scale. Sharing meaning unambiguously requires two things: a formal system of semantics capable of precise entailment, and globally unique identifiers that can be resolved to authoritative definitions. Without formal semantics, agents cannot reason reliably about what follows from what. Without resolvable identifiers, "customer" in System A and "customer" in System B remain dangerously ambiguous - they might align, or they might not. These are not novel requirements. They are the foundational principles of the semantic web: RDF for formal semantics, URIs for identification, and HTTP for resolution. Anyone building agent interoperability from scratch will either fail to meet these requirements, or meet them and arrive at substantially the same architecture. The real question is whether to adopt these principles in open or proprietary form. Proprietary approaches face a structural problem: interoperability requires shared definitions, but shared definitions only become valuable when widely adopted, and wide adoption requires openness. This is the same network-effect logic that made the web's openness essential. A proprietary web would have remained a collection of walled gardens. The trajectory is therefore clear: as agentic systems mature and the cost of failed interoperability mounts, the pressure towards truly open semantic standards will intensify. It is inevitable. ⭕ Semantic Bow Tie: https://lnkd.in/e6z3hFVn ⭕ The "O" Word: https://lnkd.in/e7v4AjXZ 🔗 Build Your Own Semantics: https://lnkd.in/ezHU2amU

Sustainability = Innovation 🌎 Integrating sustainability into business strategy requires continuous advancements in technology, processes, and resource management. At the same time, sustainability challenges drive research, development, and operational efficiencies that lead to new market opportunities and competitive advantages. Resource constraints drive material and process innovation. The need for alternatives to finite or harmful materials has accelerated the development of advanced composites, circular economy models, and energy-efficient production systems, improving cost efficiency and resilience. Addressing sustainability challenges requires systems-level innovation. Reducing emissions, optimizing resource use, and minimizing waste require advancements in supply chain management, product lifecycle design, and industrial processes, reshaping entire sectors. Cross-functional collaboration is critical. Sustainability initiatives require input from engineering, data science, regulatory compliance, and finance to develop integrated solutions that meet environmental targets while maintaining operational and commercial viability. Data-driven approaches enhance sustainability performance. Measuring environmental impact enables companies to identify inefficiencies, optimize resource allocation, and refine business strategies based on quantifiable sustainability metrics. Long-term sustainability targets drive investment in research and technology. Businesses are accelerating development in areas such as AI-driven resource optimization, carbon capture, and next-generation materials to align with regulatory requirements and market expectations. Nature-based solutions provide scalable innovation opportunities. Biomimicry has led to advancements in self-healing materials, passive cooling systems, and regenerative agricultural techniques, improving efficiency and resilience across industries. Sustainability is reshaping business models. The transition to circular economy principles, service-based models, and regenerative supply chains is driving competitive differentiation and long-term value creation. Innovation is fundamental to achieving sustainability objectives. The convergence of regulatory frameworks, technological advancements, and market shifts is reinforcing the role of sustainability as a driver of industrial transformation and business resilience. #sustainability #sustainable #business #esg #climatechange

🚨Techno-optimistic scientists take fewer climate actions 🚨 Technological innovation is key to mitigating climate change, yet excessive faith in technology may undermine the wider societal transformations needed to address it. In our new preprint, Viktoria Cologna, Maien Sachisthal, Jonas Haslbeck, and I examined techno-optimism among N = 9,199 scientists across 115 countries and how it relates to their civic engagement and lifestyle choices. 🔗 Read the paper: https://lnkd.in/dMjc6f5E 🔧 𝐖𝐡𝐚𝐭 𝐝𝐢𝐝 𝐰𝐞 𝐝𝐨? We estimated the share of scientists who believe that "advances in technology will largely solve the problems caused by climate change", and examined how this belief varies by research field, climate-relatedness of their work, and political orientation. Using causally informed Bayesian regression models, we then estimated how techno-optimism relates to civic actions (e.g., signing petitions, participating in protests) and high-impact lifestyle changes (e.g., reducing flying, shifting to plant-rich diets). 💡 𝐖𝐡𝐚𝐭 𝐝𝐢𝐝 𝐰𝐞 𝐟𝐢𝐧𝐝? 1️⃣ 𝐓𝐞𝐜𝐡𝐧𝐨-𝐨𝐩𝐭𝐢𝐦𝐢𝐬𝐦 𝐯𝐚𝐫𝐢𝐞𝐬 𝐬𝐭𝐫𝐨𝐧𝐠𝐥𝐲 𝐚𝐜𝐫𝐨𝐬𝐬 𝐟𝐢𝐞𝐥𝐝𝐬 & 𝐩𝐨𝐥𝐢𝐭𝐢𝐜𝐚𝐥 𝐨𝐫𝐢𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧. Applied sciences show the highest levels of techno-optimism, humanities the lowest. Levels did not differ based on whether scientists work on climate change. Scientists on the political right were more likely to be techno-optimists. 2️⃣ 𝐓𝐞𝐜𝐡𝐧𝐨-𝐨𝐩𝐭𝐢𝐦𝐢𝐬𝐭𝐢𝐜 𝐬𝐜𝐢𝐞𝐧𝐭𝐢𝐬𝐭𝐬 𝐭𝐚𝐤𝐞 𝐟𝐞𝐰𝐞𝐫 𝐜𝐢𝐯𝐢𝐜 𝐚𝐜𝐭𝐢𝐨𝐧𝐬. Across all forms of civic engagement — especially signing petitions and participating in protests — techno-optimistic scientists were less likely to participate. On average, their engagement was 28% lower. 3️⃣ 𝐓𝐞𝐜𝐡𝐧𝐨-𝐨𝐩𝐭𝐢𝐦𝐢𝐬𝐭𝐢𝐜 𝐬𝐜𝐢𝐞𝐧𝐭𝐢𝐬𝐭𝐬 𝐦𝐚𝐤𝐞 𝐟𝐞𝐰𝐞𝐫 𝐥𝐢𝐟𝐞𝐬𝐭𝐲𝐥𝐞 𝐜𝐡𝐚𝐧𝐠𝐞𝐬. Techno-optimistic scientists were less likely to engage in high-impact lifestyle changes such as reducing flying, reducing car use, and adopting plant-rich diets. There were no differences observed in switching to electric vehicles or renewable energy at home, both of which require no major lifestyle trade-offs. On average, engagement in lifestyle changes was 20% lower. 🔥🌎 𝐖𝐡𝐚𝐭 𝐝𝐨𝐞𝐬 𝐭𝐡𝐢𝐬 𝐦𝐞𝐚𝐧 𝐟𝐨𝐫 𝐜𝐥𝐢𝐦𝐚𝐭𝐞 𝐚𝐜𝐭𝐢𝐨𝐧? As trusted knowledge producers, scientists influence how societies understand climate risk and what solutions seem viable. Our findings suggest that scientists’ own beliefs about technology shape their engagement with climate action. We hope this work sparks reflection and dialogue within the scientific community about how our worldviews influence not only our research, but also the kinds of climate actions we take and promote.

Europe has no shortage of talent, ambition, or ideas but it still lacks a functional Single Market for startups to scale. Regulatory fragmentation and administrative burdens remain some of the biggest barriers to building global tech companies in Europe. 27 legal systems, conflicting rules, and incompatible digital infrastructures mean high costs and high friction. This model isn’t competitive. It slows growth, deters investment, and pushes innovation out of Europe. That’s why at Atomico we support the ambition behind the proposed 28th regime but with the clear message that we consistently hear from our founders: “Europe must get this right or risk it being ignored entirely.” To succeed, the regime must be credibly faster, simpler, and more scalable than the status quo - a successful 28th regime should deliver: ➡️ One EU-wide legal identity: one new pan-European legal entity with harmonised rules and digital credentials, portable across all Member States. ➡️ Faster, full-digital setup: 100% digital formation and operations via a single EU-wide company registry. ➡️ Credible investment-readiness: standardised investment documents for simpler, faster fundraising. ➡️ Harmonised employee stock options: to better attract and reward talent and improve global talent competitiveness.  ➡️ Simple rules for local taxes and employment. This can’t be symbolic or driven by legacy voices. It must be co-created with the founders building Europe’s tech future. I’d encourage everyone who supports this vision to fill out the EU consultation before *Sept 30* (Link in comments) EU–INC. Let’s make this a defining moment, not another missed opportunity. Let’s build an ecosystem that rewards bold bets and unlocks scale. Andreas Klinger

𝗘𝘂𝗿𝗼𝗽𝗲’𝘀 𝗹𝗮𝗴𝗴𝗶𝗻𝗴 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝘃𝗶𝘁𝘆 𝗮𝗻𝗱 𝗥&𝗗: 𝗠𝘂𝗰𝗵 𝗺𝗼𝗿𝗲 𝗿𝗶𝘀𝗸 𝗰𝗮𝗽𝗶𝘁𝗮𝗹 𝗻𝗲𝗲𝗱𝗲𝗱 ‼️ Last week the International Monetary Fund published a very interesting and comprehensive paper about the need for more venture capital in Europe to tackle our continents challenges. To name a few: ✔️productivity per hour worked is app 30% lower in 🇪🇺compared to the 🇺🇸 ✔️R&D investments are still way below the target of 3% per annum ✔️Within the top 100 tech companies worldwide merely a handful are European Is it all about 💶 I here you say? No it is about keeping up our welfare for future generations. And about a liveable planet. And increasing our innovation and competitiveness are crucial to do so. Which is also the key message of Mr. Draghi’s report I hope. The IMF report takes a deeper dive into the underlying issues: ✔️ VC investments are only 0,4% of GDP. In the US it is 3x as much ✔️Europeans park their savings in bank accounts. And banks are very risk aversie when it comes to financing hightech startups. ✔️Long term savings go primarily via pension funds, who hardly invest in VC in Europe (despite some positive signs recently) ✔️The EU has fewer and smaller VC funds leading to smaller rounds, less opportunities for scale-up financing and limited exit options ✔️ European scale-ups end up listing in the US instead of Europe itself ✔️ National fragmentation within the EU leads to a lot of barriers for scaling What has to be done? ✅ Increase efforts on a real single European market, for example by consolidating stock market exchanges and diminishing cross border red tape ✅ Make it more attractive for pension funds and insurers to step into VC ✅ Enhance the capacity of European Investment Bank (EIB), European Investment Fund (EIF) and national promotional institutes, like Invest-NL ✅ Implement preferential tax treatments for equity investments in startups and VC funds ✅ Encourage more funds-of-funds And I would like to ad to the findings in the report two things: 1️⃣ We need a cultural mind shift, more urgency and embracing true entrepreneurship 2️⃣ We have to step up our game when it comes to tech transfer. Transforming our high quality academic knowledge into economic and societal impact via startups.

As Mario Draghi’s report released today demonstrates, the EU is falling behind global rivals because of limited innovation. Since 2019, the EU has created over 100 pieces of digital regulation. Whether you’re a technology startup or a small retailer, regulatory complexity is a minefield. Developing, launching or just using technology is harder in Europe than elsewhere in the world. Of course, “anything goes” is not an option and rules are required - but the EU is holding itself back at a time where it could be thriving. Our research with Public First shows that generative AI alone could add €1.2 trillion to the European economy. Much of Google’s innovation is led from Europe. We work with talented European entrepreneurs, businesses and innovators every day and see first-hand the benefits that the single market could yield for them. But a new approach is needed if Europe is not to miss the moment. Here’s what needs to change: 1️⃣ Shift from regulatory growth to economic growth: Europe doesn’t just create a huge number of regulations related to digital society - the regulations they create are often conflicting, untested and inconsistently implemented. The explosion of rules makes it almost impossible for Europe to create and nurture the next tech unicorns. Draghi is right that the EU now needs to focus on enabling innovation: promoting the use of digital technologies to innovate and drive through breakthrough advances. 2️⃣ Invest in R&D: To compete in AI, the EU needs to prioritise research and development, working with the private sector to incentivise it and make funding more accessible. The EU currently lags behind the US, Israel, South Korea, Japan, the UK and China on R&D investment. Without the right incentives to develop and roll out new technology, Europe is stifling its talent. 3️⃣ Build the right infrastructure: AI breakthroughs are only possible with the right computing technologies and data centres - plus the renewable energy to run them. So the EU needs to allocate more funding towards financing such infrastructure, as well as incentivising and enabling the private sector to do the same. 4️⃣ Prioritise skills & education: People will need support to seize the benefits of AI in their work and life. A revitalised European Skills Agenda should put skills and education at the centre, while AI should be added to school curriculums. Google wants to help Europe seize the benefits of innovation. Over the last decade, we’ve worked hand in hand with Governments to build new technology responsibly; train over 13 million Europeans in digital skills; and support over €179 billion in economic activity across the EU. As a European, I’m proud of this work, but I know there’s much more to do. Read Draghi’s report here: https://lnkd.in/epBxtymw

There’s a shift happening in our discussions with customers about digital transformation. We're actively observing the unfolding of the Industrial Metaverse, propelled by foundational technologies like Digital Twin and IIoT, alongside simulation, automation, and AR/VR. These technologies are not future aspirations; they're present-day tools, forming the building blocks of the Industrial Metaverse today. In a recent Forrester study conducted by Paul Miller and Martha Bennett, we learned that an impressive 55% of business and technology professionals in production and manufacturing organizations are planning to adopt metaverse technologies within the next 12 months. So, what is the Industrial Metaverse? It's a place where engineers make quick decisions using accurate and robust data in a real and virtual space, collaborating, and interacting with a comprehensive digital twin. It allows us to visualize the digital twin in its real-world context, gain insights in a realistic environment, and collaborate in real-time to make immediate changes. Companies are now harmonizing data that was once fragmented across various applications, creating a unified "single pane of glass." This unified platform facilitates collaboration, interaction, and full immersion into a digital representation where real-world physics and data transform the digital experience into something incredibly lifelike. This empowers companies to connect their workforce, suppliers, and customers, forging a new path for future business endeavors. The journey has started. The Industrial Metaverse is taking the comprehensive digital twin to the next level – showing how operational data, plant information, geospatial insights, simulations, and 3D data can be combined to provide a genuine real-world perspective.