Ludwig Bull

UCL are using no code to generate additional revenue 🎓 Here's how they built a SaaS tool on Bubble👇🏽 UCL are one of the top universities in the world. The Medical School were fed up of their dated Microsoft Access system. Things were moving too slow, and they needed a change. They explored traditional software development but it was too expensive. They came across Bubble and got building. #1 They built a personal tutoring app The idea was to assign students a tutor. Students could see the notes, and the tutor could respond. #2 They built an app to manage student records #3 They build a curriculum map app - which now has 12,000+ users It guides students on their learning and helps them prepare for exams. The tool is so good that other departments and universities started reaching out. So good that UCL have started reselling this as a SaaS tool with customisations... Goodspeed Bubble Taylor Bennie

25

2 Comments

DeepSeek V3 - new open model that is better than Claude Sonnet at the Aider Polyglot coding benchmark and - perhaps most importantly - has 10x cheaper inference costs. The massive price drop means you can experiment with having a Aider build you a feature for $2 instead of $20. $20 is a little too pricey for rapid experimentation in your development workflow. Unsurprisingly, Deep Seek's site is down. Other Model Inference providers like Hyperbolic seem to have also taken Deep Seek V3 down for maintenance https://lnkd.in/gtn8Hb36

9

It’s finally here. Chapter 2/4 on The Post Web And it’s a beast. A thorough examination of The Stack and it’s implications.

49

6 Comments

It’s interesting to see that each LLM model has its own style of writing. After 2 years of using ChatGPT and reading people’s posts, I’m sure I’m not the only one that can tell which posts and articles are copy pasted from ChatGPT. They lack original content, structured in the same way, and use similar words across posts. This is a critical example of how AI won’t just replace humans IMO. If I see a post that looks like it was written by AI, I tend to gloss over it more than those that are actually written. Spelling errors and grammatical errors, lack of political correctness etc make the posts original as well. These platforms don’t have as much original content as most would believe. 90% of content out there is just copying other creators! I hope we don’t get to a place where it becomes 99% non-original because of AI/LLMs!

35

5 Comments

The amount of stuff you learn along the way whilst building a start up that no one tells you is intense, here's a few to keep in mind: 1. When the developers say 2 months, double it, when they say 3 months double it, when they say 6 months, sack them. 2. Cash is everything, build your business to collect and retain cash. 3. Remote teams need to buy into what you are building, otherwise they will do a shit job. They need to run with your culture and vision. 4. Don't listen to the negatives. For your vision to flourish, you will probably have to walk a pretty lonely path. 5. This shit will cause you heart problems but will also make you feel more alive than you have ever been 😂. Thank you for reading my Ted talk. #Startup #business #thoughtleadership #entrepreneur

40

(Gift link) - Nice oped from Martin Casado and Katherine Boyle at a16z VC on the blatant bias for Big Tech in D.C. on all things AI. I've never seen such government favoritism for incumbent tech companies in my 40+ year career. It's frankly obscene, self-destructive behavior. It may be some part exploitation of ignorance on the part of Big Techs (you've been duped), it may be some part faulty ideology, but I don't think there is any question it's also political corruption. It's one of the most extreme examples of government capture we've ever seen. To clarify my position -- I don't think it was necessary to even mention open source in this op-ed as it muddies the water with two issues, even if related (I support some open source, though in LLMs it does have security considerations -- all LLMs do). I'm also not against all regulation in AI, but I am fiercely opposed to regulation that favors dominant monopoly incumbents and/or new ventures with multiple billion dollar valuations that don't even have a product. As I repeat often, the super majority of problems and risks that need regulating in AI are caused by LLM firms and their Big Tech partners. That's where the focus should be, and since they are conflicted, the government shouldn't be relying on them for advice like they are, or internal AI advisers who came from them who are on the payroll of non-profits funded by tech billionaires with a myriad of conflicts. It's ugly. Bottom line: While the U.S. has a serious problem with the Big Tech oligopoly and their oligarchs, we are not quite yet an oligarchy, but we're getting very close to meeting that definition. We better fix it before it's too late. ~~~~~ Quote: Although the public-facing argument for AI regulation is to promote safety, we believe the true purpose is to suppress open-source innovation and deter competitive startups. Many emerging technology companies are well-equipped to help solve national-security challenges, but of the 22 members on the board, none represent startups, or what we call “little tech.” Only two are private companies, and the smallest organization on the board hovers around $1 billion in value. The AI companies selected for the board either are among the world’s largest companies or have received significant funding from those companies, and all are public advocates for stronger regulations on AI models.

1

The UK government's decision to shelve £1.3bn in funding for tech and AI projects is a significant setback for innovation in the UK. This funding was earmarked for groundbreaking initiatives like the Edinburgh exascale supercomputer, which had the potential to revolutionise fields like AI, drug discovery, and climate change research. We need the government to come forward with new proposals quickly and ensure the UK doesn't fall behind on the global stage. #AI #Fintech #Technology

19

7 Comments

https://lnkd.in/eu2Bb-x2 Large language modules like Chat GPT 4o are set to revolutionise the car-buying experience, which, according to the Telegraph, The customer is ready for. The article discusses the decline of traditional salespeople due to the agency model. However, based on our experiments with Chat GPT 4o, the salespeople's most significant threat might be AI. Our experience shows that most customers don't know exactly what car they want when initially searching for a new car, so ask Chat GPT 4o, 'What is the best car for a family of four with a wet dog? ' and you will receive a personalised response, validated by web links to actual recommendations. Our first very basic AI sales agent, Autoowl.ai, launched last week. It links the AI responses to a stock feed of cars the customer can buy inside our smartphone application or website. Our team is blown away by the large language module's ability to respond accurately to topics related to the car-buying experience, finance, and insurance. The next step is to train the AI to interact with other 3rd party services to apply for finance and insurance and book a test drive. We can see no reason why the whole car buying journey won't be consolidated in one conversation with an AI salesperson available 24 hours a day and seven days a week, providing a service many car buyers might prefer.

7

8 Comments

AI is shaping our future, but who exactly is shaping AI? The reality is stark. ❌ While women make up half of the UK population, only 22% of AI professionals are female. This isn't just about diversity - it's about the future of technology itself. At today's Women in Silicon Roundabout event, I witnessed more than discussion - I saw action. A room full of brilliant minds actively building the future we need to see. Because when we limit who shapes AI, we limit what AI can achieve for humanity. The brilliant Naomi Pearse put it perfectly: ↳ We need to democratise AI knowledge, build truly inclusive learning environments, and ensure every woman who wants to shape the future of tech has the tools to do so. Because the next breakthrough in AI shouldn't depend on gender - it should depend on brilliance. And right now? We're missing out on far too much of it. #ai #womenintech #artificialintelligence #machinelearning #wosr24

51

Is it PMF when your own product is saving you days of work trawling through hundreds of applications? It's been less than 24 hours and we've had hundreds of applications for our Mid React Dev role (link in comments). Popp AI is pulling every applicant from LinkedIn, analysing them on autopilot, categorising and scoring, and surfacing the very best - with all the evidence clearly referenced. Time saved thus far: about 2 days 🤯 #productivity #hiringhack #talent

44

8 Comments

Trends in AI July 2024: As Anthropic launched Claude 3.5, undercutting GPT-4o’s price with a competitive performance, and Google DeepMind’s Gemma 2 - 27B emerges as the new strongest ‘somewhat open’ model, ARC-AGI has put a $1M bounty for solving a deceptively simple task where current AI falls short. AI Startup Etched scores $120M to develop a Transformer specific ASIC, and CuspAI, a new European startup secures millions for developing carbon capture technology. Dutch AI pioneer Textkernel was sold for $300M to Bullhorn. In California, discussion is heating up about SB 1047, the state’s newly proposed AI regulation Bill, as AI products from Apple and others are delayed in Europe in a stare-down around new legislation. Join us for the next edition of the monthly Trends in AI webinar this Friday, July 5th at 12 PM CEST to stay up-to-date with the latest advancements in AI R&D, covering AI industry news and trending research papers like Prompts as Auto-Optimized Hyperparameters, TextGrad, and much more! Plus previews of SIGIR and ICML. Sign up here: https://lnkd.in/ef-kP5PA

25

A signature moment for UK tech. I wrote an oped today for The Times on how this moment for Wayve can trigger a movement for UK tech. In the same edition Scale AI's Alexandr Wang announces he is setting up a UK HQ and University of Cambridge announce they are doubling down on spin outs. Four main externalities helped the Wayve story this far: - top Cambridge academics, UK skills density in AI - immigration policies attractive to talent - corporates partnering with startups - UK government regulatory arbitrage Rishi Sunak's support for the tech sector can be seen by 2 quotes from him welcoming tech companies today. These conditions should unleash more AI leaders from the UK Policies do matter for entrepreneurs. For founders what has helped and what has hindered ? cc Microsoft NVIDIA SoftBank

817

68 Comments

I want to run a bit of a crazy experiment... (even by my standards 👀) With just under a week left of the no-code competition I'm running with OPENFORMAT and BuildShip, I want to help 5 people go from NEVER having built an app or product before, to having something genuinely innovative that they can submit for our competition. If you've had an idea on your mind for a while, want to learn a new (and VERY valuable) skill or you're just a bit mad like me, I think you'll this very intersting. I don't plan on having much of a life this week so you can have a lot of 1-1 access to me for brainstorming, building and to give you a roadmap for getting started. I reckon with 10-20 hours time invested before Sunday we could get to something pretty damn cool. I know it's late notice, and I know it's a bit wild, but I'll ask the question anyway.... Anyone in? 👀 If you're on the fence, you can check out some more information about the competition below (the prizes might just sway you) 👇 https://lnkd.in/ebkAbiR7

22

11 Comments

Handy example of just how useful ChatGPT (or other LLM's) are for everyday task The turnout for yesterdays UK election was 60% and I wanted to see how that compared to previous years Searching Google led to a PDF on the UK parliament site that had the data from 1918 to 2019 as bar chart So I 1) Copied the orginal image into ChatGPT 2) Asked it to extract the data into a table 3) Add a new row for 2024 4) Create a new bar chart with the data in The whole process took under 5 mins

25

4 Comments

Exciting that OpenAI have launched another AI video generator - Sora… Well exciting if you live in the US (not available in the UK) and if you pay for membership (£20 a month). And if you only want to make 20 second videos (longer videos are for the £200 a month Pro members). Now while I love using AI to create new graphics, it’s not that new or innovative right. Hell Canva lets me create images using words. OK they aren’t cinema quality, but I’m also not a film maker. You know what I’d love from AI companies even more?!? AI that can help me pack for my family of 5 to fly to visit my family in San Diego before returning to the UK to spend New Years in Norfolk. And the presents. All must fit into the limited number of suitcases that will fit into a rental car and manageable by sleep deprived children being hauled through international airports. That would really be some valuable intelligence! …Oh could it please also find the 50 missing single socks I seem to have lost in my house somewhere… https://lnkd.in/e_FAQzUi

13

2 Comments

I’m sliding in late tonight because I’ve been digging into Bun’s new C interop feature, and couldn't help but think: just because you can run C in the browser doesn’t mean you should. Remember Spectre and Meltdown? Same deal—code running where it shouldn’t be. Fast forward a bit, and we’re dealing with the sort of CVEs that script kiddies salivate over. Now let us all say this in unison: "write code for the right platform". C is great for building browsers, not for running web apps in them. To all my C engineers out there, I’ve always respected your craft, but why are you out here making us lose sleep? Now we’re one log4j-type exploit away from having C code breach the Linux kernel and hold entire enterprises hostage. Just thinking about it makes me WannaCry. So what do you think? 🤔 #WebDevelopment #CyberSecurity #DevOps #SoftwareArchitecture #JavaScript #Innovation

1

1 Comment

💡 Super Weights in LLMs - How Pruning Them Destroys a LLM's Ability to Generate Text ? TLDR - Super weights are crucial to performance of LLMs and can have outsized impact on LLM model's behaviour The presence of “Super weights” as a subset of outlier parameters. Pruning as few as a single super weight can ‘destroy an LLM’s ability to generate text – increasing perplexity by 3 orders of magnitude and reducing zero-shot accuracy to guessing’. 📜 https://lnkd.in/g_ps-CRf 💕 Subscribe to receive more such articles to your inbox - vevesta.substack.com #LLM #Machinelearning #research #latestResearch #datascience

6