Chris "The Wiz 🪄" Alexiuk’s Post

𝐭𝐡𝐢𝐧𝐤𝐦𝐚𝐩.𝐚𝐢 is now colorful! I've already gotten feedback from users: - that the variety of colors is more pleasing than black-and-white boxes, - and that it is easier to distinguish your displayed "think maps" from one another. Thinkmap helps you quickly orient your mind to the right context for the project you are working on. If the colors help you with that, please let me know! If they hinder it, or make it hard to read, doubly let me know!! You don't have to worry about selecting colors for each task. Tasks are colored by their 𝐦𝐞𝐚𝐧𝐢𝐧𝐠, as encoded by a language model. So, similar tasks will have a similar vibe.

CodeChef Starters 153 Solved 4 / 8 👨🍳 All the problems were very simple. A: If X <= 60 then YES else NO B: If Y - X == Z - Y then 0 else 1. C. By using Combinatorics, The equations that satisfy the conditions are: R + G - B > -1 && R + B - G > -1 && G + B - R > -1 then YES else NO D. Initialize max to -1 and calculate max where i != Ai. The answer is max + 1. #codechef

I recently got access to play around with IAN by Cyphae. IAN, short for Intelligent Agent Network, is a network of LLM agents with a meta-agentic analysis system on top that collates and provides a systematic summary of the outputs of the agent network apart from their individual responses. To demonstrate IAN's capabilities and how this approach of aggregating multiple models is revolutionizing generative AI with smarter, more accurate insights, I asked IAN to help me figure out how to reverse engineer its web UI, hacking together a TypeScript client along the way and documenting the entire process in this little dev.to piece. Check it out to explore the kind of superpowers that are unlocked when you make multiple foundation models work together to solve a problem.

Today, I faced a challenge: finding the length of the longest common substring between two strings. Initially, I tried a brute-force approach, which worked but was painfully slow for larger strings. Determined to find a better solution, I explored various methods and even consulted with a friend. Their suggestion led me to a simpler, more efficient approach: iterating over possible substrings from the first string and checking each against the second string. With this new method, I was able to solve the problem efficiently and effectively. It was a great reminder that sometimes, the simplest solutions are the most powerful. Feeling accomplished and grateful for the support and insights along the way. Here’s to finding elegant solutions and learning from every challenge! 🚀 #geekstreak2024 #ProblemSolving #CodingJourney #Efficiency #Algorithm #TechChallenges

From the blog - I continue to use GenAI for completely serious, enterprise reasons: "Using GenAI to Help Pick Your D & D Class" Makes use of Google Gemini 1.5 - their latest model. https://lnkd.in/e6NhV5jY

AI-Generated Tests: The Future of Bug Preservation! 🎉 Dev: 'Look! AI wrote our unit tests. 80% coverage!' Manager: 'Impressive! What does it mean for our code quality?' Dev: 'Well, in the absolute best case, it might catch a regression.' Manager: 'And... in other cases?' Dev: 'It'll meticulously verify that our bugs are present and accounted for.' Manager: '...But we have 80% coverage, right?' Dev: 'Oh yes, we're thoroughly testing our mistakes.' #AITesting #BugsAsFeatures #CoverageIsKing

I am planning 2025-2027 with my AI friends. 1. Converted my older blogs into single Markdown files, conveniently readable by LLMs. 2. Put the blogs and my diary into Gemini 1.5 Pro with 2M context (only 500k used so far). 3. Asked to analyze and summarize my life and career 2000-2010, 2010-2020, 2020-now. 4. Asked to project for 3 more years with high-risk high-reward, mid-risk mid-reward, and "surprise me" scenarios. I love the output a lot! Below is high-risk high-reward scenario in a form of a catchy rock song. (Verse 1) DeepMind dreamin', code in my veins AGI's calling, breaking the chains 2025 fire, burning bright Research warrior, day and night (Chorus) Silicon sunrise, a new age begun Algorithms racing, the battle is won Creative thunder, a digital roar Roman's coding, forevermore! Link to the Suno-generated song is in the comments.

If you want to build your own AI-powered Discord bot to have fun with your friends, check out my code lab, in which I have shown how to create one simply using JS and Gemini. Written in collaboration with Neha Keshri Please let me know what you think about this. Discord Google Gemini

Today the world has changed. People don't know it yet, but it has. Anthropic is the first frontier model provider to release a product that allows an AI to control your computer. It's still experimental. But it's a game changer. I'm not one of those people who go on YouTube and post hyperbolic headlines. In this case, I feel like AI has crossed another rubicon. Mind you, there have been some demos, often overblown or fake, of AI controlling screens. (ahem, Devin). There have been folks sending screenshots in sequence to good effect. These were interesting efforts. But a frontier model will give AI access to do tasks for you on your computer. At scale. It will give devs access to offer you so many features on your computer via APIs. At scale. Agent: Unlocked. In the image attached, you see a screenshot of Anthropic's computer use feature, asking Claude 1, to go to the browser, navigate to claude.ai, ask it to create a webpage with a 90s style theme. In the video, claude goes on to download the html locally. Pop open a code editor to improve it. Mindblown. This is the GitHub repo where you can try it using a Docker, Inc container: https://lnkd.in/gt8QjuDA Enjoy. P.S. any thoughts Kwasi Ankomah, Aniruth Sarathy, Felicia Schwartz, Mohan Gulati

My insights on the recent "DEVIN" release that makes many go crazy with excitement: 1. It's NOT the "1st AI software engineer". I'm definitely in awe of the founders' achievements and like their AI innovation – but it's not even close to being 1st. 2. From what I can see, it's an "autonomous AI agent swarm with coding and web capabilities" – of which I have at least half a dozen studies and GitHub repos of existing projects on my hard drive and could surely find even more because they came to my attention 1,5 years ago. (And I'm sure late to the game.) 3. DEVIN proves how important GOOD branding & marketing are because they did many things right: - DEVIN doesn't talk about "autonomous AI agents" and only uses simple language. - DEVIN is basically like Apple because it's not the 1st in tech innovation but makes it usable for the mass market – by offering the simplest, best and most beautiful UX. - DEVIN's product demos are great, enjoyable, understandable and impressive. - DEVIN's founder is likeable and an old video of him from a school event resurfaced (and went viral), where he wins a math competition by answering even the most difficult questions before the moderator even read 1/3 of the longer questions. - It's the perfect timing because we're at an inflexion point where AI desire, understanding and belief meet business needs, global crisis, etc. - And their backing is some of the most renowned names in Silicon Valley and VC funding, which proves that it also matters who you know. - It also seems like a pretty well-coordinated PR campaign because "coincidentally" many influencers – some from tech and others who are simply famous aka with Klout – come out in unison to praise Devin. Just wanted to point that out since I live, work and understand both worlds – business and tech – and heard many IMHO not-on-point takes. Anyhow. I wish DEVIN and his team all the best. Great AI innovation (just not the 1st, but the 1st to market it smartly):