Alan Kent

Running a Company on Markdown Files CompanyOS: a skills-only system that turns Claude Code into the operating layer for an entire company. No application code, no web UI - just markdown files that teach Claude Code how to run business operations. https://lnkd.in/gACFhTMk #AdventuresInClaude #ClaudeCode #CompanyOS

226

45 Comments

Meet the robots of re:Invent. The startups behind these systems are pushing the boundaries of physical AI, and Amazon Web Services (AWS) is helping them go from research to real-world impact. Applications for the 2026 Physical AI Fellowship are now open: https://lnkd.in/gzr5nvhA

574

36 Comments

Amazon has added an S3 Vector store to its AWS offering, which effectively competes against Pinecone, currently the most widely distributed vector store. (Thanks to David Lee for this). Vector stores are interesting - they underlie almost all of core search technologies (especially anything that utilizes TF/IDF type text token encoding systems) and act as a bridge to LLMs and similar transformers. Their primary advantage is their ability to perform similarity analyses between encoded documents, which differ from knowledge graph searches in SPARQL or OpenCypher (or XPath/XQuery, for that matter) which can do direct faceted searches but are less than ideal for thematic or conceptual similarity. My suspicion is that long-term most knowledge graph engines (KGEs) will have some vector store support as well, primarily for textual rather than structural encodings. https://lnkd.in/g_muyjnQ

26

6 Comments

In the past few years, we’ve seen a cambrian explosion of new columnar formats, challenging the hegemony of Parquet: Lance, Fastlanes, Nimble, Vortex, AnyBlox, F3 (File Format for the Future). The thinking is that the context has changed so much that the design of yore (the previous decade) is not going to cut it moving forward. This seemed a bit intriguing to me, especially since the main contribution of Parquet has been to provide a standard for columnar storage. Parquet is not simply a file format. As an open source project hosted by the ASF, it acts as a consensus building machine for the industry. Creating six new formats is not going to help with interroperability. I spent some time to understand a bit better how things actually changed and how Parquet needs to adapt to meet the demands of this new era. I’m discussing my findings on my blog: https://lnkd.in/gDzb8XH6

399

15 Comments

TECHNICAL PERFECTIONISM: THE MOST EXPENSIVE HOBBY IN STARTUPS This team spent 4 months perfecting IPMI card configurations while their competitors made millions with bash scripts. The Perfectionism Timeline: 📅 Month 1: "Optimizing orchestration architecture" 📅 Month 2: "Implementing advanced networking protocols" 📅 Month 3: "Debugging SSL certificate automation" 📅 Month 4: "Finalizing production deployment pipelines" Customer feedback during this period: "Can I rent a GPU by tomorrow?" "Why does provisioning take 30 minutes?" "Do you have a simple pricing page?" The brutal punchline: They built a Formula 1 race car when customers needed a reliable Honda Civic. OpenStack vs. Reality: OpenStack implementation: $400K, 3 months What customers wanted: Basic VMs, predictable pricing, simple dashboard Final article reveals why boring technology beats brilliant architecture every single time. #TechnicalDebt #ProductDevelopment #TechStacks #EngineeringManagement #StartupTech https://lnkd.in/gSCeWwve

315

221 Comments

Anthropic just published best practices for using Claude Code and coding agents generally. This is the best AI coding agent guide I've seen, and it applies (modulo some Claude-isms) to agentic coding with all models. The key is you must be open to adapting your prompts to "empathize" with the model. You need to work /with/ the model. Everyone tiptoes around this suggestion because it sounds like blaming the user: "the product should handle this automatically, don't make the user adapt!", they say. Meanwhile, the devs who adapt get immense value. https://lnkd.in/eGthJUMf

229

4 Comments

Despite the many benefits of Data #CleanRooms they may turn participating parties into Data Brokers as per California’s DELETE Act if they happen to bring 3rd party data (no direct relationship) into the platform, triggering a registration and tech alignment requirement (and hefty fines if dates are missed). (Presentation by California Privacy Protection Agency and Ben Isaacson on the DELETE ACT and the new #DROP platform that integrates all data brokers for consumers to place a single request.) Privacy Law Section of the California Lawyers Association. #AdTech

54

4 Comments

Great piece by Emilia David on the LangChain open source integration ecosystem, the power of LangGraph, and our newly released LangGraph Platform Helps put everything you see us working in perspective Link: https://lnkd.in/ghajS7zx

100

5 Comments

"Graphs make provenance first-class. Instead of burying "where did this come from?" inside a dense vector, a graph records sources, activities, and agents as nodes and edges you can query..." https://lnkd.in/efAdV9dx

26

🚨Our obsession with aligning computer computation to human cognition is almost absurd. 👉🏻Not from human self-importance, but from a misunderstanding. AI’s functionality is fundamentally antithetical to how humans think. 🙂The opportunity isn’t alignment—it’s contrast. We should celebrate the difference and understand it through a cognitive parallax. https://lnkd.in/eUGSDaCH #AI #LLMs #cognition

37

6 Comments

LLMs and Codegen have collapsed the time and cost of building new software and much can be generated. Now, more emphasis is needed on the ability to scope requirements and write specifications. Specifying intent can be incredibly difficult given the fuzziness and ambiguity of natural language. Maybe the new 'time to build' is actually 'time to think'.

30

6 Comments

New peer-reviewed paper: John Snow Labs medical text #deidentification compared to AWS, Azure, GPT-4o, and Claude 2.7 Sonnet. John Snow Labs is the only one that reaches human-level accuracy (>95%), while being 80% cheaper than Azure and >90% cheaper than GPT-4o. Paper: https://lnkd.in/gt7BR5rJ (accepted to Text2Story at ECIR 2025) #ai #generativeai #datascience #nlproc #textmining #dataprivacy #healthcareai #healthai

83

2 Comments

Fixinator version 6.1 has dropped. One of the key features in this release is the enhancement of its ColdFusion compatibility code scanner to look for usage of undefined remote arguments. This breaking change was part of the May ColdFusion Security Hotfix (ColdFusion 2025.2, CF 2023.14, and CF 2021.20): https://lnkd.in/gCGQY7qE

18

"Claude Code costs up to $200 a month. Goose does the same thing for free. Claude Code, Anthropic's terminal-based AI agent that can write, debug, and deploy code autonomously, has captured the imagination of software developers worldwide. But its pricing — ranging from $20 to $200 per month depending on usage — has sparked a growing rebellion among the very programmers it aims to serve. Now, a free alternative is gaining traction. Goose, an open-source AI agent developed by Block (the financial technology company formerly known as Square), offers nearly identical functionality to Claude Code but runs entirely on a user's local machine. No subscription fees. No cloud dependency. No rate limits that reset every five hours." Worth a look. https://lnkd.in/gAVijtMi

103

79 Comments

Openclaw / Clawdbot / Moltbot are overrated. But things are moving so fast and we should be reviewing our entire personal AI / Agent workflow and tools every few months. 6 months ago I set up my VPS on Hostinger with Claude code before Claude Cloud Coding launched and before OpenClaw. This was before Skills, MCP was barely useful and not many integrations. Persistent memory wasn’t really discussed, and tools and best practices weren’t available. Today I’m setting up a new VPS for scratch and it feels much lighter. I’ve set up these things since the beginning: - Termius for Desktop / Mobile - Claude Code - Jina AI v4 Embeddings per user - Local vector database for memory retrieval - Brave MCP for search (free) / Perplexity is better but has cost - Context7 MCP for up to date libraries Now I start with: Strongest Coding Partner Persistent Memory with efficient retrieval Up to date information on libraries and recent changes And I can build custom skills from here to tailor to my needs. There are amazing subagent libraries and skills libraries here you can use: https://lnkd.in/gcsarCpY https://lnkd.in/g-nzjeT9 Share this with those who are setting up or refreshing their agent stack today.

16

2 Comments