Aram Drevekenin’s Post

Deciding which punctuation to use at work...,?!!!!!"-":) I don't know if anyone else feels like this (especially with the increasing use of AI language tools) but this definitely resonated with me...! credit: johnschauster

Almost half of the time saved using AI is spent correcting outputs: Inconsistent outputs from large language models are introducing hours of additional labor for workers. http://dlvr.it/TQTSJm

Almost half of the time saved using AI is spent correcting outputs: Inconsistent outputs from large language models are introducing hours of additional labor for workers. http://dlvr.it/TQTSK2

Almost half of the time saved using AI is spent correcting outputs: Inconsistent outputs from large language models are introducing hours of additional labor for workers. http://dlvr.it/TQTSJt

I'm so sick of seeing this AI construct: it's not (about) X it's (about) Y I understand that in some cases (as hopefully in this one) there's an author's original thought behind it, but I cannot force myself into reading anything that is introduced like that dear gods of AI please hear me out - there are other ways to express ideas, please add them to your AIs because users won't change what your LLMs spit out

Isn't it ironic that the people who promote the idea that the future is to tell an AI what to build in "plain English" and let the AI write the code, (because everyone can speak "plain English"!) are the same ones that don't even write their own posts, but they delegate that task to AI?

AI conversations sometimes feel like everyone else got a glossary you never received. Context window. Temperature. Tokens. RAG. Agentic....Most definitions use three more terms you also don't know. I got tired of this, so I built a simple AI glossary on my website. Plain English. No assumed knowledge. There's a flashcard mode too if you want to test yourself. - Link in the comments. If you're a senior professional who's been quietly nodding along in AI conversations while Googling terms under the table, this might help. What's one AI term you wish someone would just explain simply? #BusinessFirstAI #AICoaching #AILeadership

Over the past days, something clicked for me about AI and legacy IT modernization. Recursive Language Models (RLMs) show how to break gigantic, business-critical codebases into recursive, fresh-context steps and then reassemble precise, architectural insight. In my experiments, this drastically reduced token cost while increasing depth and reliability of understanding. If huge contexts and messy systems are your bottleneck, RLM-style agentic systems are worth a serious look. #AI #LegacyModernization #AgenticSystems #RLM #SoftwareEngineering

A practical, end-to-end course that explains how large language models work, surveys the fast-moving open-source landscape, teaches the Model Context Protocol (MCP), and walks you through modern agentic patterns—culminating in a hands-on build where you create and evaluate your own tool-using AI agent.

The problem you hear is rarely the real problem you need to solve... I recorded this video with Zain Haseeb on how we think about solving problems with AI. I thought it could be helpful because I think we overcomplicate things a lot of (all) the time. A few general principles I use: - The problem people come to you with is almost always a symptom of the problem, not the problem itself - When solving the problem you can break down the workflow into three steps: 1. Capture 2. Storage 3. Delivery - People don't want more tools, they want more progress - Ugly and simple beats fancy and complicated every time Below is a cheat sheet I put together and referenced in the video. https://lnkd.in/erEc-Tx9