🤖 What's your organization's #AI strategy? You don't have one? We can help! Written AI strategy + 2-3 production workflows... in weeks, not months. No data science department needed. www.ombulabs.ai/ai-strategy-... #AIStrategy #AI #GenAI

What's gradient boosting? Each tree trains on the residuals of the previous ones. It's a relay race, not voting. go.ombulabs.com/wed #MachineLearning #DataScience #AI

🔥 AI tool adoption without an assessment burns licenses and creates security gaps. Our 4-week AI Readiness Assessment ships a roadmap, security framework, and config files you own. Is your engineering org ready for a 100% AI-DLC? go.upgradejs.com/ip8 #AI #SDLC #Security #DevSecOps

How to implement ReAct agents in #Ruby: a THOUGHT/ACTION/OBSERVATION loop with the Anthropic SDK and JSON-schema tool definitions. Reasoning + acting in one pass. go.upgradejs.com/3d0 #AI #ReAct #Anthropic

How to implement semantic search with Sequel + pgvector: 1536-dim ada-002 embeddings, ivfflat index, <=> nearest-neighbor, 0.75 cutoff to drop noise. go.upgradejs.com/pp4 #pgvector #sequel #AI

How to build your own brand-aware LLM newsletter summaries in Ruby: Sinatra UI, pgvector retrieval of 3 nearest past summaries, critic-LLM review loop. go.upgradejs.com/od1 #AI #Sinatra #pgvector #LLM

Parallax wanted a predictive ML model for hourly resource allocation. We shipped a regression model in Python (FastAPI + scikit-learn + MLflow on Azure) that forecasts daily hours per role with confidence intervals. go.ombulabs.com/ree #ML #AI #Python #Azure

Built a GenAI card generator for EO Philadelphia in days, not months: FastAPI + LlamaIndex + Gemini 3 Pro. go.ombulabs.com/tkl #GenAI

Williams College ran RAG over 822 historical syllabi with Ollama + OpenWebUI, fully offline. Local AI in higher ed is more achievable than people think. go.ombulabs.com/dew #HigherEd #LocalAI

UCLA's account provisioning chatbot used custom GPTs grounded in Responsible AI Principles. Trust framework was part of the spec, not bolted on. go.ombulabs.com/k7o #HigherEd #ResponsibleAI

A 5-tier framework for safely using AI: level 1 is policies + managed platforms, level 5 is self-hosted. Pick the tier before the tool. go.upgradejs.com/mce #AISafety #DataPrivacy

Made a Rails Superhero Card Generator: gpt-4o-mini for the name, gpt-image-1 for the portrait, LlamaIndex tying it together. Code on GitHub. go.ombulabs.com/uos #GenAI #LlamaIndex

Hot take: most AI initiatives fail because they're automating work that shouldn't exist. Start with friction, not the model. www.ombulabs.ai/blog/finding...

PMs: WCAG 2.2, plain English. POUR principles, free tools, AI helpers. No law-firm-style jargon. www.ombulabs.ai/blog/wcag-ac...

Workflows > agents, especially when a human reads the output. Case study: how we automated our quarterly maintenance reports without handing the wheel to a single agent. www.ombulabs.ai/blog/automat...

Now David presents his talk on LLM Telemetry & Evals as First-Class Rails Concerns at @blueridgeruby.com “How do we evaluate the quality of the responses of a model?” RubricLLM is a tool to help us judge the answers provided by a model. #llm #telemetry #genAI #blueridgeruby

At the same time, we have been building our own AI-powered tooling at @ombulabs.ai and doing extensive research on how AI can help teams upgrade applications faster. If you want an AI-enhanced #Rails Upgrade Roadmap 👉 www.fastruby.io/automated-ro... #AI #GenAI

It was great catching up with Ernesto! I was particularly interested in talking to him about DX and a product idea I had last year for measuring qualitative aspects of working in a codebase.

Next in our #MachineLearning series: We build our very own decision tree (a Random Forest!) using #Ruby and Scikit-learn 🐍 This is a very popular tree-based ensemble model. Check it out 👉 www.ombulabs.ai/blog/introdu...   #AI #ML #RandomForest #Python

The AI assistant we built for the #FastRuby newsletter has helped streamline our workflow, allowing our marketing team to focus on the creative aspects of curation while automating the repetitive tasks of gathering and summarizing links.

Let’s walk through the ReAct pattern and how we can use it to build a simple AI Agent that writes blog posts in Ruby. #AI #ArtificialIntelligence #ReAct #Ruby go.ombulabs.com/aoz

As AI becomes increasingly integrated into business operations, organizations face a critical challenge: how to leverage the power of Large Language Models (LLMs) while maintaining the privacy and security of sensitive data. #AI #LLM #DevSecOps #CISO #Security go.ombulabs.com/7n6

How ready is your dev team for AI? The AI revolution is transforming how development teams work, and engineering leaders are grappling with critical questions: Which AI tools will improve our engineering processes? How do we integrate AI without disrupting productivity? go.ombulabs.com/vun #AI

🤖 Still wondering how our AI-enhanced Rails Upgrade Roadmap works? Watch our latest demo... www.youtube.com/watch?v=mDjB... #AI #TechDebt #RubyOnRails #LegacyCode

- What's coming down the pipeline as we work on internal tooling that will leverage all the work with #AI and #ML that we are doing at @ombulabs.ai - How Developer Experience ( #DX ) can help engineering teams prioritize #TechnicalDebt remediation efforts

💸 Don’t have $50,000 to upgrade your #Ruby application? Neither do we! Check out our monthly, fixed-cost, top-notch upgrade plan (starting at $2,000/month) 🚀 go.fastruby.io/9ct #TechDebt #InfoSec #DevSecOps #RailsUpgrade

Interested in AI/ML and #Ruby? Amanda's article focuses on decision trees, namely Classification And Regression Trees, and the mathematics behind them. go.fastruby.io/a78 #AI #ML #DataScience

Next in our #MachineLearning series: We build our very own decision tree (a Random Forest!) using #Ruby and Scikit-learn 🐍 This is a very popular tree-based ensemble model. Check it out 👉 go.fastruby.io/l1y  #AI #ML #RandomForest #Python

I’m so grateful for @rubycentral.org and #RailsConf: It was great to meet old friends and make new #RubyFriends! It’s amazing to be part of this community, both inspiring and inclusive. I hope we can meet again soon at other regional conferences! ♥️ #RailsConf2025

Looking forward to connecting at #RailsConf! 🚀 You can find us on Wednesday, July 9th, at the #RailsConf2025 Hack Spaces to pair on some of our favorite #OpenSource projects.

What do `next_rails`, `skunk`, and `RailsBump` have in common? All #opensource tools maintained by the #FastRuby team. Ready to ship your next upgrade? Check out 8 tools they use in their fixed-cost monthly maintenance services for #rails applications 👉 go.fastruby.io/os-tools

🤖 Excited for tonight's #ArtificialRuby meetup at @betaworks.com! Join Amanda Bizzinotto and a great group of folks to explore the amazing things people are building with #AI and #Ruby. Tonight Amanda is going to show a 10' demo titled "Talk to Me: Building an AI-powered Slackbot"

One of the biggest challenges of advising students is architecting AI solutions that get the big picture of a student’s performance, struggles, and vision of success. #AI #NERCOMP #NERCOMP25 #LMS #HigherEd

Anna Ahearn and Krithik Udayashankar from UCLA are sharing their process for implementing AI-enhanced tools to help with account provisioning. Their vision was to introduce an AI-powered chatbot to assist employees with questions about account provisioning. #nercomp #nercomp25 #ai #highered

🤖 How do we mitigate the bias of AI models? Interesting discussions in this session. If we know AI is biased because of the training data, how can we improve the usage of its data? #AI #NERCOMP25 #nercomp

Great way to start NERCOMP: Keep your data close and your #AI closer at #nercomp25. 3 case studies about solutions that couldn’t leverage cloud tools and required solutions that were locally deployed and accessed. - Ollama - OpenVINO - Raspberry Pi + llama #AI #nercomp25 #nercomp #highered