# 2 Acre Studios — Full Site Content > Pittsburgh AI lab building private, local AI systems. 60+ open source repos. Zero vendor lock-in. Est. 2010. ## Navigation - [Manifesto](#manifesto) - [Capabilities](#capabilities) - [Lab](#lab) - [Team](#team) - [Terminal](#terminal) - [AI ROI Calculator](#calculator) - [FAQ](#faq) - [Contact](#signal) ## Agent-Readable Files - `/llms.txt` — Summary for AI discovery (this spec: llmstxt.org) - `/llms-full.txt` — Complete site content as markdown (you are here) - `/agent.json` — Structured data payload (JSON) --- ## Hero **WE BUILD AI THAT ACTUALLY WORKS.** Pittsburgh AI Lab. 60+ Open Source Projects. Zero Vendor Lock-in. Most AI projects fail because teams build demos instead of production systems. We don't do demos. --- ## Manifesto **WE BUILD. THEN WE SHIP.** Since 2010, we've operated like a lab — hands on the keyboard, not just the whiteboard. When we advise, it's because we've already built it ourselves. Every line of code we write ships. Every model we train solves a real problem. Every system we deploy runs on YOUR hardware, with YOUR data, under YOUR control. > We believe AI should be private, local, and owned by the people who use it. 60+ open source repositories. 624 GitHub stars. 2 shipped AI products. Zero bullshit. --- ## Capabilities ### 01 — Local AI Deployment Private, on-premise AI systems running on your hardware. We deploy Ollama-based large language models behind your firewall — no data leaves your building, ever. Your IT team gets full control over model selection, fine-tuning, and access policies. We handle the infrastructure: GPU allocation, model optimization, load balancing, and monitoring. Most deployments are production-ready within two weeks. Ideal for healthcare, finance, legal, and any industry where data sovereignty is non-negotiable. Supports models from 7B to 70B+ parameters running on consumer GPUs or enterprise hardware. ### 02 — Multi-Agent Systems Autonomous agent workflows that actually complete tasks. Not chatbots — workers. We build multi-agent systems on AutoGen and CrewAI where specialized AI agents collaborate to solve complex business problems: research agents gather information, analyst agents process it, and execution agents take action. Each agent has defined roles, tools, and guardrails. We've shipped agent teams that handle everything from competitive intelligence gathering to automated code review to content production pipelines. The result is work that gets done while you sleep — with audit trails, error handling, and human-in-the-loop checkpoints where you need them. ### 03 — Sales Enablement AI-powered proposal generation, lead scoring, and pipeline intelligence that delivers real revenue impact. Our systems cut proposal creation time by 40% by automating research, competitive positioning, and document assembly. Lead scoring models analyze behavioral signals, firmographic data, and engagement patterns to surface your highest-probability deals. Pipeline intelligence dashboards give sales leaders real-time visibility into deal health, risk factors, and forecast accuracy. Built on RAG pipelines that learn from your winning proposals and adapt to your specific sales methodology — whether that's MEDDIC, Challenger, or something custom. ### 04 — Document Intelligence Turn your document chaos into a searchable, queryable knowledge base. We build RAG (Retrieval-Augmented Generation) pipelines that process thousands of documents in hours — PDFs, Word files, spreadsheets, emails, Slack threads, whatever you have. Your team can then ask natural language questions and get accurate, sourced answers in seconds instead of hours of manual searching. Vector database backends ensure fast retrieval across millions of document chunks. We handle OCR for scanned documents, table extraction, and multi-language support. Reduce document processing time by 70% while making institutional knowledge accessible to everyone on your team. ### 05 — Customer Support AI Intelligent ticket triage, response generation, and escalation at $0.50 per interaction versus $15+ for human support. Our systems automatically classify incoming tickets by urgency, topic, and sentiment, then route them to the right team or generate draft responses for agent review. We achieve 65% automation rates on Tier 1 support while maintaining customer satisfaction scores. Complex issues get escalated with full context summaries so human agents never start from scratch. Integrates with Zendesk, Freshdesk, Intercom, and custom helpdesk systems. Built-in analytics track resolution times, automation rates, and cost savings in real time. ### 06 — Custom Development Full-stack web applications, APIs, and integrations — built in Python, JavaScript, or whatever the problem needs. We've shipped production systems for Fortune 500 companies and startups alike. Our approach: understand the problem first, build the simplest thing that works, then iterate based on real usage data. We specialize in AI-integrated applications where traditional development meets machine learning — think internal tools with natural language interfaces, automated workflows with human oversight, and data pipelines that feed both dashboards and AI models. Every project includes proper error handling, monitoring, documentation, and deployment automation. No throwaway code. Everything ships. --- ## Lab — Open Source Projects ### Ollama-Workbench - URL: https://github.com/marc-shade/Ollama-Workbench - Stars: 47 - Language: Python - Description: Comprehensive platform for managing and testing local Ollama models ### TeamForgeAI - URL: https://github.com/marc-shade/TeamForgeAI - Stars: 27 - Language: Python - Description: AI agent framework for managing teams of agents with common goals ### ai-persona-lab - URL: https://github.com/marc-shade/ai-persona-lab - Stars: 8 - Language: Python - Description: Create and manage dynamic AI personas for interactive group chats ### Reddit-Marketing - URL: https://github.com/marc-shade/Reddit-Marketing-Assistant-Workflow - Stars: 7 - Language: n8n - Description: n8n workflow for identifying marketing leads from Reddit posts **Stats: 60+ repos | 624 stars | 2 products shipped | 100% open source** --- ## Team ### Marc Shade — Agentic AI Chief Engineer Building software since the 1980s. Spent 20 years shipping for corporate clients — Kellogg's, Hertz, Stryker, ConAgra — with teams from Leo Burnett and Arc Worldwide. Founded 2 Acre Studios in 2010. Pivoted to AI in 2023. Now builds private, local AI systems — multi-agent workflows, RAG pipelines, and full-stack applications. 60+ open source repos. ARC-AGI-3 competitor. Zero tolerance for vendor lock-in. - GitHub: https://github.com/marc-shade - LinkedIn: https://www.linkedin.com/in/marcshade ### Scott Frederick Laughlin — Lead AI/ML Engineer / Cloud Architect Tech entrepreneur and cloud architect with a decade of pioneering AI SaaS, IoT solutions, and multi-cloud infrastructure. Founder of TechRamp. Led high-impact projects for Fortune 500 enterprises — advanced AI agents, large-scale data processing, and cloud-connected product architectures. Expertise: generative AI, AI-assisted consulting, IoT, and next-generation cloud systems. - LinkedIn: https://www.linkedin.com/in/scott-engineer-inventor --- ## AI ROI Calculator Estimate your return on AI investment using industry-validated metrics. ### Inputs - Company size (1-10, 11-50, 51-200, 201-500, 500+) - Industry (Technology, Healthcare, Finance, Manufacturing, Retail, Professional Services, Other) - Monthly customer support tickets - Hours spent on document processing per week - Monthly sales proposals generated - Average employee hourly cost (default: $45) ### Calculation Methodology **Support Automation Savings:** tickets × 65% automation rate × ($15 human - $0.50 AI) × 12 months × industry multiplier **Document Processing Savings:** hours/week × 70% reduction × hourly cost × 52 weeks × industry multiplier **Sales Enablement Impact:** proposals × 40% speed improvement × 4 hours saved × hourly cost × 12 months × industry multiplier **Implementation Costs by Company Size:** - 1-10 employees: $5K-$15K - 11-50 employees: $15K-$50K - 51-200 employees: $50K-$100K - 201-500 employees: $100K-$200K - 500+ employees: $200K-$500K **Industry Multipliers:** Technology (1.0×), Healthcare (1.15×), Finance (1.20×), Manufacturing (0.95×), Retail (0.90×), Professional Services (1.05×) **Data Sources:** McKinsey, IBM, Gartner, Industry Averages --- ## FAQ ### What is local AI deployment and why does it matter? Local AI deployment means running large language models and other AI systems on hardware you own and control — inside your data center, office, or private cloud. Unlike cloud AI services where your data is sent to third-party servers, local deployment keeps everything behind your firewall. This matters for industries with strict data regulations (healthcare, finance, legal, government) and for any organization that considers its data a competitive advantage. We use Ollama-based infrastructure that runs models from 7B to 70B+ parameters on standard GPU hardware, with no per-query API costs and no vendor lock-in. ### How much does AI customer support cost compared to human support? Our AI customer support systems cost approximately $0.50 per interaction compared to $15 or more for human-handled support tickets. At a 65% automation rate for Tier 1 inquiries, a company handling 1,000 tickets per month saves roughly $113,000 annually. The AI handles ticket classification, response drafting, sentiment analysis, and routing. Complex issues are escalated to human agents with full context summaries. Implementation costs range from $5,000 for small teams to $200,000+ for enterprise deployments, with most companies seeing full ROI within 3-6 months. ### What is a multi-agent system and how is it different from a chatbot? A multi-agent system uses multiple specialized AI agents that collaborate to complete complex tasks autonomously. Unlike a chatbot that responds to one query at a time, a multi-agent system assigns roles — researcher, analyst, writer, reviewer — and agents work together through defined workflows. We build these on AutoGen and CrewAI frameworks with guardrails, error handling, and human-in-the-loop checkpoints. The result is AI that does work, not just answers questions. ### What is RAG and why do you use it for document intelligence? RAG (Retrieval-Augmented Generation) is a technique that connects a large language model to your specific documents and data. Instead of relying on the model's training data alone, RAG retrieves relevant passages from your knowledge base and uses them to generate accurate, sourced answers. We use RAG because it dramatically reduces hallucinations, provides citations for every answer, and works with any document format — PDFs, Word files, spreadsheets, emails, Slack threads. ### How long does it take to deploy a private AI system? Most local AI deployments are production-ready within two to four weeks. Week one covers infrastructure assessment, model selection, and environment setup. Week two handles model deployment, fine-tuning, and integration with your existing systems. Weeks three and four focus on testing, optimization, and team training. More complex projects — like multi-agent systems or enterprise-wide document intelligence platforms — typically take four to eight weeks. ### What hardware do I need to run local AI models? For small to medium deployments (7B-13B parameter models), a single workstation with an NVIDIA RTX 3090 or 4090 GPU (24GB VRAM) is sufficient. For larger models (30B-70B parameters), you'll need enterprise GPUs like the A100 or H100, or multiple consumer GPUs. We also support Apple Silicon deployments (M2/M3/M4 Ultra) for organizations using Mac infrastructure. ### Do you offer ongoing support after deployment? Yes. Every deployment includes 30 days of post-launch support covering monitoring, optimization, and issue resolution. After that, we offer ongoing maintenance retainers that include model updates, performance tuning, security patches, and scaling support. We also provide training for your technical team so they can manage day-to-day operations independently. ### How is 2 Acre Studios different from other AI consulting firms? Three things set us apart. First, we build and ship — 60+ open source repositories, 624 GitHub stars, and two shipped AI products. We're not advisors who've never written production code. Second, we prioritize private, local AI over cloud dependencies. Your data stays on your hardware, under your control, with zero vendor lock-in. Third, we've been building software since the 1980s and shipping for Fortune 500 clients since 2006. ### What industries do you work with? We work across technology, healthcare, finance, manufacturing, retail, and professional services. Our AI ROI calculator applies industry-specific multipliers: finance and healthcare see higher returns due to the premium on data privacy and compliance automation. Manufacturing and retail benefit most from document intelligence and supply chain optimization. Technology companies typically engage us for multi-agent systems and developer tooling. --- ## Contact **READY TO BUILD SOMETHING REAL?** No pitch decks. No 6-month discovery phase. We start building in week one. - Email: marc@2acrestudios.com - Phone: (412) 407-6170 - Location: Pittsburgh, PA, USA - GitHub: https://github.com/marc-shade - LinkedIn: https://linkedin.com/company/2-acre-studios - Discord: https://discord.gg/9DvzfsrYFA --- *2 Acre Studios / Pittsburgh, PA / Est. 2010* *Built with code. Powered by local AI. Owned by us.* *This site runs zero trackers. No cookies. No analytics. Just code.*