>FOUNDRY_AI
MEETING_RECORD :: 2026.02.16

AI Strategy & Infrastructure Discussion

A strategic meeting between Four Leaf and Lutz Tech exploring hybrid AI infrastructure, data security, and the future of enterprise AI adoption.

Meeting Overview

This meeting brought together representatives from Four Leaf (a technology consultancy specializing in AI and infrastructure) and Lutz Tech (the technology division of Lutz, a top-70 accounting firm). The discussion centered on Lutz Tech's current AI adoption journey, challenges with cost and security, and exploration of hybrid infrastructure approaches to balance enterprise AI needs with data governance requirements.

Attendees

Four Leaf

Jason Flippin
Leadership / Founder

Background: 11-12 years at Cisco covering the Midwest region

Expertise: Cybersecurity, infrastructure, partner/VAR business models

Role: Founded Four Leaf 14 months ago on principles of value-driven, non-transactional partnerships

Eric (Mentioned)
Technical Lead

Four Leaf's main technical AI specialist who will conduct future demo of infrastructure automation

Foundry AI

Scott Oppliger
Leadership / Founder

Background: Serial entrepreneur (8 companies), first commercial ISP in Kansas City (1992), 7 years at Cisco

Current Role: Founder of Foundry AI Partners

Expertise: Cybersecurity, AI/ML (10+ years), RAG pipelines, agentic workflows

Clients: Netflix, Adidas, Goldman Sachs

Lutz Tech

Kirk Montagne
IT Operations Manager

Background: 8 years at Lutz, previously focused on software consulting and digital transformation

Role: Leads AI initiative internally and client advisory

Involvement: Member of CIO/CTO peer group for accounting firms

Ryan Wade
Software Solutions Director

Responsibilities: Oversees SOC team and security vetting processes

Focus: HIPAA compliance and data governance

Role: Handles approval requests for new AI tool implementations

Ryan Longnecker (Mentioned)
Director / Partner

Partner in charge of Lutz Tech, will be involved in future technical strategy discussions

Scott Kroeger (Mentioned)
Partner

Partner in charge of Lutz Tech, advocate for on-premise infrastructure approach

Key Points

Lutz Tech Current State

Organizational Structure

  • Approximately 400 employees total across all divisions
  • Accounting is the largest division; Technology (Lutz Tech) is the second largest with 50-70 employees
  • Serves primarily privately held businesses (50-250 employees typically)
  • Works with 5,000 businesses ranging from small firms to billion-dollar companies

Technology Services

Managed IT Services

Supports ~250 small-to-medium businesses as their full IT team

Data Analytics/Reporting

Microsoft Fabric-based platform with Power BI; operates as an outsourced reporting department (not project-based)

Software Consulting

Focuses on helping clients select and implement solutions rather than custom development

AI Advisory

Currently providing fundamental AI training to clients (prompt engineering, model selection)

Current AI Adoption

  • Employees using AI daily with usage policies in place
  • Recently rolled out Claude Enterprise to a pilot group of 6 users (4 Claude Code users, 3 general users)
  • Also deploying ChatGPT Teams accounts for better control
  • 90% of their tenant is shared across multiple client workspaces
Major Challenges Identified

Cost Concerns

  • Significant "sticker shock" discovered after moving to Claude Enterprise
  • Enterprise API rates are approximately 10x more expensive than individual Max accounts ($200/month)
  • Concern that rolling out enterprise licenses to all 70 Lutz Tech employees would create unsustainable costs
  • One example cited: a company spending $50,000/month on API calls alone

Security & Data Governance

  • Strong audit/trust ethos due to accounting firm roots
  • Handling sensitive data including financial information, SSNs, and HIPAA-protected healthcare data
  • Concerns about data leakage into public AI models, even with enterprise agreements
  • Daily requests from "upstairs" (accounting firm) for new AI tools requiring SOC team vetting
  • Philosophy of distrust: "accidents happen" even with well-intentioned cloud providers

Strategic Uncertainty

  • Pressure from firm shareholders who attend conferences and return demanding AI implementations
  • Need to master internal AI strategy by May 1st (end of tax season) before advising clients
  • Uncertainty about when to use cloud models vs. on-premise infrastructure
  • Confusion about building contextual layers, ontologies, and MCP implementations
Four Leaf's Proposed Approach

Hybrid Infrastructure Model

  • Run on-premise models (e.g., Qwen, Llama) for routine inference, low-hanging fruit, and sensitive data processing
  • Use commercial cloud models (Anthropic Claude, OpenAI) for complex tasks like coding and advanced reasoning
  • Implement model routers to intelligently direct workloads based on sensitivity and complexity
  • Leverage Cisco's purpose-built, stackable infrastructure (not DIY GPU solutions)

Data-First Strategy

  • Emphasized that data classification must be the starting point
  • Infrastructure decisions (cloud vs. on-prem vs. hybrid) flow from understanding data sensitivity
  • Compared approach to "zero trust" philosophy: secure data equally regardless of location

Phased AI Maturity

Layer 1: Personal Productivity

AI assistants for individual tasks (Lutz feels confident here)

Layer 2: Business Data Integration

Connecting AI to business data to level up teams (currently exploring ROI)

Layer 3: Agentic Workflows

Transform how work is done, not just automate current processes (future focus)

Action Items & Next Steps

Four Leaf Deliverables

Send email on data classification frameworks

To Ryan at Lutz to initiate strategic planning conversation

Prepare recommendations for hybrid AI strategy

Focus on Lutz Tech's internal needs, addressing cost optimization and security

Schedule technical demonstration with Eric

Demo: AI agents querying network infrastructure to auto-generate real-time Visio diagrams

Follow-up questions

Send additional questions as data classification and infrastructure recommendations develop

Lutz Tech Commitments

Review Four Leaf's recommendations

Evaluate readiness to proceed and provide feedback on proposed approach

Loop in Ryan Longnecker

Include him in technical strategy discussions going forward

Internal timeline: Focus on Lutz Tech through May 1st

Master internal AI strategy by end of tax season, then roll out to broader firm

Define use cases and ROI

Determine specific problems to solve with AI beyond generic "connect to our data" approach

Strategic Context

Timeline

Lutz Tech is in planning mode through tax season (now through May 1st). The accounting firm is in "autopilot" during this period, but Lutz Tech is using this time to finalize their internal AI strategy before broader rollout.

Philosophy

Lutz Tech wants to "eat their own dog food"—master AI internally before advising clients. This mirrors their successful approach with other technology transformations.

Market Position

As a top-70 accounting firm with a mature technology practice, Lutz is uniquely positioned to offer AI advisory services to their 5,000 clients, but they recognize the need to move from fundamentals (prompt training) to advanced capabilities (agentic workflows) to differentiate.