Building Enterprise AI Tools: From Concept to Production

Organizations rush to implement AI but see minimal ROI by treating it as a feature rather than a fundamental shift in how teams work. This gap costs companies millions in wasted licenses and missed opportunities.

This session cuts through the hype to show our proven approach for developing AI-powered solutions that integrate seamlessly with existing enterprise systems. We'll demonstrate real tools built for clients, including Tessera—our AI solution that transforms project management by intelligently generating comprehensive Jira epics, issues, and tasks from high-level requirements.

You'll see how we leverage Elvex for integration orchestration, n8n for workflow automation, and modern AI technologies to create maintainable, scalable systems. This is battle-tested methodology from enterprise development trenches, covering architecture decisions, integration patterns, and practical considerations separating successful implementations from expensive experiments.

Live demonstrations show these tools connecting with Drupal APIs and content management systems, but principles apply to any enterprise platform. You'll learn to evaluate AI use cases, design robust architectures, create intuitive interfaces, and maintain systems in real-world environments using approaches you can implement immediately.

Presented at DrupalCon Chicago 2026 by Mike Wilbur and Mark Shropshire

Transcript

1. Current currentagency.com March 26, 2026 DrupalCon Chicago 2026 Building Enterprise AI Tools From Concept to Production Mike Wilbur & Mark Shropshire
2. Today’s Speakers PROBLEM & APPROACH 2 DrupalCon Chicago 2026 Mark Shropshire Group Director, Technology Mike Wilbur Group Director, Technology
3. The AI Budget Drain DrupalCon Chicago 2026 3 PROBLEM & APPROACH Millions spent Minimal ROI The technology isn’t the problem
4. The hype trap OUR PHILOSOPHY DrupalCon Chicago 2026 4 Who are we? CEOs What do we want? AI AI to do what? We don’t know When do we want it? RIGHT NOW! “AI will solve everything.” AI is good at specific, well-defined tasks. Not magic Not autonomous Not a strategy The difference The gap between these two beliefs is where budgets disappear VS
5. OUR PHILOSOPHY DrupalCon Chicago 2026 5 AI Hype Perfect Automation Idealistic Visions Unbridled Optimism Overpromisses Exaggerated Claims Overstated Dangers AI Fear Algorithmic Control Dystopian Concerns Job Loss Bias & Prejudice Human-centric Design Responsible Development Informed Adoption Ethical Principles Realistic Expectations Balancing The Narrative: Navigating AI Hype and Fear THE Sweet Spot
6. What you’ll walk away with PROBLEM & APPROACH 6 DrupalCon Chicago 2026 architect enterprise AI tools using real integration patterns How to 01 evaluate AI use cases before you build How to 02 apply proven patterns and ship production-ready solutions How to 03
7. Terms we’ll use Today 7 PROBLEM & APPROACH DrupalCon Chicago 2026 API (Application Programming Interface) how software systems talk to each other. LLM (Large Language Model an AI engine that understands and generates text. Workflow Automation software that executes repeatable processes without manual intervention. Orchestration coordinating multiple tools and services to work together as one system. “Work Slop” low-quality, AI-generated content that lacks depth, accuracy, or real substance. MCP (Model Context Protocol) a standard for connecting AI to external tools and data sources.
8. Our Philosophy Building Enterprise AI Tools From Concept to Production 8
9. What’s the shift that changes everything? 9 Automate the mundane. 01 Preserve human judgment. 02 Build systems that make humans better at their jobs. 03 This is what the rest of the talk is about. PROBLEM & APPROACH DrupalCon Chicago 2026
10. Automate the mundane. Free humans for what matters. OUR PHILOSOPHY DrupalCon Chicago 2026 10 Complex, contextual thinking Humans excel at AI excels at The goal Pattern recognition at scale Redirect humans to the work only they can do
11. ROI isn’t theoretical. It’s measurable. OUR PHILOSOPHY DrupalCon Chicago 2026 11
12. We didn’t just theorize this. We built it. Meet Tessera an AI agent that transforms raw project requirements into structured, testable Jira tickets. Built for our teams. Refined in production. Used on real client work every day. This is the case study. 12 DrupalCon Chicago 2026 OUR PHILOSOPHY
13. CASE STUDY 13 Case Study - Building Tessera Building Enterprise AI Tools From Concept to Production DrupalCon Chicago 2026
14. Why this matters for your Drupal work. 14 Not everything in Drupal needs AI. But AI can make your Drupal work better. The tools are platform-agnostic. The opportunity for your workflow is specific. DrupalCon Chicago 2026 CASE STUDY
15. Every project. Same problem. So we built Tessera. 15 1. Requirements come in. 2. Tickets need to happen. 3. Hours disappear. CASE STUDY DrupalCon Chicago 2026 Input: requirements in any form Output: Structured epics, stories, and tasks ready for Jira.
16. How it’s built Platform-agnostic by design. 16 Agent Layer LLM Configurable Orchestration Output: Jira CASE STUDY DrupalCon Chicago 2026
17. AI generates the structure. Humans make the calls. 17 CASE STUDY Tessera produces the output Generate 01 A human reviews, refines, and approves Review 02 Only then does it enter the backlog Ship 03 DrupalCon Chicago 2026 This is what separates useful automation from workslop.
18. What changed when we shipped it. 18 CASE STUDY Ticket creation time reduced significantly Minutes instead of hours per project kickoff Speed Same ticket quality, same format, same best practices — every project, every team, every sprint 30+ concurrent workstreams, one standard Consistency Technical leads redirected to architecture, strategy, and client relationships Focus We didn't eliminate any roles. We changed where people spend their time. DrupalCon Chicago 2026
19. Tessera v1 vs. v2 19 CASE STUDY DrupalCon Chicago 2026 Same philosophy. Different engine. This is what platform-agnostic actually means in practice. Built the pattern, proved the concept, validated the architecture V1 V2 Same architecture, moved the agent layer, expanded capabilities
20. Tessera in Action Building Enterprise AI Tools From Concept to Production
26. DEMO Drupal + Tessera Integration New Feature Work 26 DrupalCon Chicago 2026 Drupal Site n8n Workflow Tessera JSON:API Field and Schemas Relationships Fetch Content Enrich Prompt Route Output Context-aware Gen Epic / Story Builder Human Review Gate New Feature Request Jira
27. IMPLEMENTATION 27 Implementation Building Enterprise AI Tools From Concept to Production DrupalCon Chicago 2026
28. Before you build anything, run it through this framework. 28 Strong candidate • High volume — happens constantly • Clear patterns — similar inputs, similar outputs • Defined output — you know what done looks like • Low judgment — verifiable without deep expertise Poor candidate • Low volume — rare or one-off • High variance — every instance is different • Ambiguous output — "done" is subjective • High judgment — requires context only humans carry DrupalCon Chicago 2026 The mundane test: tedious but straightforward? Patterns across inputs? Output is verifiable? If yes to all three — build it. IMPLEMENTATION
29. The stack that actually holds up in production. 29 Orchestration layer first API-first design Fallback + version control Human review gates IMPLEMENTATION DrupalCon Chicago 2026
30. Don't learn these the hard way. 30 DrupalCon Chicago 2026 IMPLEMENTATION workslop erodes trust fast and it’s hard to rebuild ✕ Skip human review validate the core use case first, then expand ✕ Over-engineer for edge cases it can't read your mind ✕ Expect AI to understand business context time saved, errors caught, consistency achieved. Not theoretical gains. ✓ Measure actual impact version your prompts, monitor for drift, document your review processes ✓ Maintain like it’s code build for swap, not lock-in. Agents, models, and tools are evolving fast ✓ Keep things modular
31. 31 Q&A Building Enterprise AI Tools From Concept to Production Scan QR Code and give us feedback! Mike Wilbur mike.wilbur@currentagency.com Mark Shropshire mark.shropshire@currentagency.com