Work

Things I’ve been exploring.

I like solving complicated problems.

Actually, let me correct that.

I like figuring out why something became complicated in the first place. 😂

After more than 20 years working with business processes, enterprise systems, and technology, I’ve learned that many difficult problems share surprisingly familiar patterns.

So I’ve been exploring how Business AI Architecture and an AI-driven development lifecycle (AI-DLC) could solve them differently. Each area below sits somewhere between canvas and customer.

01

Concept exploration

What if a BPO could scale without growing headcount at the same rate?

ClarisTXM · Total Execution & Experience Management

Better Service Delivery. Lower Cost to Serve. More Room to Grow.

ClarisTXM as an operating layer Concept diagram. Clients, services and SLAs sit on top. In the middle, the ClarisTXM execution and experience layer orchestrates people, workflows, automation and AI; governs approvals, exceptions and evidence; and keeps service quality, performance and cost-to-serve visible, with people setting direction and owning outcomes. Systems of record such as CRM, ERP, ticketing and data sit underneath. Clients · services · SLAs · experience ClarisTXM · execution & experience layer Human in the lead: people direct, approve, own outcomes Orchestrate· People· Workflows· Automation· AI Govern· Approvals· Exceptions· Evidence· Accountability Experience· Service quality· Performance· Cost-to-serve· Capacity Systems of record: CRM · ERP · ticketing · data

Picture a business process outsourcing provider serving multiple clients across finance, customer operations, procurement, and shared services.

Each client has different processes, service levels, systems, exceptions, and reporting needs. Adding another client shouldn’t always mean adding the same proportion of people and complexity.

ClarisTXM is my independent concept for exploring a different operating model: a shared execution and experience layer connecting people, work, automation, and AI.

Three promises to BPO providers

Deliver better services. Meet client commitments, improve quality, and manage work consistently across accounts.

Improve profitability. Reduce manual effort and cost per transaction while protecting service quality.

Grow without growing costs at the same rate. Take on more clients and higher work volumes using better processes, automation, and AI.

What would it help manage?

Work: cases, tasks, queues, exceptions, and service commitments.

People: specialists, supervisors, decisions, and accountability.

AI and automation: approved assistance and workflow actions with appropriate controls.

Experience: service quality, client visibility, employee experience, and continuous improvement.

What would success look like?

Improved productivity, consistent quality, measurable SLAs, lower cost-to-serve, and capacity to scale without proportional headcount growth. These are proposed outcomes to validate, not measured product results.

This is a conceptual, independently presented exploration. No employer or client implementations, source code, or confidential information are represented here.

Explore the architecture approach

02

What if AI had boundaries?

1 Who is asking?2 What may it touch?3 Approved actions only4 Re-check before it changes data5 Write it all down

Everyone wants AI agents that can do more. I’m interested in what they should be allowed to do.

Read information? Sure, with appropriate access. Recommend an action? Possibly. Change financial records without approval? Yeah, we should probably talk about that. 😂

I explore architectures that combine identity, permissions, approved capabilities, transaction controls, and auditability.

Making AI useful is important. Making it accountable is just as important.

  • Governed AI integration
  • Identity & permissions
  • Transaction controls
  • Audit

AI-DLC stage · Architecture → Test · approve Exploration · independent work

03

What if a business conversation could become a working application?

ConversationRequirementsDesignBuildTestLive shorter distance, same engineering

A business owner describes a problem. Someone writes requirements. Someone designs the solution. Someone builds it. Someone tests it. Eventually, something goes into production.

What if we could make that journey simpler?

I’m interested in how AI-assisted requirements engineering, reusable architecture patterns, and software development tools can help reduce the distance between an idea and a working solution.

Not by skipping engineering. By making engineering more accessible.

  • AI-assisted solution engineering
  • Requirements
  • Architecture patterns

AI-DLC stage · Canvas → AI builds (the whole lifecycle) Exploration · independent work

04

What if AI could check its own homework?

not approved? back to the start AI writesAI reviewsTests runHuman approves

AI can generate software. It can also generate mistakes.

So what happens when we combine AI-assisted development with independent review, automated testing, security checks, and human approval?

Can we make software development faster without making it less reliable?

That’s a question worth testing.

  • AI software engineering
  • Independent review
  • Automated testing
  • Human approval

AI-DLC stage · AI builds → Test · approve Exploration · independent work

05

What if smaller businesses had access to better architecture?

enterprise budget the processthe datathe right tool sized for the business →

Large enterprises spend significant resources designing how their systems should work.

Small businesses have many of the same operational challenges. Just with smaller budgets.

What if we could simplify enterprise architecture into practical methods, reusable components, and affordable solutions?

Not enterprise complexity at a discount. Good engineering, sized appropriately for the business.

  • Right-sized architecture
  • Reusable components
  • Small business

AI-DLC stage · Architecture, sized for the business Exploration · independent work

Why I’m sharing this.

These topics aren’t presented as proprietary products or client case studies. They’re areas of engineering and business architecture that interest me.

Where I publish a demonstration, template, or tool, I’ll distinguish independently developed examples from conceptual designs and published research. Illustrations on this page use fictional businesses and synthetic data.

My hope is that someone finds an idea useful, improves on it, and builds something of their own. That would be a pretty good outcome.