Gordon Barlow | Human-Centered Systems for AI Trust, Governance, and Workflow Architecture

Human-Centered Systems for AI Trust, Governance, and Workflow Architecture

I help leaders make complex systems easier to trust, use, and govern — especially when AI, data, policy, and human judgment all have to work together.

What I do:
I turn unclear systems into visible structures people can trust, use, explain, and improve.

Where I fit:
Where AI, data, policy, workflow, and human judgment all have to work together — and the cost of confusion is high. This often maps to principal-level systems, product, UX, AI trust, workflow architecture, governance, or decision-support roles.

What it delivers:
Clearer decisions. Fewer escalations. Safer automation. Visible accountability. Systems that support the people doing the work.

Gordon GPT opens in ChatGPT and may require sign-in.

Why my work exists

When systems drift, people compensate.

The signs are easy to recognize: duplicated work, contradictory rules, delayed decisions, unclear ownership, and avoidable escalation.

My work restores alignment between policy, workflow, data, and interface so systems support the people using them — and teams can make clearer decisions with greater confidence.

Where this fits

This work fits when the system is already important — but no longer clear enough to trust.

I am most useful when leaders need to see how policy, data, workflow, tools, and human judgment actually interact before they modernize, automate, govern, or scale.

Common situations include:

AI governance: outputs need visible sources, human review, fallback paths, and accountable escalation.

Workflow modernization: policy, queues, tools, and handoffs no longer match how work actually moves.

Decision intelligence: leaders need current, evidence-backed views instead of delayed reports and manual reconciliation.

Service design: ownership, next steps, and constraints need to be visible so people can move through the work with clarity.

How I work

I start by looking at what the system is already carrying.

Every workaround, delayed decision, ignored handoff, unclear rule, and patched process leaves a mark. Over time, those marks become friction: firefighting, technical debt, duplicated work, brittle tools, and processes that no longer match reality.

My work is to remove the unnecessary weight and restore alignment between policy, data, workflow, tools, and human needs.

Practically, that means listening to the people closest to the work, mapping where the system has drifted, and turning what I find into structures teams can build from — clearer workflows, decision paths, interface patterns, oversight points, and prototypes.

When systems drift, people end up supporting the system instead of being supported by it.

My work restores that relationship. A well-aligned system gives people clear next steps, trustworthy information, and room to use their judgment — so teams move with clarity, confidence, and momentum.

Use Cases

These examples are sanitized because much of the work occurred in defense, finance, HR, and proprietary enterprise environments. Internal artifacts, sensitive data, and implementation details are not shown.

What remains is the transferable pattern: what was unclear, what I was expected to make usable, what I did, and what became easier for people and teams to do.

AI Trust, Governance, and Data Modernization U.S. Air Force

These three examples show the same alignment pattern across different Air Force needs: making AI outputs, data, workflows, and human review clear enough for people to trust and use.

AI Postal Operations Assistant

Human-Centered Postal Assistant (RAG + Guardrails)
Retrieval-augmented assistant for base postal operations — providing sourced answers, clear handoffs, and policy-aligned guidance through a conversational interface (U.S. Air Force installation operations, 2023–2025).

What happened: Base postal operations needed a clearer way to answer recurring questions about policy, forms, hours, routing, cut-off times, and package status. Staff and patrons needed answers that matched real procedures, not generic automation.

What was expected of me: Design the user experience architecture for a dependable AI assistant that could provide sourced answers, respect operational limits, and preserve human authority at critical points.

What I did: I structured the RAG interaction flow, citation patterns, escalation rules, and human review points. This was Human-Centered Systems work: making policy, data, workflow, interface, and human judgment support each other in the actual flow of use.

What changed: Staff and patrons had a clearer path to accurate answers, fewer repeat inquiries, and better handoffs between AI assistance and human support. That is Trust Engineering in practice: people could see the source, understand the boundary, and know when a human needed to step in.

If your operations teams need AI assistance that respects policy, workflow, and human authority, connect on LinkedIn.

Provenance & Oversight

What happened: Finance and operations teams were working across multiple data sources, dashboards, and NLP-generated outputs. The information existed, but users needed to understand where answers came from, what they meant, and how much confidence to place in them.

What was expected of me: Turn existing NLP outputs into an AI-integrated interface pattern people could interpret, verify, and use — without changing the underlying data services or models.

What I did: I designed the query → source → answer experience, mapped how policy and workflow context should appear, and built frontend UI components that brought provenance forward. This was UX Architecture applied to data-driven systems: making sources, explanations, dashboard context, and human review points visible in the flow of work.

What changed: Operators, analysts, and leaders could see the source and context behind AI-supported answers. The trust work was no longer hidden in the data layer; it became part of the user experience. That is Trust Engineering: making confidence, accountability, and human judgment visible at the point of decision.

If your data teams need visibility, auditability, responsible AI UX, or AI governance patterns, connect on LinkedIn.

Decision Intelligence & Leadership Dashboards

What happened: Leaders were relying on delayed reports, manual reconciliations, and ad-hoc slide decks to understand financial and operational conditions. By the time the information arrived, the situation had often already changed.

What was expected of me: Create executive-facing dashboards that gave leaders a clearer, more current view of financial and operational reality.

What I did: I defined executive information needs, structured dashboard narratives, aligned data views with real decision patterns, and designed layouts that made risk, tradeoffs, and status legible at a glance. This was UX Architecture for decision support: organizing complex information so leaders could understand the situation quickly and act with confidence.

What changed: Leaders had faster access to current, evidence-backed views instead of waiting for stitched-together reports. The dashboard became part of the organization's decision-support structure: a shared place to see reality, weigh tradeoffs, and move decisions forward with greater confidence.

If your leadership needs real-time, trustworthy visibility into what's actually happening, connect on LinkedIn.

HR Service Design & Digital Transformation U.S. Navy

Human-centered workflow redesign across enterprise HR services — clarifying ownership, making escalation paths visible, and aligning policy with how work actually moved.

Across multiple HR domains — from benefits and separations to case management and self-service portals — I helped make complex HR services easier to understand, operate, and improve.

HR Service Blueprints

Enterprise HR Service Journeys & Ownership Maps
75+ interviews and service blueprints clarifying who owns which step in emotionally charged HR journeys (2021–2023).

What happened: HR cases were moving across desks, queues, systems, and policy boundaries in ways that were hard for sailors, staff, and leaders to see. The official process did not always match how work actually moved, which created confusion, escalations, and unclear ownership.

What was expected of me: Lead human-centered systems discovery across enterprise HR workflows so teams could see the real journey, not just the intended process.

What I did: I conducted interviews, mapped service journeys, facilitated working sessions, and produced service blueprints that made ownership, handoffs, decision points, and policy conflicts visible. This was Workflow Architecture grounded in human-centered discovery: listening to the people closest to the work and turning what they knew into shared structure.

What changed: Teams had a clearer picture of how HR cases actually moved, where they stalled, and who needed to act. The blueprints gave staff, product teams, and policy stakeholders a common reference point for reducing escalations, orienting new staff, and prioritizing improvements by human impact.

If your HR organization needs shared blueprints to clarify ownership, reduce escalations, and align policy with delivery, connect on LinkedIn.

Case Escalation Pathways

Escalation Rules, Workflow Automation Readiness & Digital Alignment
Clarified escalation triggers and designed handoff patterns so complex cases could move through consistent, policy-aligned pathways (2021–2023).

What happened: Complex HR cases were escalating through inconsistent paths. Staff needed to know which cases truly required escalation, what authority was needed, what “done” looked like, and how to explain decisions clearly to sailors waiting for answers.

What was expected of me: Translate policy, edge cases, and operational judgment into workflow patterns that product teams could reflect in tools, training, and queue logic.

What I did: I mapped escalation triggers, decision points, authority requirements, queue rules, and handoff patterns. This was Human-Centered Systems work: making policy, workflow, interface behavior, and human judgment point to the same path.

What changed: Staff had clearer pathways for moving complex cases forward, and sailors had a more consistent basis for understanding decisions. The work made escalation logic visible, repeatable, and easier to encode without flattening the nuance of real cases. That is Trust Engineering in practice: people can see how decisions move, who has authority, and what happens next.

If your escalations need clearer rules, handoffs, and decision paths, connect on LinkedIn.

Low-Bandwidth HR Portals

HR Self-Service UX for Constrained Environments
Prototype flows and content strategy tuned for low-bandwidth, high-security access to HR services (2022–2023).

What happened: Sailors needed access to critical HR services from ships, remote bases, shared workstations, and unreliable network conditions. The tasks were sensitive and important, but the environment made normal self-service patterns harder to use.

What was expected of me: Design UX architecture and portal patterns that made essential HR tasks clear, usable, and secure under real operational constraints.

What I did: I defined navigation, content hierarchy, simplified task flows, fallback states, and interaction patterns that reduced unnecessary round-trips. This was accessibility and inclusion applied to constrained environments: shaping the experience around limited time, intermittent connectivity, shared machines, and sensitive user needs.

What changed: The portal patterns gave teams a clearer structure for building HR services that could work under pressure. Users could find critical information, understand next steps, and complete high-impact tasks with more confidence, even when bandwidth, access, or device conditions were limited.

If your HR tools need to work in low-connectivity, high-security, or high-pressure environments, connect on LinkedIn.

Financial Systems Re-Architecture Lendwise

Financial product and workflow redesign across lending applications — unifying patterns, reducing duplication, and making complex financial tasks easier to complete.

Across mobile native, mobile web, desktop web, and internal consoles, I helped turn fragmented lending tools into a more coherent financial system for borrowers, staff, partners, and risk teams.

Loan Origination

End-to-End Loan Application UX
Unified multi-step loan application journeys across web and mobile, improving completion and data quality (2017–2020).

What happened: Borrowers were moving through dense, form-heavy loan applications with eligibility rules, disclosures, required fields, and decision points that were difficult to understand. The process needed to stay accurate and compliant, but it also needed to feel navigable to people making high-stakes financial decisions.

What was expected of me: Design the UX architecture for a guided loan application experience that worked across channels while respecting regulatory, data, and underwriting requirements.

What I did: I designed step structures, validation patterns, help-in-context, error recovery, and reusable components for cross-channel origination. This was Human-Centered Systems work: making financial rules, data collection, interface behavior, and borrower needs support each other in the actual flow of applying.

What changed: Borrowers had a clearer path through the application, with better visibility into obligations, requirements, next steps, and decisions. The experience supported better completion and cleaner data by making the process easier to understand and act on.

If your loan applications need to be accurate, compliant, and easier for people to complete, connect on LinkedIn.

Point of Sale & Title

In-Store & Dealer-Facing UX
Simplified POS and title loan flows for staff and partners handling applications live with customers (2017–2020).

What happened: Staff and dealer partners were walking customers through financial products in real time while also managing system steps, compliance checks, required disclosures, and product rules. The interaction needed to stay accurate without forcing staff to rely on memory, side notes, or repeated backtracking.

What was expected of me: Design workflow patterns that made in-person and dealer-facing workflows easier to complete correctly while preserving compliance and customer rapport.

What I did: I structured POS and title loan flows, tuned field order and branching logic, reduced unnecessary steps, and embedded prompts and guardrails for required disclosures. This was UX Architecture for live service environments: shaping the interface so staff could stay present with the customer while the system supported the required process.

What changed: Staff and partners had clearer paths through complex product flows, with fewer keystrokes, less backtracking, and better support for required steps. The tools helped the work move smoothly while keeping product terms, risk rules, and customer-facing guidance aligned.

If your in-person or dealer tools need to support compliance, speed, and customer trust at the same time, connect on LinkedIn.

Collections Console

Unified View for Collections & Support
Desktop console that brought account status, promises to pay, and prior contacts into one coherent view (2018–2020).

What happened: Collections and support agents were working across multiple systems to understand account status, payment history, promises to pay, prior contacts, and risk signals. Customers were often already under stress, and agents needed the full story before asking the first question.

What was expected of me: Design a decision-support view that brought the right account context, actions, and prioritization into one console so agents could work fairly, consistently, and quickly.

What I did: I designed the console layout, information hierarchy, prioritization rules, action flows, and scripting aids around the real conversations agents had every day. This was Human-Centered Systems work: aligning servicing data, collections policy, customer history, and agent judgment in one usable workflow.

What changed: Agents had a clearer view of the customer's situation, the account history, and the next best action. The console supported faster handling, stronger follow-through, and more consistent decisions by making the full context visible in the flow of work.

If your collections or support teams need clearer context, better handoffs, and fewer blind spots across systems, connect on LinkedIn.

Across these examples, the pattern is the same: when systems support the people using them, work becomes clearer, decisions move with confidence, and accountability is easier to see.

I design systems where data, workflow, interface, and human judgment reinforce each other — so teams can understand what is happening, know what to do next, and act with momentum.

Connect on LinkedIn

Definitions

These definitions explain the disciplines behind the work.

The use cases show what happened. These terms explain the structure behind those results: how systems are made clearer, safer, easier to trust, and easier for people to use.

Human-Centered Systems

Designing policy, data, workflow, tools, and interfaces around the people who have to use them.

Human-Centered Systems means the system is designed to support human judgment, responsibility, and action. It looks beyond the screen to the full environment people are working inside: policies, handoffs, tools, data, constraints, and decision paths.

In my work, this means listening to the people closest to the work, mapping how the system actually operates, and turning that understanding into structures people can trust and use.

The goal is a system where people can see what is happening, understand what matters, know what to do next, and act with confidence.

Trust Engineering

Making trust visible in the system — through sources, safeguards, boundaries, review points, and accountable decisions.

Trust Engineering means reliability is not assumed. It is designed into the way people see information, understand risk, review outputs, and know when human judgment is required.

In my work, this means making sources visible, showing where answers come from, defining escalation paths, clarifying system boundaries, and placing human review where it matters most.

The goal is a system where people can understand why something is shown, how much confidence to place in it, who is accountable, and what should happen next.

UX Architecture

Structuring how people move through information, decisions, tasks, and interfaces.

UX Architecture is the structural layer beneath the visible interface. It defines how information is organized, how tasks flow, how data meaning appears, how navigation works, and how reusable patterns keep the experience coherent.

In my work, this means designing the paths people use to understand what is happening, complete important tasks, review information, recover from errors, and know what to do next.

The goal is an experience that is easier to build, easier to explain, and easier for people to move through with confidence.

AI-Integrated Design

Designing AI into products and workflows in service of a clear purpose, problem, or human need.

AI-Integrated Design means AI is not the starting point. The starting point is the intent of the product, feature, workflow, or decision.

When AI is added without a clear purpose, the result is usually noise: confusing outputs, unclear ownership, weak trust, and features that feel impressive but do not support the work. When AI is integrated around a real problem, it can strengthen the system by helping people find context, interpret information, reduce manual effort, and make better decisions.

In my work, this means identifying where AI actually supports the job to be done, then designing how its outputs appear in the workflow: what it helps with, what evidence supports it, where its limits are, and when human judgment needs to guide the result.

The goal is AI that serves the product's purpose instead of distracting from it — making the system clearer, more useful, and easier to trust.

Workflow Architecture

Structuring how work moves across people, policy, tools, data, decisions, and handoffs.

Workflow Architecture means designing the path work follows through a system. It looks at how requests begin, where decisions happen, who owns each step, what information is needed, when handoffs occur, and where the system needs to support human judgment.

In my work, this means mapping how work actually moves, identifying where it slows down or becomes unclear, and turning that understanding into repeatable structures: clearer workflows, decision paths, queue logic, escalation rules, interface patterns, and operational handoffs.

The goal is a system where people can see what is happening, understand who owns what, know what to do next, and move work forward with clarity, confidence, and momentum.

Decision Support

Structuring information, context, and evidence so people can understand what is happening and decide with confidence.

Decision Support means the system helps people see the situation clearly enough to act. It does not replace judgment; it gives judgment the right context, evidence, timing, and structure.

In my work, this means designing dashboards, source-linked answers, status views, tradeoff visibility, and decision paths that bring operational reality forward. The goal is not simply to show more data, but to make the right information understandable at the moment a decision needs to be made.

A strong decision-support system helps people see what changed, what matters, where the information came from, what options are available, and what should happen next.

Accessibility & Inclusion

Designing systems that support people across real constraints, environments, abilities, and access conditions.

Accessibility and inclusion mean the system is designed for the conditions people actually face. That includes physical and cognitive accessibility, but also limited bandwidth, shared devices, time pressure, sensitive tasks, unclear environments, and high-stress moments.

In my work, this means shaping structure, navigation, contrast, content, task flow, fallback states, and interaction patterns so people can understand what is happening and complete important work under real-world constraints.

The goal is a system that works for more people in more situations — not because accessibility was checked at the end, but because human constraints were considered from the beginning.