Overview
Tanglewood is a strategic concept exploring how enterprise configuration and onboarding could work if designed AI-first, rather than retrofitting AI into legacy workflows.
This case does not document a shipped product. Instead, it demonstrates how I approach complex systems, reframe entrenched problems, and design toward scalable, human-centered outcomes, especially in environments where AI plays an active role.
The Problem Space
Enterprise platforms often fail before users ever reach “day one.”
Across finance, HR, and operations platforms, onboarding typically involves:
Long setup timelines
Fragmented ownership across teams
Heavy reliance on spreadsheets and tribal knowledge
UX that assumes structure before understanding exists
Traditional design approaches rush toward:
Forms
Step-by-step wizards
Dashboards and validation rules
But these solutions presuppose that users already know what they are configuring. In reality, much of the early onboarding phase is about aligning people, policies, and intent before the system can be meaningfully shaped.
In many enterprise environments, this gap is filled by external consultants who translate organizational intent into system configuration. While effective, this approach is expensive, time-bound, and difficult to scale. Much of the knowledge produced during implementation disappears once the engagement ends, forcing organizations to repeat the process during future changes.
AI is frequently introduced too late, framed as...
Autofill
Recommendations
Help text
...rather than as a core collaborator in the process.
Strategic Reframe
Tanglewood begins with a different premise: Enterprise onboarding is not a setup task. It is a translation problem.
Organizations are attempting to translate:
Policies into structures
Language into data
Human intent into system behavior
Instead of starting with screens, Tanglewood starts with conversation and narrative.
The system’s first job is not to validate inputs—it is to help users:
Articulate what they mean
See assumptions made explicit
Understand consequences before committing
In this model:
AI acts as an orchestrator, not a shortcut
Users collaborate with the system instead of feeding it
Progress happens through dialogue, not clicks
Tanglewood reframes this dynamic by embedding those translation and sense-making functions directly into the system itself. Rather than relying on temporary external expertise, organizations collaborate with an AI agent that continuously performs the same interpretive work: documenting intent, surfacing assumptions, and maintaining alignment over time.
The result is not faster setup alone, but clearer, more resilient configuration.
High-Level Flow
At its highest level, Tanglewood operates as a continuous loop, not a one-time setup sequence.
The flow is designed to:
Establish shared understanding
Make assumptions visible
Preserve decision context over time
Rather than guiding users through a fixed wizard, Tanglewood cycles through three repeating states:
Express – users articulate intent in human terms
Interpret – the system translates intent into structured meaning
Align – humans and AI converge on a shared configuration
This loop may run multiple times before anything is finalized.
Nothing is committed by default.
The Tanglewood Concept Model
1. Input
Users provide whatever materials already exist:
Documents
Spreadsheets
Written explanations
Partial data
Tanglewood assumes inputs are incomplete and contradictory.
That is treated as normal, not erroneous.
2. Synthesis
The AI agent:
Analyzes provided materials
Drafts an initial enterprise blueprint
Surfaces assumptions explicitly
Flags uncertainty and missing context
The system explains its reasoning in plain language.
3. Review
Users examine the draft blueprint and can:
Accept or reject assumptions
Edit outputs directly
Ask clarifying or hypothetical questions
This phase is optimized for sense-making, not speed.
4. Refinement
As feedback is incorporated:
The configuration evolves
Rationale is updated
A change history is maintained
Understanding deepens with each iteration.
5. Validation
Only after alignment is reached does the system move toward:
Formal validation
Deployment readiness
Ongoing maintenance
At this stage, decisions are explicit and ownership is clear.
Artifact 1 of 4 – Business Narrative Document
The Business Narrative Document captures how the organization believes it operates—before that understanding is forced into system structures.
Rather than starting with predefined fields or schemas, this document accepts inputs in human terms: policies, goals, constraints, exceptions, and unresolved questions. It is intentionally permissive, allowing ambiguity and contradiction to exist long enough to be examined.
In Tanglewood, this narrative becomes the primary substrate for AI interpretation. The system does not treat it as static documentation, but as a living source of intent that can be revisited, challenged, and refined over time.
Why this matters:
Prevents premature formalization
Gives non-technical stakeholders a first-class voice
Replaces consultant-led discovery with system-owned understanding
Establishes a shared reference point before configuration begins
The Business Narrative Document is not an output to finalize—it is a foundation to reason from.
Artifact 2 of 4 – Configuration Summary
The Configuration Summary represents the system’s current understanding of how the organization intends to operate.
Unlike traditional setup screens or configuration wizards, this artifact is not manually assembled by the user. It is generated and continuously updated by the AI agent as it interprets narrative inputs, documents, and ongoing feedback.
The summary translates human intent into structured concepts—entities, rules, relationships, and constraints—while explicitly marking assumptions, uncertainties, and inferred decisions.
Why this matters:
Bridges narrative understanding and system structure
Makes AI interpretation visible and inspectable
Gives users a concrete artifact to react to, not abstract prompts
Replaces consultant-produced configuration outputs with system-owned synthesis
The Configuration Summary is not a final state.
It is a working draft that evolves as understanding deepens.
Artifact 3 of 4 – Decision & Change Log
The Decision & Change Log captures how understanding evolves over time.
As assumptions are accepted, revised, or rejected, the system records not just what changed, but why the change occurred. Each entry preserves context, rationale, and ownership, creating a durable record of decision-making.
Unlike traditional audit logs or version histories, this artifact is designed for human comprehension. It explains shifts in reasoning, not just state transitions.
Why this matters:
Preserves institutional knowledge beyond implementation
Makes configuration decisions traceable and defensible
Reduces rework during organizational change
Replaces consultant-owned rationale with system-owned memory
The log is continuously available and grows alongside the system.
Artifact 4 of 4 – AI Narration / Reasoning Trail
What the System is considering
Why This Matters
Tanglewood is not a faster setup flow.
It is a shift in how enterprise systems come into being.
Most enterprise onboarding fails not because the software is incapable, but because the work of interpretation—translating human intent into system structure—is fragmented, expensive, and temporary.
Traditionally, this gap is filled by external consultants. They gather context, reconcile contradictions, document decisions, and translate policy into configuration. While effective, this approach is costly, difficult to scale, and prone to knowledge loss once the engagement ends.
Tanglewood internalizes that work.
By embedding sensemaking, interpretation, and documentation directly into the system, Tanglewood enables organizations to:
Reduce dependence on long-running consulting engagements
Preserve institutional knowledge over time
Make configuration decisions explicit, inspectable, and revisitable
Adapt continuously as the organization evolves
From a UX perspective, this reframes onboarding as a collaborative process, not a hurdle to clear.
From a business perspective, it replaces episodic, high-cost intervention with durable, system-owned understanding.
Most importantly, it establishes a model where AI does not obscure decision-making—but strengthens it.
The result is not just a configured system, but a shared, living understanding of how the organization intends to operate.
Design Principles
Tanglewood is guided by a small set of deliberate design principles:
Start with meaning, not mechanics
Documents before dashboards
Visibility over automation
Conversation before configuration
Humans stay accountable; AI does the heavy lifting
These principles shape not just UI decisions, but the order in which the system reveals itself.
Rather than asking users to “fill out everything correctly,” Tanglewood helps them understand what the system believes to be true, and invites them to correct it.
