AI Solution Platform - Heidi Lowry

AI-Powered Solution Development Platform

Transforming how businesses build custom AI solutions through strategic design leadership

AI platform entry point — users describe their business problem in natural language

The Challenge

Businesses were forcing SaaS products to fit their needs instead of getting custom AI solutions. Our platform had all the pieces – Processes, Workflows, Apps, Forms, Data, and Document Generation – but they weren't working together to give AI the business context it needed.

Role

Product Design Manager

Timeline

2025 to present

Team

1 lead designer (direct report), 6 engineers, 1 PM, multiple senior leadership and C-suite stakeholders

The Core Problem: Disconnected Capabilities

Before — Fragmented

Processes

Workflows

Apps & Forms

Data

Document Generation

Five capabilities, no shared context. AI has no business intelligence to draw from — it can't build anything meaningful.

After — Unified Platform

AI Solution Layer Context-aware across all capabilities

Processes Workflows Apps Forms Data Docs

All capabilities feed business context upward. AI can now generate genuinely custom solutions instead of generic outputs.

The strategic opportunity: unify six platform capabilities so AI has the business context to build solutions that actually fit

Strategic Approach

1. Identifying the Market Opportunity

Understood gap where businesses needed custom AI solutions
Leveraged our platform's unique advantage: combining all capabilities to provide AI with rich business context
Positioned us for competitive differentiation in AI-powered business solutions

AI Solution Engine Rich business context from all platform layers

Processes

Workflows

Apps

Forms

Data

Doc Generation

Our competitive advantage: no other platform combines all six. Together they give AI the business context to build solutions that actually fit.

2. Establishing Human-AI Partnership Principles

"The hardest part isn't the AI generation—it's designing the right amount of AI output that humans can meaningfully review and act on."

Challenged industry assumption that "AI should do everything"
Established principle: AI generates, humans guide
Designed granular interaction patterns for different personas without cognitive overload

Business User

Describes problem in plain language, reviews AI-generated solution as a whole

natural language

AI Generation Layer

Generates assets, workflows, and data structures from business context

structured output

Technical User

Validates and refines at the component level, customizes generated assets

Same AI output, different entry points. The design challenge was ensuring neither user had to work at the wrong level of granularity.

3. Leading Through Complexity

Stakeholder Management: Guided C-suite presentations and provided real-time meeting direction
Team Protection: Defended small, empowered teams to ensure streamlined leadership understanding and prevent scope creep
Technical Partnership: Collaborated with engineering to ensure designs matched AI capabilities possible to deliver
User Focus: Maintained continuous learning loop despite rapid AI technology changes

Design Process

Discovery & Alignment

Built continuous feedback loop with internal and external users
Balanced quick ideation for stakeholder alignment with engineering feasibility
Created shared vision across multiple senior stakeholders

Strategy to Execution

Elevated established senior designer to higher-visibility role, coaching them through executive presentations
Championed designer's ideas and created space for them to focus wholly on AI exploration
Mentored designer through executive stakeholder management, positioning them for promotion
Set team workloads and negotiated scope with PM to shape realistic roadmaps

Key Design Decisions

Adaptive AI interaction modes: Combined chat-style AI for open-ended exploration, inline AI for contextual assistance, and quick manual edits for finishing touches – matching users' mental models as projects evolve from ambiguous to refined
Progressive personalization strategy: Created different entry points for technical vs. non-technical users, with plans for role-based prompts and customizable AI behavior to meet each persona where they are
Human-in-the-loop validation: Designed preview and validation steps that mimic peer collaboration, allowing early error detection and correction to prevent hallucination cascades and maintain user confidence

Adaptive AI Interaction Model

Exploration

Open-ended problem framing — users describe in natural language

Chat-style AI

Refinement

AI assists in context — suggestions appear where the user is already working

Inline AI

Finalization

Precise manual control — user makes targeted edits before committing

Manual Edit

Each mode matches the user's mental state as projects evolve from ambiguous to refined — AI steps back as clarity increases

Impact & Outcomes

Designer Promoted

New Interaction Modes

✓

Patterns Adopted

6→1

Capabilities Unified

Business Impact

Will shape platform product cohesion across all capabilities
Informs AI asset generation strategy company-wide
Fundamentally evolves how we communicate as an integrated platform

Team Development

Elevated senior designer to executive-facing role, resulting in promotion trajectory
Established design patterns now used across multiple teams
Created framework for future AI feature development
Built team capability in AI design through focused mentorship

Reflection

What I learned about leading AI design: The intersection of AI capabilities and human needs requires constant recalibration. Success came from protecting the team's focus while managing upward to keep stakeholders aligned.

What I'd do differently: Keep the team smaller and more focused from the start. Too many stakeholders got involved too early, making it hard to narrow scope and costing us valuable learning time through mid-development changes. Small, empowered teams with clear decision rights actually move faster.