Legacy Logix

Estate management is one of those problems everyone faces but nobody is taught how to handle. When a parent dies, families scramble through boxes of documents, navigate state-specific legal requirements, and pay attorneys thousands of dollars for guidance that could be systematized. Legacy Logix set out to change that — and I joined three co-founders I’d worked with across multiple previous ventures to build it from the ground up.

The Problem We Saw

The estate planning industry operates on a model designed for attorneys, not families. Organizing an estate, keeping it current as life changes, and managing the distribution process after death all require professional guidance that most people can’t afford or don’t seek until it’s too late. We believed technology could make this accessible to everyone — starting with document organization and lifecycle management, then expanding into AI-powered estate analysis that could tell someone exactly what they need to do based on their specific situation and state laws.

Our initial go-to-market hypothesis, driven by strong VC partner advocacy, was a B2B2C play targeting estate attorneys and financial brokers. The logic was sound on paper: professionals already had the client relationships, so partner with them to reach families faster. What we discovered was that this kind of SaaS sales process was completely foreign to these professionals. Their practices weren’t built for technology partnerships, and the sales cycle dragged. We pivoted to direct-to-consumer — which had always been in the long-term plan, but the B2B2C detour taught us that sometimes the faster path to revenue isn’t the obvious one.

Building the AI Pipeline

The estate analysis feature was our most ambitious bet. The original approach used licensed professionals to generate and review every estate analysis before it reached a user. This delivered high-quality output but created a per-customer cost of $500 to $700 that made the product economically unsustainable at scale.

I led the redesign of this pipeline through a combination of product strategy and AI architecture decisions. We built a multi-agent system with extremely strong guardrails: separate AI agents — sometimes different models entirely — would cross-analyze each other’s output against a defined quality criteria set. The system generated three separate analysis attempts and selected the best output. Any result that fell outside quality thresholds or presented an uncommon estate situation triggered escalation to a human reviewer.

The insight that made this work was recognizing that the vast majority of estates are relatively consistent and even straightforward once you understand the patterns. Complex, exceptional situations absolutely require human expertise. But for the 90-plus percent of cases that follow predictable structures, well-designed AI guardrails and cross-validation could deliver the same quality as manual review. This approach cut operating costs by over 95%, transforming what had been an unsustainable product into a viable revenue stream while maintaining the enterprise-grade privacy and security standards that estate data demands.

Design Operations as a Force Multiplier

When I arrived, the product was moving slowly — not because the team lacked talent, but because the design-to-engineering handoff created constant churn. Inconsistencies lived primarily at the workflow level rather than the component level. The same user task might flow through three different interaction patterns depending on which part of the platform you were in, and the documentation supporting each feature varied wildly in depth and format.

I ran a systematic audit across the platform and found that roughly 60% of our design surface carried some form of debt — redundant flows, undocumented interaction patterns, and assumptions baked into the UI that no longer matched the product’s direction. Eliminating that debt wasn’t just cleanup work. It was the foundation for everything else.

I leveraged new capabilities in Figma to accelerate how we delivered mocks and specification details, then established structured patterns with engineering that reduced the back-and-forth between design intent and implementation. Documentation standards, design system tokens, and shared component libraries gave engineers the context they needed to build correctly the first time. The rework cycle — that expensive loop where engineering builds, design reviews, engineering rebuilds — shrank dramatically.

The cumulative effect of these changes doubled our release velocity. Concepts that previously took months to ship were reaching production in six weeks. This wasn’t one silver-bullet fix; it was process improvements, tooling upgrades, and collaboration patterns all reinforcing each other.

What I Took Away

The Legacy Logix experience reinforced something I’ve seen across every venture I’ve been part of: the design operations layer is where velocity lives. Individual design talent matters, but the system around that talent — how decisions get documented, how handoffs work, how quality gets validated — determines whether a team ships fast or spins. Building that system from scratch with co-founders who shared a decade of working history made it possible to move quickly and honestly when things weren’t working, whether that was a go-to-market strategy or an AI pipeline architecture.