Stacks & Vera

stacks UI

Building what I couldn't find

Building what I couldn't find

Year

'25

Client

Side Project

Service

Design & Engineering

When research drives a design decision, the reasoning should be traceable.

Which evidence, which synthesis, which assumption. In practice, that lineage disappears within weeks. Stacks and VERA are tools I designed and built to solve that problem: one tracks how evidence connects to artifacts, the other generates structured deliverables from evidence logs with built-in confidence scoring.

Stacks The Lineage Layer

Stacks is a spatial canvas built on tldraw that treats research and strategy work as a system with explicit lineage. Sources, synthesis, assumptions, and design artifacts are placed on the canvas and connected directionally so you can trace from a raw interview quote through synthesis to the specific screen or decision it informed. When an upstream source changes, Stacks surfaces which downstream artifacts may need review. The interaction model is intentionally calm: things arrive slowly, hover states reveal connections without overwhelming the view, and every element is inspectable.

VERA The Generation Layer

VERA (Validated Evidence, Ready Artifacts) is a CLI-based AI agent that generates structured discovery deliverables from evidence logs. It reads the evidence you've collected, generates artifacts like journey maps, problem statements, or synthesis documents, and tags each output with a confidence score based on how much supporting evidence exists. The human reviews, validates, or rejects the AI does the assembly work, the human retains judgment. VERA runs in production and integrates with Stacks through an API layer, so generated artifacts inherit the lineage of their source evidence.

Why This Matters Beyond Design

The problems Stacks and VERA solve upstream change propagation, confidence in derived outputs, human oversight of automated generation are the same challenges data platforms face with pipeline lineage, schema evolution, and AI-assisted analytics. I built these tools because the problem is personal to how I work. That they mirror the challenges of data infrastructure is why this project keeps sharpening my thinking on the exact systems I want to design professionally.



© Stacks & Vera

login
fieldbooks list
login
login
login
login
login
login
login
login

The Problem

Evidence accumulates everywhere. Transcripts, spikes, prototypes, docs, Jira. But lives nowhere structured. Confidence is implicit. Artifacts get produced by humans synthesizing the same inputs in parallel, which means deliverables look complete but can't trace back to what was actually validated. Polish gets mistaken for confidence.

The Solution

Stacks is the evidence layer. It captures and structures signal as a single source of truth across an engagement. VERA is the agent that takes that evidence in and produces artifacts out. It reads from Stacks and generates deliverables (strategy, roadmap, backlog, cost model) that reflect actual confidence.

Together Every artifact traces to evidence. Every recommendation carries a confidence level. Humans spend their time generating signal; VERA generates the artifacts derived from it.

Role

Strategy, Design, Engineering

API Integration Hub