ProjectCase Study

Aperture Financial Intelligence

A portfolio-aware fintech workspace for holdings review, cited AI briefs, goal scenarios, risk explanation, and billing-aware research flows without trade execution or personalized advice.

Date

2026

Complexity, made quiet.

Aperture is designed around a small research loop: load a portfolio, inspect the evidence, review scenarios, and keep the final judgment with the person using the workspace.

The current product includes public marketing routes, Clerk-authenticated workspace routes, a Go-backed finance API layer, Supabase persistence, AI safety fixtures, and verified billing and security paths.

Load the portfolio

Start from holdings, allocation, and freshness metadata.

Inspect the evidence

Review source chunks, assumptions, and risk context.

Keep the boundary visible

Use scenario review and research prompts without trade instructions.

02Product page

The live product language stays explicit.

The public product page presents the core workspace surfaces and keeps the guardrails visible: freshness labels, saved citations, source context, and no return promises.

Aperture Financial Intelligence product page

03Workspace surfaces

What is implemented in the workspace.

These are the routes and interactions currently present in the product. Research, goal, and risk views use deterministic fixture mode by default so safety behavior and citations can be reviewed without external model spend.

Portfolio intelligence

A portfolio workspace for holdings, allocation, concentration, intraday movement, provider freshness, and visible unavailable states.

Market movement review

A watchlist surface for provider-backed equities and ETFs, daily movement, research notes, and an add-ticker flow.

Research room

A deterministic local-demo filing summary with cited source chunks, filing highlights, material risks, and saved research controls.

Goal intelligence

A deterministic local-demo scenario planner with milestones, contribution scenarios, and research-discipline prompts.

Risk explainer

A deterministic local-demo risk review for concentration signals, allocation context, limitations, and review actions.

Settings and billing

Workspace controls for privacy, providers, plan entitlements, saved research limits, and billing-period usage.

Visible guardrails

The boundaries appear in the interface.

The product does not hide its constraints in fine print. Research-only framing, freshness labels, and unavailable states are part of the visible experience.

Research and portfolio decision support only.

No brokerage or trade execution.

No personalized investment advice.

No promised returns or suitability claims.

Visible provider timestamps and freshness labels.

No estimated values when market data is unavailable.

04Current state

Built enough to audit.

The public marketing routes, authenticated workspace routes, Go API tests, deterministic AI evals, and billing and security verification paths are in place. Portfolio and watchlist views expose provider freshness and failure states.

The case study stays clear about what is real today: deterministic AI behavior is the default review mode, direct model calls require explicit environment opt-in, and the product does not present research support as financial advice.

MoreSelected Work