SparkData AnalyticsApplied AI Portfolio
OpenAI Research Study · SDA Portfolio Case Series

Navigating ambiguity.

Five recorded cases show how SparkData Analytics turns complex problems into clear decisions and working systems.

SparkData Analytics founder Ryan Zimmerman developed and demonstrated the work as a participant in the OpenAI study.

5 recorded cases4 matched ChatGPT and Claude comparisons1 end-to-end AI workflow
The evidence

Five cases. One operating standard.

The OpenAI study, High Stakes Multi-Homing (Pro Users), examined how complex, real-world work moves across ChatGPT and Claude: where each product is strong, where it struggles, and how the workflow changes between the two.

S01 through S04 replicate the same task and starting prompt independently in ChatGPT and Claude, across retirement planning, federal tax, local demand generation, and brand evaluation. The matched conditions make every comparison inspectable and reveal where the products converge, diverge, or require further verification.

S05 brings the complete SDA method together in the MGM ARIA Rewards Optimizer, carrying the work from structured research and modeling through reconciliation and validation to a working decision tool.

The stakes in each case are real: a retirement runway, a tax year, a local business's growth, a brand's identity, and a family trip. People have to live with these decisions; the method has to hold up beyond the first answer.

The standard

Evidence before adjectives.

The operating rule is direct: no model is the sole validator of its own output. Disagreement between models is treated as signal, not dismissed as noise.

SparkData Analytics builds governance and verification infrastructure around that rule. The recordings are the argument: the standard made visible before anyone is asked to trust it.

Sonic case-study media

A creative system, heard through its decisions.

Watch carries one bounded case-study artifact. The complete lab identity and method live at sonic.sparkdatalab.ai.

Abstract Sonic Architecture Lab signal paths converging into one warm waveform Case-study media · Listening-led Sonic Architecture LabMany Versions. One That Feels True.A visual case study of how story, contrast, careful listening, and human feedback shape one coherent creative direction.Follow the signal
Flagship applied-AI builds

Working systems, shown in practice.

Explore four SDA builds: a persistent project workspace, a live tax workflow, predictive investment intelligence, and human-gated customer support that carries context through a real call.

Cinematic connected-workflow artwork for Tax Workflow Agent Flagship · 9-month build AI CPA agent · Accounting automationTax Workflow AgentMore than 5,000 controlled artifacts. Twenty-one current workpapers. One agent that carried the work into live tax software.Watch the proof Cinematic full-stack predictive intelligence system for Investment Intelligence Predictive ML · Full-stack build Investment intelligence · Predictive analyticsInvestment IntelligencePoint-in-time data, ten-factor market intelligence, transparent ML, thirteen promotion gates, three identical reproductions — the complete predictive chain, running.Explore the full stack Cinematic project-workspace artwork for From a Real Question to Finished Work SDA consulting field demonstration Applied intelligence consulting · Navigating ambiguityFrom a Real Question to Finished WorkA professional working in youth and fitness programs needed context, decisions, and next actions to carry across sessions. SDA found the real assignment, built the workspace, and recorded the proof.See how Ryan framed and directed the work Cinematic human-gated support workflow for SDA / SRS CSR AI SDA / SRS capability · Real call Customer support intelligence · Human-gatedSDA / SRS CSR AIThe AI prepares, waits for the human cue, presents the issue, and converts a closed support line into a concrete next action.Watch the handoff happen

All five walkthroughs are private. Approved viewers verify their email before viewing.