GW
HomeWorkAboutNotesStackLab
Experience atOmnicom Group

Driving financial systems and automation across global teams.

Builder ofEnterprise AI Systems

AI agents, audit workflows, and decision intelligence at scale.

Based inNew York City

Building systems for enterprises around the world.

HomeNotesWorkAboutStackLabPartners
XLinkedInGitHubEmailllms.txt
© 2026 · New York City
All enterprise work

Decision intelligence case study

AI finance dashboard studio.

A source-flexible, HTML-based dashboard experience that refines data and uses AI to propose useful executive views, commentary, and user-requested edits.

HTML dashboardsData viewsAI orchestrationExecutive reporting
Ask the project copilotGrounded in this public-safe case study.

Business problem

What people needed to solve.

Finance audiences need different views of the same data, but rebuilding a dashboard for each request is slow and tends to separate the commentary from the view it describes. Teams need a way to go from approved data to a decision-focused executive view, and to take plain-language edit requests, without treating every request as a fresh build.

Decision frame

Questions the work needed to answer.

  • Which visualization is appropriate for this decision?
  • What should an executive see first?
  • Can a user ask a question or request an edit without rebuilding the dashboard?
Interactive operating flowSelect any node for context or ask the AI guide.

Selected Operating layer

AI visualization loop

Built an HTML-based dashboard workflow that consumes approved sources or views, runs a refinement loop to improve data readiness, then uses an AI-assisted loop to propose a chart treatment, a lead story, and executive commentary. An ask-and-edit layer lets users request analysis or adjust the presentation in plain language, with source selection and transformations kept traceable.

Implementation

From the business problem to a working operating model.

Built an HTML-based dashboard workflow that consumes approved sources or views, runs a refinement loop to improve data readiness, then uses an AI-assisted loop to propose a chart treatment, a lead story, and executive commentary. An ask-and-edit layer lets users request analysis or adjust the presentation in plain language, with source selection and transformations kept traceable.

  1. 01

    Accepted data from approved sources and existing views.

  2. 02

    Created a refinement loop to improve data readiness before presentation.

  3. 03

    Used an AI-assisted visualization loop to select an appropriate story, chart treatment, and executive commentary.

  4. 04

    Added an ask-and-edit layer so users can request analysis or adjust the presentation in plain language.

Controls, approvals, and delivery constraints

The work around the work.

Enterprise systems change only when data, access, testing, ownership, and evidence move together.

  • The visual recommendation is reviewable and should not override finance judgment.
  • Source selection, transformations, and user edits need traceable context.
  • The public case study uses illustrative examples only, never internal finance data.

Prior art

How leading teams approach this.

Public references for the same class of problem, so this work can be read against how other organizations are building it.

Anthropic

Advancing Claude for Financial Services

Anthropic's Claude for Excel reads, analyzes, and builds workbooks from a sidebar and explains every change with cell references, the same reviewable natural-language build-and-edit loop over finance data this studio applies to dashboards.

Microsoft

Introducing Microsoft Copilot for Finance: Transform finance with next-generation AI in Microsoft 365

Copilot for Finance takes prompts like explain forecast-to-actuals variance and returns a contextualized view with supporting data, the plain-language ask-and-edit pattern this dashboard workflow provides.

Databricks

AI/BI Genie is now Generally Available

Databricks pairs AI-assisted dashboard creation with conversational follow-ups on the same governed dataset, the model of generating a view and then refining it in natural language used here.

Public-safe outcome

Illustrative of a reusable path from approved data to an executive-ready dashboard, using sample data only, where the AI recommendation is reviewable and does not override finance judgment.