Score the alternatives
Compared speed to deploy, data sensitivity, fit with the existing agent-platform strategy, vendor maturity risk, and ongoing cost.
All work01 02
CASE STUDY / AI STRATEGY / RAPID PROTOTYPINGPrototype / decision artifact
Build-vs-Buy Evaluation — Compensation Analytics
A structured platform evaluation and clickable fictional-data prototype to support a purchasing decision.
The vendor product bundled two different things: analysis workflows that could be built in-house, and external market benchmarking data that could not be recreated in-house at any price and would still have to be licensed. That reframed the question: build the workflows we can, and license only the data we cannot.
MY ROLE
Sole author of the evaluation and interactive prototype
PROJECT STATUS
Prototype / decision artifact
FOCUS
AI Strategy / Rapid Prototyping · Enterprise AI
01 / BUSINESS CHALLENGE
Start with the problem.
A business team wanted a third-party AI compensation-analytics platform. Leadership needed a timely comparison with an in-house alternative before a purchasing decision.
02 / SOLUTION APPROACH
A practical path forward.
Ran a structured build-vs-buy evaluation and, in parallel, used Claude Code to create a working interactive prototype of the in-house alternative. The prototype was a conversational compensation-analysis agent with saved analyses, generated charts, and a live audit-trail panel, running entirely on fictional data.
My contribution
Sole author of the evaluation and interactive prototype.
04 / IMPLEMENTATION
From design to workflow.
Make the argument tangible
Built a clickable prototype with Claude Code so stakeholders could inspect the proposed analysis workflows, saved analyses, generated charts, and audit-trail panel using fictional data.
Design decisions & tradeoffs
Separate workflows from licensed data+
Workflow capability could be built in-house. External market benchmarking data would need to be licensed regardless of the chosen platform path. The decision artifact separated those two needs.
07.5 / TRUST & GUARDRAILS
How the system fails safely.
Fictional data throughout
The interactive prototype ran entirely on fictional data. Its audit-trail panel demonstrated the concept; this case study does not claim production integration or a production control system.
Decision support, not a purchasing outcome
The evaluation and prototype were delivered for stakeholder review. No purchasing decision or production deployment is claimed.
08 / OUTCOMES
What the work demonstrates.
- Delivered the structured evaluation and clickable prototype as decision artifacts for stakeholder review.
- Reframed the platform question around buildable workflows and separately licensed market benchmarking data.
Prototype and decision-support deliverables only. No purchasing outcome, production deployment, or measured business gains are claimed. The interactive prototype used fictional data and was delivered for stakeholder review.
10 / REFLECTIONS
What I take forward.
- Evaluate the source of a product’s value before treating every feature as something to rebuild.
- An interactive prototype helps stakeholders inspect an argument quickly.
Finance & Supply Chain Copilot
05 /A CONVERSATION AWAY
Let’s build
what’s next.
Interested in enterprise AI, agentic systems,
or intelligent automation? Let’s connect.