Building with AI

How I approach problem-solving with AI

AI-assisted development and AI-powered product features — from CleerSplit and Specly to the live RuthVerse Product Guide on this site.

  • Start with a real user problem — not the tool.
  • Use AI to move faster; use judgment to ship responsibly.
  • Scope LLM features with guardrails, retrieval, and clear fallbacks.

iOS app + Python CLI

CleerSplit

GitHub CopilotCodexDockerGitHub ActionsFastlaneMCP

Problem

Shared expenses create awkward tension — and validating split logic across iOS, CLI, and deployment workflows needed to be fast, reliable, and portable across environments.

Approach

Built a Python CLI with Docker for consistent multi-environment runs, GitHub Actions for CI, Fastlane for TestFlight delivery, and AWS/Azure setup scripts for future scale. Used Copilot and Codex to accelerate split-logic tests and edge-case handlers, with MCP tooling for agent-friendly integration — reviewing every output before shipping.

Outcome

CleerSplit is live in TestFlight with automated delivery pipelines, containerized CLI validation, and infrastructure ready to scale when user growth demands it.

Social product prototype

Specly

CursorSwiftUIFigma

Problem

Specly needed to explore how people turn shared interests into real plans — requiring rapid UI iteration and product thinking, not months of setup.

Approach

Built with Cursor as an AI pair-programmer: prototyping flows, refining SwiftUI screens, and iterating on copy and interaction patterns while keeping product intent in the lead.

Outcome

Specly is heading to TestFlight with a polished product world on ruthokolo.com and a clear vision for discovery, invites, and meetups that actually happen.

Live AI assistant

RuthVerse Product Guide

CursorVercel AI SDKGroqRAG

Problem

Visitors needed a fast way to understand which RuthVerse app fits their need and how to join beta — without reading every page.

Approach

Designed an agentic guide with retrieval over product knowledge, tool calling for recommendations and beta links, scoped prompts, and friendly error handling — built and shipped on the live portfolio.

Outcome

A working AI product feature on ruthokolo.com that demonstrates RAG, tool orchestration, and product UX in production.