A AI Product Sprints
From idea to product shape
For founders and teams with a promising AI idea but no clear product shape yet.
Open to selective AI product work
AI Product Builder for founders using coding agents.
I turn vague AI ideas into scoped products, prototypes, internal tools and live MVPs — using Claude Code, Codex and modern full-stack tooling.
$keiran build --idea ""
clarify the actual problem
shape the product
define flows + architecture
build with coding agents
instrument + iterate
SchoolAILive
AI education platform scaled with zero paid acquisition.
12,000+active users
0
Users on SchoolAI
zero paid acquisition10+
Shipped builds and prototypes
LLMnesia · Job Radar · Flow402 · sites0
Years in tech and IT
IT · schools · SMEs · product2022
AI and product build focus since
Claude Code · Codex · full-stack AI01 — Engagements
What I help with
Three ways to work together, depending on how far along the idea is.
B MVP Builds
Ship the first real version
For scoped AI products, internal tools, SaaS features and workflow systems.
C Fractional AI Product Lead
Judgment without the full hire
For teams that need AI and product judgment, technical scoping and execution support without hiring full-time.
02 — Proof
See all workRecent work
● Chrome extensionLive
LLMnesia
Local-first Chrome extension for searching AI conversations across major LLM platforms.
Local-first · private by default · Chrome Web Store
● AI education platformLive
SchoolAI
AI education platform scaled to 12,000+ users with zero paid acquisition.
12,000+ users · 0 paid acquisition
● Consumer AI productLive
LunaCradle
AI-powered baby sleep guidance using structured knowledge and personalised LLM outputs.
Structured knowledge · personalised LLM output
● Agent payments R&DPrototype
Flow402
Agent payment prototype using x402, Base and USDC, built to explore machine-to-machine payment flows.
x402 · Base · USDC · prototype and discovery work
03 — Method
How I work
Coding agents compress the loop from idea to prototype. The judgment is in what to build.
Clarify the actual problem
Turn messy ideas into product shape
Define user flows and technical architecture
Build fast with coding agents and modern full-stack tooling
Instrument, test and iterate using real usage signals
04 — Approach
Product judgment first
I work remotely from Thailand across product, UX, technical architecture and build.
01 Decide
What should exist?
The useful work starts with scope: who the product serves, what risk needs testing, and what should stay manual for now.
02 Build
What can ship?
Coding agents compress the loop from idea to working software, but the output still needs product taste, review and restraint.
03 Prove
What did users do?
The next decision comes from usage signals: adoption, failure points, support friction and whether the workflow saves real time.
Let's build
Have an AI idea, internal workflow problem or prototype that needs turning into something real?
Book an AI Product Fit Call. Ten minutes, no pitch — just whether it's worth building.