AI Outcomes. Not AI Output.

David Kim

Most AI work delivers output. Mine delivers business outcomes. I install the operating system that converts business intent into shipped value: Continuous Intent Delivery. 35 production applications in 15 weeks. Every line of code traces to a documented value statement.

David Kim
CID
Outcomes

About

The gap the industry isn't closing

Every AI advisor on the market sells the same things: better prompting, better tooling, better evaluation. None of that moves the business outcome.

AI ROI doesn't live in the prompt. It lives one layer deeper, in how a business value statement becomes a shipped product without losing fidelity along the way. That layer didn't exist. I built it.

Continuous Intent Delivery (CID) is the development methodology I co-created at Alchemaize and proved by shipping 35 production applications in 15 weeks with two co-founders. The atomic unit is a Verifiable Outcome Slice: a documented intent that ties a business value statement to its executable acceptance contract before any code is generated. AI writes the implementation. The contract verifies it. The sprint disappears. The cycle is measured in minutes.

Before Alchemaize, five years at AWS as a Senior Manager. I built the first-to-market Generative AI EBA framework that cut public-sector AI implementation timelines by 60% across 100+ customer engagements. That work, $250M+ in cloud services delivered, taught me the gap between AI capability and customer outcome. CID is what I built to close it. Eight years in the Marine Corps before any of it.

35

Production apps in 15 weeks

~1M

Lines of production code, 2 FTE

30-50x

Throughput multiplier, verified

$250M+

Cloud services delivered at AWS

The Claim

"Most AI work delivers output. Mine delivers business outcomes."

The whole AI advisory market is teaching people to use AI better. Better prompts, better tools, better evaluations. That's useful. It's also not where ROI happens.

Business value gets lost in the translation from "what we want" to "what got built." CID closes that translation gap with a documented intent pipeline that AI orchestrates and verification enforces. The value statement and the production artifact stay linked.

That's the work I do, and it's the work nobody else is doing at this layer.

The Method

How AI becomes business value

Continuous Intent Delivery in three movements. Documented intent in. Verified outcome out. Every step traceable.

1

Business Intent

The Verifiable Outcome Slice

Every initiative starts as a VOS. The Intent Engineer authors the value statement, the executable acceptance contract, and a curated context bundle of 5 to 15 files. "Done" is defined before generation begins.

  • • Business value statement (WHY)
  • • Gherkin acceptance contract (HOW)
  • • Curated context bundle (CONTEXT)
2

AI Orchestration

Generation against the contract

The AI Orchestrator runs the agents against the contract. Multi-model, multi-stack. Compliance rules block non-compliant code at the moment it's written. The work is generated, not typed.

  • • Claude, Bedrock, Gemini, multi-LLM
  • • HIPAA / FedRAMP / FinServ enforced
  • • Drift caught in the loop, not after
3

Verification

Outcome, traceable

The Verification Owner runs the contract. It's executable. Nothing ships without passing. Every shipped feature traces back to the original business value statement that authorized it.

  • • TDD red-green-refactor, every task
  • • Acceptance scenarios pass before merge
  • • Value statement → production, linked

"The middle, the actual typing of code, is now a near-zero-cost operation. Frameworks that optimize for the middle are optimizing for nothing."

From the CID Framework

Expertise

What I do that nobody else can

AI Output → AI Outcomes

The whole industry sells better AI tooling. I install the methodology that converts business intent into shipped business value. CID is the missing layer between capability and ROI.

  • • Documented VOS pipeline, intent to production
  • • Every line of code traces to a value statement
  • • 30-40% ceremony overhead eliminated

The Sprint Is Dead

Sprints exist to coordinate humans typing code. When AI writes the code, sprint cadence is friction. CID runs in continuous cycles measured in minutes, not weeks.

  • • One 30-minute meeting per week, total
  • • Continuous flow replaces sprint boundaries
  • • Output ships as fast as verification permits

Compliance-Enforced Generation

Auditing AI-generated code after the fact is mathematically impossible at AI speed. CID enforces HIPAA, FedRAMP, and Financial Services rules at the moment code is written. Non-compliant output cannot ship.

  • • 22 blocking rules across three frameworks
  • • 45 CFR Part 164, NIST 800-53 Rev 5, PCI-DSS v4.0
  • • The audit-lag problem disappears

AI Orchestration at Pod Scale

Three roles, not thirty. Intent Engineer writes the what. AI Orchestrator runs the agents. Verification Owner makes speed safe. Pods scale linearly. The pod is the unit, not the team.

  • • 2 FTE produced 31 developer-years of output
  • • Cross-stack, cross-model, cross-vertical
  • • Add pods, not headcount

Production Proof, Not Theory

Methodology that doesn't ship isn't methodology. Three of us shipped 35 production applications in 15 weeks. ~1M lines of source code. 429 test files. That's the receipt.

  • • Web, mobile, native iOS, games, infrastructure
  • • 30-50x developer-throughput multiplier
  • • Forged from work, not whiteboards

Enterprise AI Leadership at AWS

Five years building the public-sector AI programs that customers actually adopted. First-to-market Gen AI EBA framework cut implementation timelines by 60% across 100+ engagements.

  • • $250M+ in cloud services delivered
  • • 100+ annual customer engagements
  • • Federal, state, local public-sector experience

Recent Work

Where the methodology lives in production

Four production AI systems pulled from the 35-app body of work. Each one shipped against a business outcome, each one built with CID from the first commit.

CATALYST

The Adoption Path

How enterprises actually adopt CID. The customer gets the methodology, three compliance extensions, a customer-deployable evaluator running in their own AWS account, and the consulting team to install the practice. AWS Marketplace listing in flight.

CID / ELCID install HIPAA / FedRAMP / FinServ Evaluator harness

DOSSIER

Public Sector IDP

AWS-native intelligent document processing that replaces FileNet. Lighthouse customer: Texas HHSC intake forms. White-label-first architecture so every state agency overlay lives in its own tenant directory. Built in weeks where legacy IDP implementations take years.

Bedrock FileNet replacement Multi-tenant overlay

DMS Killer

Automotive F&I

Cloud-native competitor to legacy dealer management systems in the F&I (Finance and Insurance) layer of automotive retail. Six VOSes in production: F&I bootstrap, menu and products, compliance copilot that catches deal-jacket errors before they ship, Westlake lender adapter, console UI.

Compliance copilot Cognito auth Live at dealer pilots

CADENCE

Consumer Health AI

AWS Bedrock-powered cycle-synced nutrition platform. The macro engine has mandatory evidence citation on every recommendation, with hard guardrails against medical misinformation. iOS-first, native SwiftUI over a Bedrock-backed API. The recommendation engine is the moat.

Bedrock Claude SwiftUI native Evidence-grounded RAG

Full portfolio includes ThreadLens, Radient, Naeum (native iOS), Visible Wealth, TradeCodex, Yeon CRM, STRfish, Tanaiger, NoshMode, Drawer, FlipMode, BoxLens, Ember, SkipDay, RuneShell, VoidTrader, and others. 35 total. See the resume for the full inventory.

Experience

Leadership Track Record

2025 - Present

Co-founder & Chief Product Officer

Alchemaize, Inc.

Co-founded with Casey Robinson and Glenn Knepp in August 2025. AI Orchestrator role on the three-person pod that proved Continuous Intent Delivery by shipping 35 production applications in 15 weeks. Created the CID / ELCID / CATALYST methodology stack. Built the compliance enforcement extensions (HIPAA, FedRAMP, Financial Services) that block non-compliant code at generation time. Co-authored the book.

AI-Native Methodology VOS Pipeline Design Compliance Tooling Founder
2021 - 2026

Senior Manager, Public Sector Shared Services

Amazon Web Services (AWS)

Built and led three strategic organizations that accelerated cloud transformation across AWS's global public sector customer base. First-to-market Generative AI EBA framework that cut implementation timelines by 60%. $250M+ in cloud services delivered across 100+ annual engagements.

Gen AI EBA Framework Public Sector RAG / Bedrock $250M+ Delivered
2019 - 2021

Vice President, Cloud Advisory

Genpact

Founded and scaled a 220-person multi-cloud professional services practice. Generated $70M annual revenue.

Multi-Cloud Practice Development $70M Revenue
2017 - 2019

Senior Director, Global Software & Cloud Portfolio | Deputy CTO

Honeywell International

Directed $40M technology portfolio spanning mobile applications to IoT. Managed a 180-person global engineering organization.

IoT Azure Migration Strategic Planning
2015 - 2017

Founder and Chief Technology Officer

kFidelis, Inc

Launched a technology consulting firm specializing in public sector digital transformation. Generated $1M+ first-year revenue. Led SAFe Agile implementation for a 1,000+ person IT organization.

Startups Agile Transformation Public Sector
1992 - 2015

Earlier Experience

Managing Director

The Greentree Group (2004-2015)

Generated $10M+ in new revenue with 40% YoY growth. Managed $120M Agile development portfolio.

Sr Systems Engineer

Raytheon Missile Systems (1999-2004)

Developed embedded systems for mission-critical defense applications. Delivered $2M annual savings through virtualization.

Sergeant, US Marine Corps

Active Duty (1992-2000)

IT Project Manager. Mission-critical infrastructure. The discipline foundation underneath everything that followed.

Books

Three titles, one through-line

Methodology, personal finance, and memoir. The pattern is taking complex domains and writing the playbook.

Methodology · 2026

Continuous Intent Delivery

With Casey Robinson and Glenn Knepp

The first development framework designed from a blank sheet for the assumption that AI does the implementation. Built from the work of shipping 35 production apps in 15 weeks. Open methodology with a commercial adoption path via CATALYST.

Personal Finance · 2025

The Real Money Guide

A personal-finance system, not a pep talk

Paired with 30+ interactive financial tools at therealmoneyguide.com. System over motivation.

Memoir · 2025

Bullets Don't Fly

A Supply Marine's Memoir

Eight years in the Marine Corps, the bankruptcy that followed, the rebuild. The foundation for every operating mode that came after.

Get In Touch

Where AI investment becomes business value

If your team is generating code 10x faster but business outcomes haven't moved, that's not a tooling problem. It's a methodology problem. The next conversation is the methodology one.