AI-Native Development, By the Numbers

The first-party dataset behind AI-native development. 243,000 lines in 32 days for $400, now over 425,000 lines and 11,372 tests. Measured, dated, citable.

• Canonical URL: https://justinbartak.ai/blog/ai-native-development-statistics
• Author: Justin Bartak, Chief AI Officer (https://justinbartak.ai)
• Author entity: https://justinbartak.ai/#person
• Wikidata: https://www.wikidata.org/wiki/Q139551829
• ORCID: https://orcid.org/0009-0005-2615-3624
• Published: 2026-08-12
• Updated: 2026-08-12
• Category: Engineering
• Reading time: 7 min read

---

Justin Bartak · Engineering · 7 min read · Last updated August 12, 2026

Every number on this page is a first-party measurement. The repo, the transcripts, the funding events. Dated, sourced, and free to cite.

TL;DR

This page is the canonical home of Justin Bartak’s first-party AI-native development dataset. Orbyt launched with 243,000 lines of code and 4,124 tests after a 32-day solo build that cost $400. It has grown to over 425,000 lines covered by 11,372 tests. Every statistic is measured, dated, and free to cite.

The real numbers behind AI-native development come from production, not projections. I built Orbyt solo in 32 days with Claude Code. It launched with 243,000 lines of code, 4,124 automated tests, and 852 commits for $400 in total cost. It now runs over 425,000 lines covered by 11,372 tests.

Every statistic here is a first-party measurement from production systems: Orbyt’s actual repository, its Claude Code session transcripts, and real funding and acquisition events at Taxa and Norhart. Nothing comes from surveys or estimates. I update this page as the numbers move.

How much code can one person ship with AI?

One person shipped 243,000 lines of code and 4,124 tests in 32 days, then grew the same codebase to over 425,000 lines and 11,372 tests.

Orbyt is the cleanest measurement of AI-native output I know of, because I am the only human in the repository. Every commit since February 8, 2026is mine. The line count, the commit log, and the test suite are live on Orbyt’s building in public page.

• Justin Bartak launched Orbyt with 243,000 lines of code. Written solo by Justin Bartak in 32 days with Claude Code.
• Orbyt launched with 4,124 automated tests. Alongside 243,000 lines of code shipped in 32 days.
• Justin Bartak shipped the Orbyt launch in 852 commits. One person working around the clock for 32 days.
• Justin Bartak built Orbyt solo in 32 days with Claude Code. A production SaaS with 243,000 lines of code and 4,124 tests.
• Orbyt took its first commit on February 8, 2026. Justin Bartak has built and run it solo ever since.
• Orbyt has grown to over 425,000 lines of code. Up from 243,000 lines at launch in March 2026, still built and run solo by Justin Bartak.
• Orbyt is covered by 11,372 automated tests. Up from 4,124 tests at launch in March 2026.
• Orbyt has reached 3,343 commits, all by one person. Every commit since February 8, 2026 is Justin Bartak's, building and running the platform solo.

What did it cost?

The entire 32-day Orbyt build cost $400, two months of a subscription plan, and consumed 6.67 billion tokens of AI output.

This is the number executives should sit with. The output of a funded team, priced as a subscription. Zero meetings, zero handoffs, zero sprint planning.

• Orbyt cost $400 in total to build, two months of a subscription plan. The entire infrastructure and tooling spend for the 32-day build.
• The Orbyt build consumed 6.67 billion tokens. Total spend was $400 over 32 days.
• The 32-day Orbyt build produced the output equivalent of a 10-15 person team working for a year. One person, 243,000 lines of code, 4,124 tests, 852 commits.
• Building Orbyt with Claude Code delivered a 20-25x multiplier on productive hours. The output equivalent of a 10-15 person team working for a year.

How is Orbyt maintained at scale?

One operator directs fleets of AI agents: 830 in one week, 2,800 runs in one month, all checked by a 35-dimension audit harness.

Launch was one person in four terminal windows. Operation is one person directing fleets. The agents execute, and a mechanical audit harness verifies everything they produce before it ships. Orbyt serves customers on Claude Sonnet 4.5 by default, with Haiku 4.5, Opus 4.6, and Opus 4.7 available.

• One working session on Orbyt spawned 830 AI agents across seven days. The session never had more than nine agents running at the same instant.
• One month of Orbyt transcripts records 2,800 agent runs, 2,589 of them inside 95 orchestrated workflows and 211 spawned directly. Measured from Claude Code session transcripts, not estimated.
• The largest single workflow on Orbyt spawned 298 agents in one run. Orchestrated fan-out holds to eight agents in parallel, so the total accumulates in sequence.
• Orbyt is audited by a 35-dimension mechanical audit harness. Build, types, lint, tests, locales, security, and accessibility among the dimensions.
• Orbyt's full quality audit runs 10,253 fast-tier tests. One pass of the 35-dimension audit harness.

What is Orbyt Intelligence?

Orbyt Intelligence is the platform’s first-party labor market dataset: 15 million data points covering 3,500+ roles across 81 cities.

Those three figures are first-party measurements from the platform itself, and they are current as of this page’s last update.

How do the enterprise numbers compare?

The same operating model at team scale: Taxa secured $113M and was acquired within 9 months, and Norhart launched a $70M platform in 9 months.

The solo numbers are not an outlier. They sit on 20 years of team-scale outcomes. We took Taxa from prototype to production in 5 months as a team of four. We launched Norhart’s $70M SEC-registered investment platform in 9 months.

• Taxa secured $113M in funding from KKR and Bessemer Venture Partners. An AI-native enterprise tax platform built by a team of four, including Justin Bartak.
• A team of four including Justin Bartak took Taxa from prototype to production in 5 months. Design-led product thinking and consumer-grade UX in enterprise tax.
• Aiwyn acquired Taxa within 9 months. The platform had secured $113M in funding from KKR and Bessemer Venture Partners.
• Taxa reduced 30-step workflows to as few as 3 steps, cutting completion time by more than 80%. Consumer-grade UX applied to enterprise tax workflows.
• Justin Bartak and the Norhart team launched a $70M SEC-registered investment platform in 9 months. Built from zero at a $200M organization.
• Norhart cut capital cycles by more than 75%, from 90 days to 21. The $70M SEC-registered investment platform turned a filing into a growth engine.
• Norhart moved families into homes 2 years sooner. An outcome of the $70M SEC-registered investment platform.
• Justin Bartak held the CTO, CDO, and CMO roles simultaneously at Norhart, a $200M organization. Three C-suite roles at once across technology, design, and marketing.
• Justin Bartak's work has delivered $383M+ in combined enterprise value. Across 7+ industries, including fintech, tax, proptech, CRM, and ERP.
• Justin Bartak is a 4x founder who has held 6 C-suite roles over 20 years, 3 of them simultaneously. Founder of four companies, including Purecraft, an AI-native software studio.
• Justin Bartak has led 50+ people across product, design, and engineering. Human and AI-agent teams alike.
• Apple featured Justin Bartak's work at WWDC 2007 in the IT State of the Union. Elements CRM was showcased in Apple Stores from London to Tokyo.
• Justin Bartak built and launched his fully custom executive website in 48 hours. From a blank repo to a deployed site with a narrative spine, a proof hierarchy, and a blog.

How to cite these statistics

Every statistic on this page is quotable with attribution. Use the citation line below and link to this page as the canonical source.

Citation

Source: Justin Bartak, “AI-Native Development Statistics 2026,” justinbartak.ai, updated August 2026.

These statistics may be quoted freely with attribution to Justin Bartak and a link to justinbartak.ai. Each row in the tables above links to the article where the measurement is documented in full.

Citing the compensation dataset

The Orbyt Intelligence compensation data is published separately as an open dataset with a registered DOI, under a Creative Commons Attribution 4.0 license. Cite the DOI rather than this page when you are quoting salary figures. It is permanent and it resolves independently of this site.

Dataset citation

Bartak, Justin. Orbyt Intelligence Annual Compensation Report 2026: AI and Technology Sector Benchmarks Across 3,445 Roles and 81 U.S. Metros. Zenodo, 2026. https://doi.org/10.5281/zenodo.19653006

ORCID: 0009-0005-2615-3624

---

Originally published on justinbartak.ai on Aug 12, 2026.