Nimble Development
← Back to articles
Strategy

Why We Decided to Build an AI-First Development and Construction Company (An Analytical View)

Why We Decided to Build an AI-First Development and Construction Company (An Analytical View)

In spring of 2025 we started a building and development company with the understanding that we would need to prioritize flexible thinking and adaptability as a core tenet of how we operate. Hence, we named the company Nimble Development so we could communicate the importance of remaining flexible and adaptable to what the future puts in front of us. Another key belief we had in starting this business was that we needed to remain open minded and rely on what data tells us instead of operating from our gut and experience so much. Our goal is to scale what we are doing over time and to do that requires solid processes. In order to understand what data is telling us, we need to be grounded in an understanding of how the world works, what financial principles are at play, and how that all fits into our current point-in-time.

So when we start looking at real estate deals and how we can bring value to a neighborhood, we start with our experience and identify what we think a particular parcel needs in order to bring it up to its highest and best use. Then we ask ourselves questions and research the data to validate our assumptions. Starting out small with flips and one-off new builds, the process is simple usually only involving a real estate agent as a sounding board. However as we get into bigger chunks of land with bigger dollars at stake, I find myself asking more and more questions that only data can help answer.

For example - there has been A LOT of material written about housing affordability and why we as a nation are in this mess (California as the poster-child). How do we sort through what is real and what matters for an individual neighborhood? As we survey different states, cities and neighborhoods, we find there is such a side range of scenarios at play.

Econ 101 teaches us that the supply and demand should solve for this problem; "We just need to build more." It sounds intuitive and simple. But if you look at actual market data — inventory levels, days on market, median incomes — the picture is more complicated and more actionable than "just build more" suggests. What should we believe is real and how do we parse through it?

Is the Conventional Wisdom Wrong?

In markets across California, Texas, and the Mountain West, active inventory has risen 30-50% from pandemic lows — and in some Central Valley metros, the shift is even more dramatic. Stockton's active listings have roughly tripled since early 2024, from ~315 homes to over 1,100. Months of supply in Patterson have reached 8.4 — deep into buyer's market territory. Across the region, months of supply are approaching balanced-market territory. Builders are offering rate buydowns and closing cost credits. By every traditional supply metric, the market is loosening.

And yet, median home prices remain 5-7x median household incomes. First-time buyers still can't close. Working families are still priced out.

This isn't a supply quantity problem. It's a supply alignment problem. We're building homes — just not the ones people can actually afford.

The Income-Price Disconnect

Here's the math that matters. Take Stockton, California — a growing Central Valley metro absorbing migration from the Bay Area. The median household income is approximately $77,000 (Census ACS 2024). At 30% of gross income (the traditional affordability threshold) and a 6.1% mortgage rate, that family can afford roughly $330,000 in home price.

Now look at what's actually being built. The cheapest new construction community in Stockton starts at $443,000. In nearby Modesto, it's $471,000. In Manteca, $552,000. The gap between what the median household can afford and what the median new home costs is over $110,000 — and that's using the most affordable product on the market. That gap doesn't close with more units at the same price point. It closes with different units at a different price point.

This pattern isn't unique to the Central Valley. Nationally, the median household earns approximately $84,000 (Census 2024), yet nearly 75% of U.S. households cannot afford a median-priced new home according to the NAHB.

What Tools Exist to Understand This?

If this problem is national in scale, surely someone has built the tools to measure it. They have — sort of. There are several established indices that track housing affordability:

  • The NAR Housing Affordability Index measures whether a median-income household can qualify for a mortgage on a median-priced home. It's the most widely cited, updated monthly, and treats 100 as the threshold — below that, the median family can't qualify.
  • The Atlanta Fed's HOAM adds taxes, insurance, and PMI to the cost picture, and drills down to the metro-county level — a real improvement in granularity.
  • The NAHB/Wells Fargo Cost of Housing Index tracks what share of income goes to a mortgage payment, reported quarterly.
  • The ULI/RCLCO Home Attainability Index takes the broadest view — affordability, connectivity, racial disparity, and growth — across MSAs, counties, and census tracts.
  • The H+T Index from the Center for Neighborhood Technology adds transportation costs to the equation, benchmarking the combined burden at 45% of income.

All of them ask some version of the same question: can people afford what exists?

None of them ask the question that matters to a builder, politician or concerned community member: is anyone building what people can afford?

That distinction is everything. These indices are diagnostic — they tell you the patient is sick. What we needed was something operational — a tool that tells a builder exactly where the gap is and whether the local construction pipeline is filling it. None of the existing indices combine income data, FHA loan limits, construction permits by unit type, and real-time market dynamics into a single score. None of them update from live data feeds. And none of them are designed to drive a build decision.

So we built one. Our Housing Attainability Calculator makes this visible for any metro in the country. Pick a market. See the gap. The data speaks for itself.

Misallocated Supply

The real story isn't in aggregate inventory numbers — it's in how inventory distributes across price bands.

In early 2024, the split was stark: homes above $500K sat 45+ days while attainable homes below the FHA limit moved in under 20. That velocity gap has since narrowed — not because luxury sped up, but because the entire market slowed. As of January 2026, even median-priced homes in Stockton sit 54 days. The broader Modesto area: 35. Patterson: 101.

The product-market mismatch hasn't gone away — it's just been complicated by a broader demand pullback at 6%+ mortgage rates. For builders, this means the attainability thesis is still valid, but the margin for error is thinner. You can't just "build down" — you have to build down precisely, at a cost basis that survives a slower absorption environment.

The permit data tells the same story. In November 2025, San Joaquin County pulled 141 single-family permits and zero — literally zero — permits for duplexes, triplexes, or fourplexes. Stanislaus County: 45 single-family, 2 duplex, 0 three-four unit. Meanwhile, Stanislaus issued 216 permits for 5+ unit apartment complexes — the county is building big apartments and single-family homes, completely skipping everything in between. The missing middle isn't just underbuilt. In our target markets, it's functionally non-existent.

This is both the problem and the opportunity. Regulatory barriers (parking minimums, lot-size requirements, utility connection fees scaled per unit) make small multifamily uneconomic at conventional cost structures. Solving the cost equation — through design standardization, modular procurement, and AI-optimized scheduling — is how a vertically integrated builder can operate where conventional developers can't.

What the Data Says Right Now

Here's a snapshot of our Central Valley target markets as of January 2026:

| Market | Median Price | DOM | Months Supply | Signal | |--------|-------------|-----|---------------|--------| | Stockton | $520,000 | 54 | 4.0 | Softening | | Modesto area | $478,500 | 35 | 2.9 | Balanced | | Patterson | $467,000 | 101 | 8.4 | Buyer's market | | Manteca | $585,000 | 69 | 2.6 | Balanced | | Tracy | $652,000 | 40 | 6.0 | Softening |

Several things jump out. Patterson and Tracy — both markets with significant new construction — show elevated supply. This is where product-market mismatch is most visible: builders brought inventory at price points above what local incomes support, and absorption has slowed accordingly.

Meanwhile, the Modesto area and Manteca remain supply-constrained under 3 months. Mortgage rates at 5.98% (as of February 2026) continue to suppress demand, but the income-price gap hasn't closed. The gap is structural, not cyclical.

Why AI Changes the Equation

We don't use "AI" as a marketing label. We use it as operating infrastructure.

Our data pipeline ingests five public data sources in real time:

  • Census ACS — Median household income by metro, county, and census tract
  • Redfin Data Center — Sale prices, days on market, inventory, months of supply, sale-to-list ratios
  • FRED — Mortgage rates and economic indicators
  • HUD — FHA loan limits by county
  • Census Building Permits Survey — New construction pipeline by unit type

These feeds combine into what we call the Nimble Attainability Index (NAI) — a composite score from 0 to 100 that measures how well a market's housing supply matches its residents' ability to pay.

This isn't about predicting the market. It's about measuring product-market fit in housing. When the NAI drops below 40, it means the local construction industry is building above the income curve. That's an opportunity for a builder disciplined enough to target the gap.

As of Q1 2026, here's how our active markets score:

  • Patterson: NAI 52 — The income-price gap is wide ($77K income vs. $467K median), but inventory is building and absorption is slow. Opportunity exists for attainable product, but cost discipline is critical.
  • Modesto area: NAI 61 — Tighter supply, lower price point relative to neighboring metros, and strong FHA utilization signal genuine demand in the attainable band.
  • Stockton: NAI 44 — Softening. Supply has outpaced absorption at current price points. The index is telling us to be selective, not aggressive.

We publish these scores not because they always tell a bullish story, but because transparency is what separates a data-driven company from one that just says the words.

The 9 Metrics We Watch

Our full attainability model tracks nine metrics that together paint a complete picture of housing market alignment:

  1. Days on market by price band — Do attainable homes sell faster than luxury? (Demand signal)
  2. Months of inventory by price band — Where is supply accumulating? (Supply signal)
  3. Absorption rate by product type — What's actually selling: SFR, townhome, ADU? (Product signal)
  4. Rent-to-price ratio — Is renting cheaper than owning? By how much? (Tenure signal)
  5. FHA utilization rate — What share of inventory falls under FHA limits? (Access signal)
  6. Building permits by unit type — Is the pipeline adding the right product? (Pipeline signal)
  7. Income-to-price gap — Can the median household afford the median home? (Affordability signal)
  8. Sale-to-list ratio + price reductions — Are sellers adjusting to reality? (Market health signal)
  9. ADU and infill permit share — Is the market adding density where it's needed? (Efficiency signal)

Update (March 2026): We originally stated that four of these metrics require MLS access. That was wrong. All nine run on free public data — Census ACS, Redfin Data Center, FRED, HUD, and Census Building Permits Survey. No MLS feed is used anywhere in the system. See Anatomy of the NAI for the full technical breakdown. Our public calculator uses the same data sources as our internal models.

What "AI-First" Actually Means

When we say AI-first, we don't mean chatbots on our website or AI-generated floor plans. We mean:

Data pipelines that replace gut feel. Every acquisition decision starts with the NAI score for that market. If the score says the gap is closing, we don't chase the deal.

Market signals that update weekly. Our systems pull fresh data from Census, Redfin, and FRED on a rolling basis. We don't wait for quarterly reports.

Deal scoring that compounds learning. Every completed project feeds back into our models. Actual vs. estimated costs. Actual vs. projected sale price. Actual days on market vs. predicted. Each cycle sharpens the next.

Cost estimation from historical data. Our construction cost models are trained on our own project data — not national averages that blur the signal. When we say a 1,400 sq ft home costs $185/ft to build in Patterson, that's derived from line-item actuals on our last three builds.

This is vertical integration powered by data. We control the acquisition, entitlement, construction, and disposition process — and we instrument every step. The result is a feedback loop that makes each project cheaper, faster, and more aligned with what the market actually needs.

We should be honest about what we don't know yet. Our construction cost models are early — trained on a small number of completed projects. The feedback loop we describe is real, but it's in its first iterations. We're betting that the compound effect of instrumenting every phase will produce meaningful cost advantages within 18-24 months. That bet is unproven. What is proven is that the current approach — building on intuition with quarterly data — consistently produces homes that most people can't afford.

Try It Yourself

We've made our core attainability analysis available for free. The Housing Attainability Calculator lets you select any major metro area and see:

  • The gap between what residents can afford and what homes cost
  • How supply, FHA access, and the construction pipeline score
  • An overall Nimble Attainability Index for that market

It's the same data infrastructure we use internally — just the public-data subset. If you're an investor, builder, or policy maker who wants to see where the real opportunities are, start there.

If you want to go deeper — partner on a project, invest alongside us, or license our data platform — get in touch.

The market data behind these scores updates weekly. This article was last verified against live data on February 26, 2026. If you're reading this more than 90 days after publication, the specific numbers may have shifted — but the structural dynamics we describe (income-price gap, misallocated supply, missing middle deficit) move on multi-year timescales.

Sources & References

Data Sources Used in NAI

Housing Affordability & Attainability Indices

Missing Middle Housing Research

Cited Statistics

← All articles