AcrePace

1. One-liner

Warns a small California grower they are on pace to overpump months before the $500 penalty bill lands.

2. Trend signal — why now?

The clock is five weeks out and it is not a rumour — it is a rate, a date, and a published schedule.

California’s Sustainable Groundwater Management Act (SGMA) has been law since 2014, but for most San Joaquin Valley growers the enforcement has been theoretical. That ends now. Multiple Groundwater Sustainability Agencies (GSAs) ran Water Year 2026 (1 Oct 2025 – 30 Sep 2026) as tracking-only — pumping measured, no fees charged. Fees and penalties begin in Water Year 2027, starting 1 October 2026.

The rates are published and brutal for a small operation:

GSA	Free/base allocation	Overage rate
East Kaweah	0.82 AF/acre native yield, 2.5 AF/acre hard cap	$500/AF + 1:1 cut to 2028 allocation
Mid-Kaweah (WY2026)	1.58 AF/acre	$500/AF + next-year loss
Greater Kaweah (WY2026)	1.37 AF/acre	$150 / $250 tiers, $500 prohibited tier
South Fork Kings	0.86 AF/acre	$500/AF + next-year reduction
Southwest Kings	0.66 AF/acre	$500/AF
Madera County	ETAW inches	$300/AF ramping to $500 by 2040
Eastern Tule	~1 AF/acre	Transitional water $245/AF

Two things make this a product rather than a news story.

First, the measurement is lagged. Most of these GSAs do not meter wells — they bill on satellite evapotranspiration from Land IQ. Land IQ’s own FAQ states results are “delivered within 25 days of the end of the previous month.” Eastern Tule tells landowners data is “available approximately 30 days after the end of each month.” Land IQ’s 2026 flyer concedes the point directly: the monthly product’s results “do not come quickly enough for irrigation management decisions.” So a grower sees July’s water use in late August — and the water year closes 30 September. Overage is discovered when the invoice arrives.

Second, nobody sells the grower a projection. Every accounting platform in this market is sold to the agency. Kaweah Water Dashboard Phase 1 is explicitly “historical evapotranspiration (ET) data.” The EDF/California Water Data Consortium Groundwater Accounting Platform is deployed by GSAs. GSA Water Dashboard is priced on “a sliding scale based on your agency size.” East Kaweah’s own allocation page tells growers to phone the office at (559) 697-6095 to find out where they stand.

The consequence is already documented. Rosie Lee, a Hmong-American farmer on 19 acres in Merced County, received a $15,335 overextraction invoice: “It just gave me a real shock, almost like a heart attack.”

Provenance:
  - Signal 1 (Demand): UC ANR / CAFF report "Managing Groundwater with Small Farms in Mind" finds small farms face the same cutbacks and fees as large ones; in Tule and Tulare Lake subbasins over 40% of pumpers account for only 1% of water use; 85% of 170 farmers surveyed feared shutdown — https://ucanr.edu/blog/food-blog/article/sgma-challenges — 29 July 2026
  - Signal 2 (Feasibility): OpenET evapotranspiration data published in the Google Earth Engine catalog under a CC-BY-4.0 licence (30m resolution), making field-level ET legally usable in a commercial product with attribution — https://developers.google.com/earth-engine/datasets/catalog/OpenET_ENSEMBLE_CONUS_GRIDMET_MONTHLY_v2_0 — accessed 24 Aug 2026
  - Signal 3 (Economic): GSAs set overage at $500/acre-foot with penalties beginning Water Year 2027 (1 Oct 2026) after a tracking-only WY2026; East Kaweah sells mitigation credits at the same $500/AF — https://ekgsa.org/allocation/ — accessed 24 Aug 2026
  Category: Underserved niche

3. The opportunity

Every piece of software in California groundwater is built for the party doing the charging. GSA Water Dashboard, the Groundwater Accounting Platform, Kaweah Water Dashboard — all of them are agency procurements. The grower gets, at best, a read-only portal showing what already happened, updated monthly, 25–30 days in arrears.

That is a billing system with a customer-facing tab bolted on. It answers “what did you use?” It does not answer the only question that matters between April and September: “at this rate, where do I finish, and what will it cost me?”

There is one grower-side product — AgWaterAI, which sells an SGMA allocation tracker and irrigation forecasting at $20/acre/year (BYOS) or $80/acre/year (managed). It is a real, shipping product with named GSAs and published validation figures, and it is honest about its own limits. But it is fundamentally an irrigation-intelligence and sensor-analytics platform: BYOS means bring your own sensors, and the forecasting rests on soil moisture probes and pump runtime. That is the right product for a well-capitalized permanent-crop operation with telemetry already in the ground.

It is the wrong product for the 77% of San Joaquin Valley farms under 100 acres. A 40-acre citrus grower is not installing Sentek probes and an IrriMax subscription to find out he is 30 acre-feet over. He needs one number, monthly, on his phone, derived from data that already exists about his field whether he instruments it or not.

The wedge is sensor-free pacing. Satellite ET is already how the GSA is going to bill him. If the agency’s meter of record is a satellite, the grower can run the identical projection against the identical data source — and see it in April instead of December.

The second half is the part nobody does at all: remedy economics. Being told you are 30 AF over is not a decision. The decision is whether $15,000 of penalty is cheaper than buying credits at $500/AF, cheaper than deficit-irrigating and accepting yield loss, or cheaper than fallowing the weakest block. East Kaweah’s mitigation credits price at exactly the penalty rate, which means the arbitrage is real and computable.

4. Target market

• Primary customer: Owner-operators farming 20–300 irrigated acres in a San Joaquin Valley GSA with an adopted penalty-bearing allocation — Kaweah subbasin (East, Mid, Greater), Kings County (South Fork, Southwest), Eastern Tule, Madera County. Skew to permanent crops — almonds, pistachios, citrus, table grapes — where fallowing means killing trees, so the decision is genuinely agonizing and genuinely worth money.
• Why they buy, in their words:
  • Rosie Lee, 19 acres, Merced County, on her $15,335 invoice: “It just gave me a real shock, almost like a heart attack.” And: “It’s preventing me from growing what I used to grow… We’re afraid that we do not make enough to pay for the water.” (LA Times, 31 July 2026)
  • Ralph Alcala, Kings County: “It’s overwhelming and it’s frustrating… I’m just trying to understand these policies myself.” (SJV Water, 6 Oct 2025)
  • Ceil Howe, South Fork Kings GSA board, on adopting the allocation: “I think it’s a necessary evil, but it scares me and it’s going to be real expensive… It might be sustainable for the groundwater, but it’s not sustainable for the farmer.” (SJV Water, 22 Jan 2026)
  • Dan Fiser, Kings County, on the timing of the decision: “I’ve got to make decisions this month, next month, whether I’m gonna spend money on those trees or pull them out now.” (SJV Water, 19 Dec 2025)
  • Arshdeep Singh, Punjabi American Growers Group: “We get growers calling us, saying, ‘Banks are sending us bankruptcy letters.’” and “our growers did not even know what SGMA stood for… They should have been counseled far better than they were.” (Ag Alert, 4 Dec 2024)
  • Makhan Singh, almond grower, Madera: “It was like a hammer to the head.” (Ag Alert, 4 Dec 2024)
  • Doug Freitas, small Kings County farmer: “I’m disgusted and discouraged. It’s easy to police the little guys, but who’s watching the big ones?” (SJV Water, 13 March 2024)
  • Travis Millwee, Pixley/Lower Tule GSAs, on the state’s GEARS reporting portal: “the whole GEARS portal is a mess” — staff tried uploading for 50 landowners; four succeeded. (SJV Water, 13 April 2026)
• Rough TAM reasoning: The San Joaquin Valley has roughly 34,500 farms, ~77% of them under 100 irrigable acres (PPIC). Statewide there are 261 GSAs and 21 critically overdrafted basins, 8 in the SJV. Not all are under penalty regimes yet, but the direction is one-way. A realistic serviceable base in the penalty-adopting Kaweah/Kings/Tule/Madera footprint is in the 8,000–15,000 farm range, expanding each year as more GSAs switch from tracking to charging.
• Why now for them: WY2026 was free. WY2027 is not. The first genuinely expensive water year in California history starts 1 October 2026, and the first bills land after it closes.

5. Product sketch (MVP)

• Enter your parcels (APNs) and your GSA. The product knows each GSA’s allocation formula, base rate, tier structure, hard cap and penalty rate, because they are all published.
• Field-level water use from satellite ET, the same measurement basis the GSA bills on — no meters, no probes, no hardware.
• The pacing number: “You are 62% through the water year and have used 78% of your allocation. On current pace you finish 31 AF over, which is $15,500 at $500/AF, plus a 31 AF cut to next year.”
• Remedy comparison: side-by-side cost of (a) paying the penalty, (b) buying mitigation credits where the GSA sells them, (c) deficit-irrigating a named block for the rest of the season, (d) fallowing the weakest block — with the dollar delta on each.
• Threshold alerts by SMS at 50/70/85/100% of allocation, in English, Spanish and Punjabi — three of the languages the growers in this footprint actually speak.
• A one-page season statement the grower can hand to a lender, a landlord, or the GSA when contesting a bill.
• Carryover and multi-year view, because the one-for-one clawback means this year’s overage is also next year’s problem.

6. AI angle — what’s load-bearing

Two places, and the product does not exist without either.

Projection under uncertainty. Turning eight months of noisy 30-metre satellite ET into a defensible end-of-season estimate is a real modelling problem: you are extrapolating crop water demand across the remaining season conditioned on crop type, phenological stage, planting date, weather forecast and historical ET curves for that field. A linear run-rate is worse than useless here — almond ET is heavily back-loaded, so naive extrapolation in June badly understates the finish and tells the grower he is fine when he is not. Getting the shape of the curve right is the product.

Reconciling the grower’s world to the GSA’s ledger. Parcel boundaries, APNs, field boundaries and cropped area do not line up cleanly; allocation rules are written in PDFs with different tier structures per agency; and the grower’s own understanding of what he farms rarely matches the parcel roll. Parsing the rulebooks and matching messy field geometry to billed parcels is exactly what current models are good at and what made this a two-person build rather than a twenty-person one.

Strip the AI out and you have a spreadsheet that divides usage by allocation — which is what the GSA portals already are, and why nobody pays for them.

7. Localization angle (if any)

Not a country play — a subbasin play, which is the same logic at smaller scale. Every GSA has its own allocation formula, tier boundaries, penalty rate, credit market and appeal process, all published in PDFs and board minutes. That per-agency rule encoding is tedious, unglamorous, and the reason a generic “water tracker” cannot compete: being exactly right about East Kaweah’s 0.82 AF/acre native yield and 2.5 AF/acre hard cap is the whole value.

Language matters more than usual here. The affected grower population includes large Hmong, Punjabi and Latino farming communities — the documented complaints come disproportionately from exactly these operators, several of whom say plainly that nobody explained SGMA to them. Spanish and Punjabi SMS alerts are not a nice-to-have; they are the distribution channel.

8. Business model — path to $1M–$5M ARR

• Pricing: $15/acre/year, minimum $600/year, capped at 500 acres. A 60-acre grower pays $900/year; a 200-acre grower pays $3,000. Undercuts AgWaterAI’s $20/acre BYOS while requiring no hardware at all — the sensor spend is the real cost being avoided, not the subscription.
• ACV: ~$1,400 blended across a 20–300 acre base.
• Why the price works: a single avoided overage on a mid-size farm is $10,000–$25,000 at $500/AF. This is a rounding error against the penalty it prevents, which is the only pricing conversation that matters.
• Rough math to $1M ARR: 715 farms × $1,400 = $1.0M. Against a serviceable base of 8,000–15,000 penalty-exposed farms, that is 5–9% penetration.
• Rough math to $5M ARR: 3,600 customers. Requires expanding beyond the Kaweah/Kings/Tule/Madera core as more of the 261 GSAs adopt penalty regimes, plus the adjacent buyers below. Realistic by 2029–2030, not by 2028.
• Expansion path: (1) Landlords and lenders — absentee owners and ag lenders want water-risk visibility across a portfolio of leased parcels, priced per portfolio; (2) PCAs, irrigation consultants and ag CPAs reselling a multi-client seat; (3) credit-market brokerage — once you know who is over and who is under in the same subbasin, you know both sides of every mitigation-credit trade before either party does; (4) geographic — Arizona’s AMAs, Colorado’s Republican River basin and Kansas LEMAs run structurally identical allocation-and-penalty regimes.

9. Go-to-market wedge — first 100 customers

The distribution here is unusually tractable because the target list is a public record. GSAs publish parcel rolls, allocation registries and assessment lists; county assessors publish APN-to-owner mappings.

1. Work the parcel roll directly. Pull the East Kaweah, Mid-Kaweah, Greater Kaweah, South Fork Kings and Southwest Kings parcel and assessment lists, join to county assessor owner-of-record data, and filter to 20–300 acre holdings. That is a named, addressed list of a few thousand operators. Send each one a free, personalized pacing report for their actual parcels — their allocation, their ET-derived use to date, their projected finish and the dollar penalty — as a physical mailer plus SMS. The report is the demo. Nobody ignores a letter with their own APN and a four-figure number on it.
2. Ride the 1 October deadline with the grower associations. Punjabi American Growers Group, Nisei Farmers League, California Farm Bureau county chapters, the UC Small Farms Network and CAFF are all actively organizing around exactly this issue and publishing about small-farm SGMA distress. Offer free multilingual “what your allocation costs you in WY2027” sessions through them in September and October. These organizations are looking for something concrete to hand their members.
3. Be the free penalty calculator that ranks. A public, no-login calculator per GSA — “what will I owe East Kaweah?” — captures the searches growers are already running as bills land. AgWaterAI has proven this channel works by building one; the counter is to be better, cover more agencies, and gate the projection rather than the arithmetic.
4. Sell through the people already in the field. PCAs, irrigation-system dealers and ag CPAs visit these farms monthly and are being asked SGMA questions they cannot answer. A referral seat lets them show up with a number.
5. Bill-shock retargeting, October–December 2026. The first real penalty invoices in California history hit in Q4 2026. Every grower who opens one becomes an inbound lead for the following water year — and unlike most bill-shock moments, this one repeats annually and gets worse.

10. Build complexity — justification

Medium. The data spine is off-the-shelf: OpenET is published in the Google Earth Engine catalog under CC-BY-4.0, which permits commercial use with attribution, and county parcel/APN data is public. There is no hardware, no field installation, no device fleet.

The custom work is threefold: encoding each GSA’s allocation rules faithfully (tedious, and the actual moat), matching messy field geometry to billed parcels, and building a seasonal ET projection that is right about crop-stage curves rather than naively linear. Call it 14–18 weeks for two people to a credible v1 covering five to eight GSAs in the Kaweah and Kings footprint.

One important build note: do not build on the OpenET REST API. Its terms of service grant only a “personal, noncommercial use” licence and prohibit developing “applications that interact with our Services without our prior written consent.” The Earth Engine CC-BY-4.0 distribution of the same underlying data is the correct legal path, with attribution. Confirm licence scope in writing before launch regardless — see risk flags.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Public ET data under CC-BY-4.0, public parcel records. No licensed activity. Advisory only — the GSA remains the system of record.
Ethical — no harm / dark patterns	✅	Helps the least-resourced party in an asymmetric regulatory regime understand a bill before it arrives.
Market exists (evidence above)	✅	Published $500/AF penalties, a dated 1 Oct 2026 start, documented five-figure invoices, an existing paid competitor at $20–80/acre.
1–5 person team can build this	✅	Two people, 14–18 weeks. No hardware.
Launchable with <$50K / ₹40L	✅	Data is free; cost is compute, mailers and two salaries.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	18/20	Five-figure invoices on 19-acre farms, bankruptcy letters, a one-for-one clawback that compounds the mistake into next year, and a hard 1 Oct 2026 start. Docked 2 because it is annual, not daily, pain.
Demand evidence	15	13/15	Published penalty schedules, a funded competitor charging $20–80/acre, a dozen sourced verbatim grower quotes, and a July 2026 academic report. Docked 2: no grower is on record saying “I would pay for a projection” — the demand is inferred from the penalty, not from purchase intent.
Build feasibility	15	12/15	Free licensed data, no hardware, but per-GSA rule encoding and seasonal ET projection are real work. 14–18 weeks for two.
Distribution clarity	15	13/15	The customer list is a public parcel roll joined to assessor records, and the personalized pacing report is a demo that mails itself. Docked 2 for the known difficulty of converting older, non-digital, sometimes non-English-speaking operators.
Revenue mechanics	15	11/15	Pricing is anchored to a $500/AF penalty and undercuts a live competitor. Docked 4: small wallets, seasonal purchasing, and a $1M target that needs 715 customers out of a base that is real but not enormous.
Time to first revenue	10	8/10	The 1 Oct 2026 deadline and Q4 bill shock create a natural purchase moment, and the free report converts warm. Not 9–10 because growers buy on the crop calendar, not on demand.
Defensibility	10	5/10	Per-GSA rule encoding and accumulated field-boundary/crop history compound, and the credit-market position is genuinely defensible later. But AgWaterAI is already here, the data is public by design, and any GSA could bundle a projection into its own portal. Execution and focus, not moat.
Total	100	80/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required

Geospatial and remote-sensing competence is required, and so is somebody who can sit in a GSA board meeting and read an allocation resolution correctly. A technical founder paired with a Central Valley ag advisor or ex-GSA staffer is the right shape. This is not a product you can build correctly from outside the valley.

Key assumptions to validate (3–5)

1. Assumption: Satellite-ET-derived projections are accurate enough by mid-season that a grower will act on them. How to test: Backtest against WY2026 tracking data — that year was measured but not billed, so the actual outcomes exist. Run the projection as of 1 May and 1 July 2026 for 200 parcels and measure error against the 30 Sept actual. If mid-season projection error exceeds roughly ±15%, the alert is not actionable and the product does not work.
2. Assumption: Growers under 100 acres will pay ~$900/year for a projection when the GSA portal shows history for free. How to test: Mail 300 personalized pacing reports with a pre-order price and count paid conversions, not replies.
3. Assumption: The Earth Engine CC-BY-4.0 path is genuinely usable commercially and does not collapse back into the restrictive OpenET API terms. How to test: Written confirmation from OpenET and an IP lawyer’s read before a single dollar is charged. This is a go/no-go, not a nice-to-have.
4. Assumption: GSAs will tolerate, rather than fight, a third party projecting against their allocations. How to test: Present to two GSA boards in public comment and gauge the response — hostility here changes the distribution strategy entirely.
5. Assumption: Enough GSAs follow the Kaweah/Kings pattern to sustain expansion past $1M. How to test: Track allocation-adoption resolutions across the 21 critically overdrafted basins through 2027.

Risk flags

1. Competitive: AgWaterAI already sells grower-side SGMA allocation tracking with forecasting at $20/acre BYOS and $80/acre managed, has real deployments and publishes honest validation figures. The differentiation — sensor-free operation and remedy economics — is real but narrow, and they can close it. This is the single biggest risk and is the reason defensibility scores 5.
2. Data-licence: OpenET’s own API terms prohibit commercial use and third-party applications without written consent. The CC-BY-4.0 Earth Engine distribution appears to be a clean path, but the GEE monthly ensemble collection runs only through Dec 2024, so current-season data sourcing needs to be nailed down precisely. Get this in writing before launch.
3. Accuracy liability: If the product says “you are fine” and the grower gets a $20,000 bill, that is a lawsuit and a dead brand. Requires conservative projection defaults, explicit confidence bands, and hard contractual disclaimers that the GSA is the system of record.
4. Platform dependency: If GSAs bundle forecasting into the agency portals they already pay for — technically easy, and the Kaweah Dashboard’s own Phase 2 and 3 roadmap points toward a “landowner water bank account” and trading — the standalone product gets squeezed. Watch that roadmap closely.
5. Market timing at the tail: PPIC projects SGMA removes 500,000–900,000 acres from full production by 2040. Some fraction of this customer base does not survive to renew, and the ones who exit are disproportionately the small farms this product targets.

14. Structured verdict

Score:                  80/100
Verdict:                STRONG GO
Confidence:             Medium
Best-fit builder:       Geospatial/remote-sensing technical founder paired with a
                        Central Valley ag advisor or former GSA staffer
Time to revenue:        8-14 weeks (deadline-driven; 1 Oct 2026 start, Q4 bill shock)
Capital to launch:      $15-25K (data is free; cost is compute, mailers, two salaries)
Top 3 assumptions to validate first:
  1. Mid-season ET projection accuracy — backtest 200 parcels against WY2026
     tracking-year actuals at 1 May and 1 July; need better than ±15%
  2. Willingness to pay ~$900/yr — mail 300 personalized pacing reports with a
     pre-order price; count paid conversions, not replies
  3. Commercial usability of CC-BY-4.0 Earth Engine ET data, in writing, from
     OpenET and an IP lawyer, before charging anyone
Kill criteria:
  - Abandon if mid-season projection error exceeds ±15% against WY2026 actuals
  - Abandon if fewer than 5 of 300 personalized pacing reports convert to paid pre-orders
  - Abandon if the CC-BY-4.0 commercial path is not confirmable in writing
  - Abandon if AgWaterAI or a GSA portal ships sensor-free allocation projection
    with remedy economics before v1

15. Next step — 1-week validation sprint

• Day 1–2: Pull WY2026 ET and allocation data for 200 parcels across East Kaweah and South Fork Kings. WY2026 was the tracking-only year, so measured usage exists with no penalty attached — a free backtest set. Build the projection and run it as of 1 May and 1 July against the 30 Sept actual. Falsifiable output: median absolute projection error.
• Day 3–4: Take the 40 parcels where the model says the operator would have finished over allocation. Generate a personalized one-page pacing report for each — their APN, their allocation, their projected finish, their dollar penalty. Mail all 40 with a pre-order offer at $15/acre, minimum $600.
• Day 5: In parallel, email OpenET requesting written clarification on commercial use of the CC-BY-4.0 Earth Engine distribution, and book two GSA board public-comment slots to gauge agency reaction.
• Decide go / no-go on: median projection error at 1 July better than ±15%, and at least 3 of 40 mailers converting to paid pre-orders within 14 days, and no written blocker from OpenET.

If the projection is not accurate by July, there is no product — the alert arrives too late to change anything, which is precisely the failure mode of the incumbent portals.