AROYA · 2-Week Facility Audit · AUTOMAIT ReadinessAROYA · Auditoría de 2 Semanas · Preparación AUTOMAIT

Insa · Easthampton

Facility 4685 · 16 Bloom rooms · Window 2026-06-30 → 2026-07-14 · Prepared for Mike Ronnie

1 · Headline1 · Resumen

Mesh health

102 / 102

radios reporting · battery 3.60–3.67 V · RSSI median −60 dBm

Rooms AUTOMAIT-blocked

Bloom 1.5 · 3.2 · 3.4 — substrate probe faults

Rooms AUTOMAIT-ready

Bloom 1.3 · 1.6 · 3.1 · 3.5

Substrate EC spread

2.7 → 10.0

mS/cm across rooms — a 3.7× range

The one-sentence version. Insa's AROYA mesh is in excellent health — every one of the 102 radios is reporting, batteries are full, signal is strong, and travel times are near zero — so the network is not what stands between Insa and AUTOMAIT. What stands in the way is three failed substrate probes and two dark climate stations, which between them blind three of sixteen rooms entirely, plus a 3.7× spread in substrate EC across rooms that should be running the same program. Fix the sensors, converge the EC, and four rooms are ready to pilot AUTOMAIT today.

Why this matters for the AUTOMAIT conversation. AUTOMAIT's control signal is the substrate probe. It reads VWC and pore EC, decides when to fire a shot, and closes the loop on dryback. A room whose probe is pegged at the channel ceiling — Bloom 1.5 at 25 mS/cm, Bloom 3.2 at 30 mS/cm — does not give AUTOMAIT a degraded signal, it gives it a wrong one. And a room with no VPD input at all (Bloom 3.4) gives it nothing to model transpiration from. Sensor integrity is not a prerequisite you do after you buy automation. It is the thing you are automating on.

Top three findings

#	Finding	Number	So what
1	Three rooms are flying blind on substrate — each has exactly one substrate zone and that zone is faulted	Bloom 1.5 pore EC pegged at 25.0 mS/cm · Bloom 3.2 pegged at 30.0 mS/cm with VWC at 0.0% · Bloom 3.4 VWC collapsed to 0.8%	These three rooms cannot run AUTOMAIT at all, and cannot be graded agronomically today. The mesh radios serving them are healthy — it is a probe-layer fault, and it is cheap to fix.
2	Substrate EC is not converged across the floor — rooms on the same wing are running wildly different root-zone chemistry	Bloom 1.4 at 2.72 mS/cm and falling (3.41 → 2.20) vs Bloom 1.3 at 10.01 mS/cm and climbing (9.31 → 11.71) — a 3.7× range	Substrate EC is the steering output. A 3.7× spread means the rooms are not being steered to a common intent. No automation can reconcile this — it is a feed and P3 decision that has to be made first.
3	Two climate stations are dark, one of them for the entire window	Bloom 3.4: 0 air-temp samples in 14 days. Bloom 3.2: silent since 2026-07-03 — 488 samples where a healthy room returns ~2,240	No VPD input means no transpiration model. These rooms are invisible to both the grower and to any autonomous controller.

What we could NOT verify this run. The AROYA SPA session is invalidated (user token cutoff), so this audit has no harvest-group roster, no cultivar list, and no read of the AUTOMAIT configuration itself. Every phase label below is signal-inferred from steering signatures and is labelled as such. We are not claiming AUTOMAIT is or is not currently enabled at Insa — we cannot prove either from public_api, so we do not assert it. Re-login at app.aroya.io restores all of it.

2 · Prior-Commitment Review2 · Revisión de Compromisos

Graded against the recommendations shipped in the Insa 2-week audit of 2026-06-12 → 2026-06-26. Note that the prior audit ran on the Bloom 1 wing only and also had no harvest-group roster, so its phase labels were signal-inferred too.

Owner	Prior recommendation (2026-06-26)	Then	Now	Grade
Grower	Bloom 1.3 — bring dryback into the phase band. Dryback 17.7 pp was below the 20–30 pp early flower / stretch band.	17.7 pp	16.1 pp	NOT DONE — dryback went the wrong way (−1.6 pp). And the room has now stacked substrate EC to 10.0 mS/cm, so the shallow dryback is actively trapping salt. This is now a bigger problem than it was.
Grower	Bloom 1.6 — bring dryback into the phase band. Dryback 9.7 pp was below the 12–18 pp bulking band.	9.7 pp	23.3 pp	DONE — dryback deepened by 13.6 pp and the room now sits squarely in the 20–30 pp early-flower band on a 172-min P0 delay. Best-executed change on the floor.
Owner	Bloom 1.6 — correct substrate EC. Pore EC 3.8 mS/cm was below the 4.0–7.0 band.	3.8 mS/cm	5.57 mS/cm (3.70 → 7.09)	DONE — a real, sustained, monotone stack into band. Paired with the dryback fix above, Bloom 1.6 went from the facility's problem room to its reference room in one cycle.
Process	Re-verify phases against a live harvest-group roster. Phases were signal-inferred; a roster check tightens every target band.	SPA down	SPA still down	NOT DONE — the AROYA SPA session is still invalidated (user token cutoff). Phases in this report are signal-inferred for the second consecutive audit. A single re-login at app.aroya.io clears this permanently.

Net: two of four closed, and the two that closed both landed on Bloom 1.6 — which is now the strongest room in the building and the natural AUTOMAIT pilot. The Bloom 1.3 dryback ask was not actioned and has compounded. The roster re-verification is a one-click fix on Insa's side that has now blocked two audits in a row.

3 · AUTOMAIT Readiness3 · Preparación AUTOMAIT

The meeting topic. This section answers three questions: is the data clean enough, are the rooms steerable, and what would AUTOMAIT actually change.

Ready

Bloom 1.3 · 1.6 · 3.1 · 3.5

Conditional

Bloom 1.4 · 3.3 · 3.6

Not yet

Bloom 2.1–2.6 — insufficient history

Blocked

Bloom 1.5 · 3.2 · 3.4 — sensor faults

Is the data clean enough?

For 7 of 16 rooms, yes. For 3, emphatically no. For 6, not yet — they simply have not been running long enough. The readiness test we applied is the one AUTOMAIT actually cares about: does the room have (a) enough continuous substrate history to baseline against, (b) a live climate feed to model transpiration from, (c) a probe whose readings are physically plausible, and (d) a dryback that repeats day over day within a few percentage points.

Room	Verdict	Substrate history	Climate history	Zones	Dryback SD	Zone spread	Basis
Bloom 1.3	READY	14 d	14 d	1	±0.8 pp	0.0 pp	14 d substrate + 14 d climate · dryback SD 0.8 pp (tightest on the floor) · probe clean
Bloom 1.4	CONDITIONAL	14 d	14 d	1	±0.8 pp	0.0 pp	14 d substrate + 14 d climate · dryback SD 0.8 pp · probe clean
Bloom 1.5	BLOCKED	14 d	14 d	1	—	0.0 pp	Pore EC pegged at 25 mS on the room's only substrate zone (81548)
Bloom 1.6	READY	14 d	14 d	1	±2.9 pp	0.0 pp	14 d substrate + 14 d climate · dryback SD 2.9 pp · drip + drain EC pair live (2.82 / 5.08)
Bloom 2.1	NOT YET	6 d	4 d	2	±0.4 pp	0.2 pp	6 d substrate / 4 d climate — online 2026-07-08
Bloom 2.2	NOT YET	6 d	4 d	2	±0.9 pp	5.3 pp	6 d substrate / 4 d climate — online 2026-07-08
Bloom 2.3	NOT YET	6 d	4 d	2	±1.4 pp	2.6 pp	6 d substrate / 4 d climate — online 2026-07-08
Bloom 2.4	NOT YET	6 d	4 d	2	±1.4 pp	1.5 pp	6 d substrate / 4 d climate — online 2026-07-08
Bloom 2.5	NOT YET	6 d	4 d	2	±0.5 pp	0.5 pp	6 d substrate / 4 d climate — online 2026-07-08
Bloom 2.6	NOT YET	11 d	4 d	3	±0.7 pp	6.9 pp	11 d substrate / 4 d climate — online 2026-07-02
Bloom 3.1	READY	14 d	14 d	3	±3.1 pp	3.1 pp	14 d substrate + 14 d climate · 3 zones · dryback SD 3.1 pp · zone spread 3.1 pp
Bloom 3.2	BLOCKED	14 d	4 d	1	—	0.0 pp	Only substrate zone pegged (EC 30 mS / VWC 0%) AND climate station dark since 2026-07-03
Bloom 3.3	CONDITIONAL	14 d	14 d	2	±8.5 pp	3.7 pp	14 d substrate + 14 d climate · but dryback SD 8.5 pp — the least repeatable room
Bloom 3.4	BLOCKED	14 d	none	1	—	0.0 pp	Only substrate zone collapsed (VWC 0.8%) AND zero climate samples in 14 days
Bloom 3.5	READY	14 d	14 d	3	±3.2 pp	5.6 pp	14 d substrate + 14 d climate · 3 zones · dryback SD 3.2 pp on a deep 31.6 pp program
Bloom 3.6	CONDITIONAL	14 d	14 d	2	±1.6 pp	7.8 pp	14 d substrate + 14 d climate · dryback SD 1.6 pp · but zone spread 7.8 pp

Read the "Dryback SD" column — it is the single best proxy for steerability. A room whose overnight dryback varies by ±0.8 pp (Bloom 1.3, 1.4) does the same thing every day and will respond predictably to a controller. A room that varies by ±8.5 pp (Bloom 3.3) does not have a program yet; it has a habit of watering. Automating an unstable program does not stabilise it — it scales it.

Are the rooms steerable?

Steerability is not just "does the probe work" — it is "does the room respond to the lever". Three signals say yes across most of the floor:

P3 depth is already being used as the primary lever. Dryback across the graded rooms runs from 15.2 pp (Bloom 3.6) to 31.6 pp (Bloom 3.5) — a 2× range that tracks inferred phase. Someone is deliberately steering these rooms differently, which means the levers work.
Substrate EC responds to that steering. Bloom 1.6 stacked 3.70 → 7.09 mS/cm and Bloom 3.5 restacked 4.50 → 7.04 mS/cm inside one fortnight. The root zone moves when you ask it to.
Zone agreement is tight in the best rooms. Bloom 3.1 holds 3 substrate zones within 3.1 pp of each other; Bloom 2.5 within 0.5 pp. Where the probes agree, the control signal is trustworthy.

The exceptions are documented: Bloom 3.6 zone spread 7.8 pp and Bloom 2.6 zone spread 6.9 pp — in those two rooms the probes disagree with each other by roughly half the room's total dryback, and AUTOMAIT would be steering on the average of a disagreement.

What would AUTOMAIT actually change?

Today (manual)	With AUTOMAIT	Evidence from this window
P0 delay is set by clock and drifts room to room	P0 ends when the plant starts transpiring, not when the timer says so	P0 delay ranges 65 min (Bloom 1.4) to 514 min (Bloom 2.1) across the floor. Bloom 1.3 and 1.4 fire their first shot ~70–80 min after lights-on — inside the veg window, not the generative one their EC implies.
P1 shot count is fixed; runoff is chased manually	P1 ramps to first runoff and stops, per room, per day	Shot counts range 1.2/day (Bloom 2.1) to 8.7/day (Bloom 3.5). Bloom 2.1's 1.2 shots/day on an 8.5 pp dryback is a room that is barely being irrigated at all.
P3 dry-down depth is a target that gets hit or missed	P3 is held to a dryback setpoint with closed-loop correction	Bloom 3.3 misses by ±8.5 pp day to day. Bloom 1.3 has been asked to deepen dryback for a full cycle and has instead gone from 17.7 → 16.1 pp.
Substrate EC drifts and is caught at the next audit	EC stack is a managed trajectory, not a discovered outcome	Bloom 1.4 bled 3.41 → 2.20 mS/cm unnoticed over 14 days while Bloom 1.3 climbed to 11.71. Both are out of band, in opposite directions, on the same wing.

The honest caveat. AUTOMAIT steers water. It does not fix feed. Bloom 1.4's EC bleed and Bloom 1.3's EC runaway are nutrition and P3-depth decisions that a human has to make; AUTOMAIT will then hold whatever you decide, precisely, every day. Automating before you have converged the EC intent across the floor would lock in the divergence rather than resolve it.

Recommended pilot

Start with Bloom 1.6 and Bloom 3.1. Bloom 1.6 is the only room with a working drip/drain EC pair (2.82 / 5.08 mS/cm), has 14 clean days of both substrate and climate, closed both of its prior-audit commitments, and now sits in-band on dryback (23.3 pp), P0 (172 min) and EC (3.70 → 7.09 mS/cm). Bloom 3.1 brings three substrate zones that agree within 3.1 pp — the most trustworthy control signal in the building. Between them they cover an early-flower program on two different wings, which is the right shape for a pilot. Bloom 3.5 is a strong third if Insa wants a deep-generative ripening room in the trial.

4 · HVAC Capacity4 · Capacidad HVAC

Dehumidification headroom is what sets achievable VPD, and VPD is what drives the transpiration that AUTOMAIT models. "Inverted" flags a room whose lights-off VPD exceeds its lights-on VPD — the plant is doing its transpiration work in the dark.

Room	Air on / off (°F)	Day–night swing	RH on / off	VPD on / off (kPa)	VPD swing	CO₂ on / off (ppm)	Substrate T on (°F)
Bloom 1.3	77.0 / 75.3	1.7	54% / 52%	1.45 / 1.44	0.01	1047 / 1133	74.1
Bloom 1.4	76.7 / 73.5	3.2	57% / 55%	1.34 / 1.26	0.08	971 / 677	74.4
Bloom 1.5	75.1 / 74.1	1.1	50% / 48%	1.49 / 1.51 inverted	-0.02	849 / 645	71.7
Bloom 1.6	77.4 / 76.0	1.4	59% / 56%	1.32 / 1.34 inverted	-0.02	999 / 956	75.9
Bloom 2.1	75.4 / 69.2	6.2	61% / 58%	1.18 / 1.01	0.17	368 / 370	73.5
Bloom 2.2	79.8 / 73.8	6.0	67% / 56%	1.15 / 1.25 inverted	-0.10	502 / 751	78.4
Bloom 2.3	75.0 / 70.1	4.9	55% / 54%	1.34 / 1.16	0.18	623 / 777	73.0
Bloom 2.4	77.1 / 71.3	5.8	59% / 51%	1.31 / 1.28	0.03	541 / 784	74.4
Bloom 2.5	77.1 / 70.1	7.0	68% / 59%	1.00 / 1.03 inverted	-0.03	531 / 1155	73.0
Bloom 2.6	75.8 / 75.1	0.7	52% / 51%	1.47 / 1.45	0.02	935 / 1245	73.7
Bloom 3.1	80.1 / 79.5	0.7	58% / 56%	1.48 / 1.50 inverted	-0.02	1055 / 1030	78.3
Bloom 3.2	76.6 / 76.0	0.6	49% / 49%	1.58 / 1.56	0.02	1695 / 1702	76.2
Bloom 3.3	81.2 / 79.7	1.6	53% / 55%	1.73 / 1.58	0.15	1735 / 1715	78.7
Bloom 3.4	NO CLIMATE TELEMETRY — 0 air-temp samples in 14 days
Bloom 3.5	78.3 / 77.4	1.0	61% / 59%	1.28 / 1.33 inverted	-0.05	1475 / 1602	77.0
Bloom 3.6	78.5 / 78.0	0.6	60% / 63%	1.34 / 1.22	0.12	1479 / 1437	77.8

What the table says. Insa's HVAC is, on the whole, capable: lights-on VPD clears 1.28 kPa in every fully-instrumented Bloom 3 room and RH is held in the 50–62% band. The three things worth calling out:

Bloom 3.3 is running the hottest air in the building — lights-on VPD 1.73 kPa against a 1.2–1.4 bulking target. That is over-driving transpiration and it shows up directly as the ±8.5 pp dryback instability.
Day–night swing is very small in Bloom 3 (0.6–1.6 °F). A near-flat diurnal profile means the photoperiod has to be inferred from a shallow temperature ramp, which is why P0 delay could not be resolved reliably in Bloom 3.2 / 3.3 / 3.4.
The Bloom 2 wing has no CO₂ enrichment reaching it — 368–935 ppm lights-on against 1000–1735 ppm in Bloom 3. Bloom 2.1 at 368 ppm is ambient air.

5 · Climate ↔ Irrigation Interplay5 · Interacción Clima–Riego

The air sets the demand; the irrigation program answers it. ET rate (the P3 dry-down slope) is the clearest measurable link between the two. Substrate-blind rooms are excluded.

Room	VPD lights-on (kPa)	ET rate (pp VWC/h)	Dryback (absolute pp)	Shots / day	Substrate EC (mS/cm)	Substrate T (°F)
Bloom 1.3	1.45	0.36	16.1 pp	5.2	10.01	74.1
Bloom 1.4	1.34	0.56	16.9 pp	4.5	2.72	74.4
Bloom 1.6	1.32	0.71	23.3 pp	6.4	5.57	75.9
Bloom 2.1	1.18	0.08	8.5 pp	1.2	7.77	73.5
Bloom 2.2	1.15	0.31	10.6 pp	4.7	7.02	78.4
Bloom 2.3	1.34	0.27	14.0 pp	2.0	6.14	73.0
Bloom 2.4	1.31	0.43	11.6 pp	4.2	6.61	74.4
Bloom 2.5	1.00	0.66	18.4 pp	5.0	7.61	73.0
Bloom 2.6	1.47	0.44	17.5 pp	4.0	6.46	73.7
Bloom 3.1	1.48	0.63	19.6 pp	6.2	6.34	78.3
Bloom 3.3	1.73	1.07	27.9 pp	5.1	3.33	78.7
Bloom 3.5	1.28	1.38	31.6 pp	8.7	5.32	77.0
Bloom 3.6	1.34	0.34	15.2 pp	3.3	6.73	77.8

The coupling is real and it is visible. The two rooms with the highest lights-on VPD outside Bloom 3.3 — Bloom 3.5 (1.28 kPa) and Bloom 3.1 (1.48 kPa) — are also the two with the highest ET rates (1.38 and 0.63 pp/h) and the deepest drybacks (31.6 and 19.6 pp). That is the system working as designed: drier air pulls more water, the bag dries back further, and the EC stack concentrates.

The counter-example is Bloom 2.1: VPD 1.18 kPa but an ET rate of just 0.08 pp/h and a dryback of 8.5 pp on 1.2 shots/day. Normal air, no transpiration. Per the AROYA diagnostic hooks, "no additional dryback with adequate VPD" points at the plant, not the irrigation volume — a crop that has not woken up, or a saturated root zone. Worth a physical look.

6 · Per-Room Deep Dive6 · Análisis por Sala

Select a room to drive every chart in sections 7 and 8. Turn on Compare mode to overlay several rooms at once. Every phase label is signal-inferred — the AROYA harvest-group roster was unavailable this run.

Bloom 1.3 AUTOMAIT: READY

Bloom 1 · 1 substrate zone(s) · signal-inferred: late flower / ripening
Basis: substrate EC stacked 9.3 → 11.7 mS · dryback 16.1 pp · VPD 1.45 kPa · 12 h photoperiod

Dryback

16.1 pp

±0.8 pp · absolute pp

VWC peak → trough

65.5 → 49.5

Shots / day

5.2

derived

P0 delay

80 min

lights-on → first shot

Substrate EC

10.01

9.31 → 11.71 mS/cm

VPD on / off

1.45 / 1.44

kPa

Air on / off

77.0 / 75.3

°F

RH lights-on

54.2%

photoperiod 12 h

ET rate

0.36

pp VWC / h (P3 slope)

Zone spread

0.0 pp

1 substrate zone(s)

Steering intent: signal-inferred late flower / ripening — and the substrate EC has run away from the band. Bloom 1.3 carries a substrate EC of 10.0 mS/cm mean, stacking 9.3 → 11.7 across the fortnight and touching 12.67. The ripening restack target is 6.0–8.0 mS/cm. This room is 45–95% above it and still climbing. Meanwhile overnight dryback is only 16.1 absolute pp against a 25–35 pp ripening band, on a 80-minute P0 delay and 5.2 shots/day.

Those two facts fit together and they are not good: the room is being watered often enough and shallow enough that the salts never leave, but not deeply enough to drive the uptake that would consume them. Substrate EC climbing without a matching dryback is the classic "stacking too aggressively for the phase" signature. The fix is P3 depth, not feed EC — lengthen the dry-down, drop the shot count, and let the plant pull the stack down.

The redeeming number is dryback SD of 0.8 pp — the tightest on the floor. Whatever this room is doing, it does it the same way every single day. That repeatability is exactly what makes it a credible AUTOMAIT room once the EC question is settled.

Bloom 1.4 AUTOMAIT: CONDITIONAL

Bloom 1 · 1 substrate zone(s) · signal-inferred: bulking
Basis: substrate EC 3.4 → 2.2 mS (falling) · dryback 17.0 pp · VWC held 55–72% · 13 h photoperiod

Dryback

16.9 pp

±0.8 pp · absolute pp

VWC peak → trough

71.8 → 54.9

Shots / day

4.5

derived

P0 delay

65 min

lights-on → first shot

Substrate EC

2.72

3.41 → 2.20 mS/cm

VPD on / off

1.34 / 1.26

kPa

Air on / off

76.7 / 73.5

°F

RH lights-on

57.2%

photoperiod 13 h

ET rate

0.56

pp VWC / h (P3 slope)

Zone spread

0.0 pp

1 substrate zone(s)

Steering intent: signal-inferred bulking — but the substrate EC is bleeding out. Substrate EC fell 3.41 → 2.20 mS/cm over the window (mean 2.72, minimum 0.99). The bulking band is 4.0–7.0 mS/cm. This room is running at roughly half the bottom of its band and the trend is still downward.

The mechanism is visible in the water: VWC sits at a mean of 65.3% — the wettest room in the building — with peaks of 71.8 and troughs of 54.9. Dryback is 17.0 absolute pp, nominally inside the 12–18 pp bulking band, but it is being delivered from a very high field capacity, so the bag is never actually allowed to concentrate. Wet bag + frequent shots = the stack washes out. That is the same bleed mechanism, in reverse, that Bloom 1.3 is suffering from.

Note the contrast that should frame the whole AUTOMAIT conversation: Bloom 1.3 and Bloom 1.4 sit side by side, on the same wing, and their substrate EC differs by 3.7× (10.0 vs 2.7 mS/cm). No automation can reconcile that — it is a feed and P3 decision, and it has to be made by a human first.

Bloom 1.5 AUTOMAIT: BLOCKED

Bloom 1 · 1 substrate zone(s) · substrate-blind — phase not inferable
Basis: the room's only substrate zone is faulted (pore EC pegged to 25 mS); climate is clean

Dryback

n/a

probe faulted

VWC peak → trough

n/a

Shots / day

n/a

derived

P0 delay

n/a

lights-on → first shot

Substrate EC

faulted

channel pegged

VPD on / off

1.49 / 1.51

kPa

Air on / off

75.1 / 74.1

°F

RH lights-on

49.8%

photoperiod 12 h

ET rate

n/a

pp VWC / h (P3 slope)

Zone spread

0.0 pp

1 substrate zone(s)

Substrate-blind. The room's only substrate zone (81548) is faulted. Pore EC on zone 81548 reads a mean of 12.4 mS/cm and pegs at 25.0 — 25.0 is the channel ceiling, not a soil reading — while VWC swings from 0.2% to 69.5% and the derived dryback carries a standard deviation of 16.7 pp, which is not a crop signal, it is noise.

Everything downstream of that probe is unusable: VWC, pore EC, dryback, shot count, ET. We are not reporting an agronomic verdict for this room, because we do not have one. The climate suite (air temp, RH, VPD, CO₂) is clean and complete for all 14 days — VPD 1.49/1.51 kPa lights-on/off, RH ~50%. The room is being managed; we just cannot see the root zone.

This is a 15-minute fix that unlocks a whole room. Re-seat or replace the probe on zone 81548.

Bloom 1.6 AUTOMAIT: READY

Bloom 1 · 1 substrate zone(s) · signal-inferred: early flower / stretch
Basis: substrate EC stacking 3.7 → 7.1 mS · dryback 23.3 pp · P0 delay 172 min · drip/drain pair live

Dryback

23.3 pp

±2.9 pp · absolute pp

VWC peak → trough

54.2 → 30.9

Shots / day

6.4

derived

P0 delay

172 min

lights-on → first shot

Substrate EC

5.57

3.70 → 7.09 mS/cm

VPD on / off

1.32 / 1.34

kPa

Air on / off

77.4 / 76.0

°F

RH lights-on

58.8%

photoperiod 17 h

ET rate

0.71

pp VWC / h (P3 slope)

Zone spread

0.0 pp

1 substrate zone(s)

Steering intent: signal-inferred early flower / stretch — and the data agrees. This is the best-run room in the building. Substrate EC stacked 3.70 → 7.09 mS/cm across the fortnight — a real, sustained, monotone stack into the 6.0–12.0 early-flower band. Overnight dryback averaged 23.3 absolute pp (54.2 peak → 30.9 trough), squarely inside the 20–30 pp early-flower target, with an SD of 2.9 pp. The P0 delay is 172 minutes — inside the 120–240 min generative window. Shot count 6.4/day. Every one of those numbers is where the phase says it should be.

It is also the only room on the floor with a working drip/drain EC pair: drip-position 2.82 vs drain-position 5.08 mS/cm. Drip below drain is the normal direction (the probes are not reversed), and a 2.3 mS/cm gradient across the bag says the stack is being held rather than flushed on every P2 pulse.

Full 14-day substrate and climate, a repeatable P0/P1/P2/P3 rhythm, a live drip/drain pair, and an EC trajectory that matches declared intent. If Insa runs one AUTOMAIT pilot room, this is the one to pick.

Bloom 2.1 AUTOMAIT: NOT YET

Bloom 2 · 2 substrate zone(s) · new run — substrate live 07-08
Basis: 6 d substrate / 4 d climate · EC 7.8 mS · dryback 8.5 pp on 1.2 shots/day

Dryback

8.5 pp

±0.4 pp · absolute pp

VWC peak → trough

46.7 → 38.2

Shots / day

1.2

derived

P0 delay

514 min

lights-on → first shot

Substrate EC

7.77

7.44 → 8.10 mS/cm

VPD on / off

1.18 / 1.01

kPa

Air on / off

75.4 / 69.2

°F

RH lights-on

60.5%

photoperiod 20 h

ET rate

0.08

pp VWC / h (P3 slope)

Zone spread

0.2 pp

2 substrate zone(s)

New run — substrate telemetry began 2026-07-08, climate 2026-07-10. Six days of root zone, four of air. We are not grading this room against a 14-day standard, because it does not have 14 days.

What the six days do show: substrate EC is 7.77 mS/cm — the second-highest on the floor — and dryback is only 8.5 absolute pp on 1.2 shots/day. A near-static bag at high EC. The ET rate of 0.08 pp/h is essentially flat: the plants are not transpiring meaningfully. In a room this early that usually means the crop has not woken up yet, or the root zone is saturated. CO₂ reads 368 ppm lights-on — that is ambient air, i.e. no CO₂ supplementation is reaching this room, against 1000–1700 ppm in the Bloom 3 wing.

Watch it. Do not automate it yet.

Bloom 2.2 AUTOMAIT: NOT YET

Bloom 2 · 2 substrate zone(s) · new run — substrate live 07-08
Basis: 6 d substrate / 4 d climate · EC 7.0 mS · dryback 10.6 pp · P0 198 min

Dryback

10.6 pp

±0.9 pp · absolute pp

VWC peak → trough

45.0 → 34.5

Shots / day

4.7

derived

P0 delay

198 min

lights-on → first shot

Substrate EC

7.02

6.94 → 7.08 mS/cm

VPD on / off

1.15 / 1.25

kPa

Air on / off

79.8 / 73.8

°F

RH lights-on

67.0%

photoperiod 12 h

ET rate

0.31

pp VWC / h (P3 slope)

Zone spread

5.3 pp

2 substrate zone(s)

New run — substrate live 2026-07-08 (6 d), climate live 2026-07-10 (4 d). Substrate EC 7.02 mS/cm, flat (6.94 → 7.08). Dryback 10.6 absolute pp on 4.7 shots/day with a P0 delay of 198 min. Lights-on VPD 1.15 kPa on 67% RH — the humidest room in the Bloom 2 wing, and the lights-off VPD (1.25) is higher than the lights-on VPD, meaning the plant is doing its transpiration work in the dark. Zone spread 5.3 pp between the two substrate zones.

Directionally reasonable for an early room. Not enough history to baseline, and the inverted day/night VPD profile is worth a look before it becomes the room's habit.

Bloom 2.3 AUTOMAIT: NOT YET

Bloom 2 · 2 substrate zone(s) · new run — substrate live 07-08
Basis: 6 d substrate / 4 d climate · EC 6.1 mS · dryback 14.0 pp · VPD 1.34 kPa

Dryback

14.0 pp

±1.4 pp · absolute pp

VWC peak → trough

56.5 → 42.5

Shots / day

2.0

derived

P0 delay

144 min

lights-on → first shot

Substrate EC

6.14

6.10 → 6.17 mS/cm

VPD on / off

1.34 / 1.16

kPa

Air on / off

75.0 / 70.1

°F

RH lights-on

54.7%

photoperiod 12 h

ET rate

0.27

pp VWC / h (P3 slope)

Zone spread

2.6 pp

2 substrate zone(s)

New run — substrate live 2026-07-08 (6 d), climate live 2026-07-10 (4 d). The tidiest of the Bloom 2 wing. Substrate EC 6.14 mS/cm and stable, dryback 14.0 absolute pp, P0 delay 144 min, lights-on VPD 1.34 kPa on 55% RH, zone spread 2.6 pp. Only 2.0 shots/day, which is low — the dryback is being delivered by a long P3 rather than by an active P1/P2 ramp.

Nothing here is alarming; it simply has no baseline yet. Re-assess after 14 continuous days.

Bloom 2.4 AUTOMAIT: NOT YET

Bloom 2 · 2 substrate zone(s) · new run — substrate live 07-08
Basis: 6 d substrate / 4 d climate · EC stacking 6.1 → 7.2 mS · dryback 11.6 pp

Dryback

11.6 pp

±1.4 pp · absolute pp

VWC peak → trough

49.7 → 38.1

Shots / day

4.2

derived

P0 delay

320 min

lights-on → first shot

Substrate EC

6.61

6.06 → 7.15 mS/cm

VPD on / off

1.31 / 1.28

kPa

Air on / off

77.1 / 71.3

°F

RH lights-on

58.7%

photoperiod 12 h

ET rate

0.43

pp VWC / h (P3 slope)

Zone spread

1.5 pp

2 substrate zone(s)

New run — substrate live 2026-07-08 (6 d), climate live 2026-07-10 (4 d). Substrate EC is stacking 6.06 → 7.15 mS/cm — the clearest deliberate stack in the Bloom 2 wing — on a dryback of 11.6 absolute pp and 4.3 shots/day. Zone spread is a tight 1.5 pp. Lights-on VPD 1.31 kPa.

This is the Bloom 2 room whose signal most resembles a room being steered on purpose. It still needs 14 days before it can be graded, or automated.

Bloom 2.5 AUTOMAIT: NOT YET

Bloom 2 · 2 substrate zone(s) · new run — substrate live 07-08
Basis: 6 d substrate / 4 d climate · EC 7.6 mS (highest on the floor bar 1.3) · dryback 18.4 pp

Dryback

18.4 pp

±0.5 pp · absolute pp

VWC peak → trough

51.0 → 32.6

Shots / day

5.0

derived

P0 delay

140 min

lights-on → first shot

Substrate EC

7.61

7.46 → 7.74 mS/cm

VPD on / off

1.00 / 1.03

kPa

Air on / off

77.1 / 70.1

°F

RH lights-on

68.4%

photoperiod 11 h

ET rate

0.66

pp VWC / h (P3 slope)

Zone spread

0.5 pp

2 substrate zone(s)

New run — substrate live 2026-07-08 (6 d), climate live 2026-07-10 (4 d). Substrate EC 7.61 mS/cm and rising (7.46 → 7.74) — the highest in the Bloom 2 wing. Dryback 18.4 absolute pp on 5.0 shots/day, P0 140 min, ET 0.66 pp/h, zone spread 0.5 pp (the tightest zone agreement in the building).

Lights-on VPD is 1.00 kPa on 68% RH — the softest air in the wing — while lights-off CO₂ climbs to 1155 ppm. The room is transpiring well despite the low VPD. Good bones. Needs history.

Bloom 2.6 AUTOMAIT: NOT YET

Bloom 2 · 3 substrate zone(s) · new run — substrate live 07-02
Basis: 11 d substrate / 4 d climate · EC 6.1 → 6.7 mS · dryback 17.5 pp · VPD 1.47 kPa

Dryback

17.5 pp

±0.7 pp · absolute pp

VWC peak → trough

58.0 → 40.6

Shots / day

4.0

derived

P0 delay

465 min

lights-on → first shot

Substrate EC

6.46

6.13 → 6.72 mS/cm

VPD on / off

1.47 / 1.45

kPa

Air on / off

75.8 / 75.1

°F

RH lights-on

51.9%

photoperiod 14 h

ET rate

0.44

pp VWC / h (P3 slope)

Zone spread

6.9 pp

3 substrate zone(s)

New run — substrate live 2026-07-02 (11 d), climate live 2026-07-10 (4 d). The most mature of the Bloom 2 wing. Substrate EC 6.13 → 6.72 mS/cm, dryback 17.5 absolute pp with an SD of 0.7 pp — a very repeatable program — on 4.0 shots/day. Lights-on VPD 1.47 kPa, the highest in the wing, on 52% RH.

The blocker is not the root zone; it is the air. Climate history is only four days because the Bloom 2 climate stations came online 2026-07-10. Zone spread is 6.9 pp across three substrate zones — the second-widest on the floor, worth checking.

This room will be AUTOMAIT-ready first out of the Bloom 2 wing, roughly two weeks from now.

Bloom 3.1 AUTOMAIT: READY

Bloom 3 · 3 substrate zone(s) · signal-inferred: early flower / stretch
Basis: substrate EC stacking 5.5 → 6.7 mS · dryback 19.6 pp · 6.2 shots/day · P0 183 min

Dryback

19.6 pp

±3.1 pp · absolute pp

VWC peak → trough

54.0 → 34.5

Shots / day

6.2

derived

P0 delay

183 min

lights-on → first shot

Substrate EC

6.34

5.54 → 6.68 mS/cm

VPD on / off

1.48 / 1.50

kPa

Air on / off

80.1 / 79.5

°F

RH lights-on

57.7%

photoperiod 12 h

ET rate

0.63

pp VWC / h (P3 slope)

Zone spread

3.1 pp

3 substrate zone(s)

Steering intent: signal-inferred early flower / stretch — and the data agrees. The strongest AUTOMAIT candidate in the building. Substrate EC stacked 5.54 → 6.68 mS/cm into the 6.0–12.0 early-flower band. Overnight dryback averaged 19.6 absolute pp (54.0 peak → 34.5 trough) — the bottom edge of the 20–30 pp early-flower window — on 6.2 shots/day with a P0 delay of 183 min, right in the 120–240 min generative target.

What sets it apart is instrumentation quality: three substrate zones, 14 full days of both substrate and climate, a dryback SD of 3.1 pp and a zone-to-zone dryback spread of only 3.1 pp. The three probes agree with each other and the room agrees with itself day over day. That is precisely the profile an autonomous irrigation controller needs — a control signal it can trust and a room that responds predictably to it.

The one adjustment: dryback wants to go a little deeper (19.6 → 22–25 pp) to sit mid-band rather than on the edge. Lights-on VPD 1.48 kPa is above the 1.0–1.2 early-flower band, which is doing some of that dryback work for you via the air.

Bloom 3.2 AUTOMAIT: BLOCKED

Bloom 3 · 1 substrate zone(s) · substrate-blind + climate-dark — phase not inferable
Basis: only substrate zone pegged (EC 30 mS, VWC 0%); climate station silent since 2026-07-03

Dryback

n/a

probe faulted

VWC peak → trough

n/a

Shots / day

n/a

derived

P0 delay

n/a

lights-on → first shot

Substrate EC

faulted

channel pegged

VPD on / off

1.58 / 1.56

kPa

Air on / off

76.6 / 76.0

°F

RH lights-on

49.3%

photoperiod 6 h

ET rate

n/a

pp VWC / h (P3 slope)

Zone spread

0.0 pp

1 substrate zone(s)

Substrate-blind AND climate-dark. Two independent failures in one room. The room's only substrate zone (84065) reads a pore EC mean of 20.6 mS/cm, pegged at the 30.0 channel ceiling, while its VWC collapses to 0.0% — a probe that is simultaneously reporting "bone dry" and "saturated with salt" is a probe that is out of media or failed, not a root zone.

Separately, the climate station stopped reporting on 2026-07-03. We have 488 air-temperature samples where a healthy room in this window returns ~2,240 — 11 of the last 14 days have no air temp, RH, VPD or CO₂ at all.

No agronomic verdict is offered for this room, because there is no usable data to base one on. Note that the mesh radio serving this room is healthy (battery 3.6 V, strong link) — this is a sensor problem, not a network problem. Both faults are covered in the companion device-health report.

Bloom 3.3 AUTOMAIT: CONDITIONAL

Bloom 3 · 2 substrate zone(s) · signal-inferred: bulking (erratic)
Basis: substrate EC 2.7 → 3.9 mS (below band) · dryback 27.9 pp but SD 8.5 pp · VPD 1.73 kPa

Dryback

27.9 pp

±8.5 pp · absolute pp

VWC peak → trough

68.5 → 40.5

Shots / day

5.1

derived

P0 delay

-295 min

lights-on → first shot

Substrate EC

3.33

2.71 → 3.91 mS/cm

VPD on / off

1.73 / 1.58

kPa

Air on / off

81.2 / 79.7

°F

RH lights-on

53.4%

photoperiod 4 h

ET rate

1.07

pp VWC / h (P3 slope)

Zone spread

3.7 pp

2 substrate zone(s)

Steering intent: signal-inferred bulking — but this room does not repeat itself, and that is the whole problem. Overnight dryback averaged 27.9 absolute pp — well above the 12–18 pp bulking band — but with a standard deviation of 8.5 pp, by far the least repeatable room on the floor. Some days it pulls 20 pp, some days it pulls 36. Substrate EC is 3.33 mS/cm (2.71 → 3.91), below the 4.0–7.0 bulking band.

Lights-on VPD is 1.73 kPa, the highest in the facility and well above the 1.2–1.4 bulking target, on lights-off CO₂ of 1715 ppm. The air is pulling water out of these bags hard and inconsistently, and the substrate EC is too dilute to buffer it. VWC swings 29.2 → 70.2%.

This is the single clearest example of a room that AUTOMAIT would make worse, not better. An autonomous controller optimises against a signal; if the signal is unstable it will chase it. Bring the VPD down toward 1.2–1.4, get the EC into band, and get the dryback SD under ~3 pp before anyone talks about automating it.

Bloom 3.4 AUTOMAIT: BLOCKED

Bloom 3 · 1 substrate zone(s) · substrate-blind + no climate — phase not inferable
Basis: only substrate zone collapsed (VWC 0.8%); ZERO air-temp/RH/VPD/CO₂ samples in 14 days

Dryback

n/a

probe faulted

VWC peak → trough

n/a

Shots / day

n/a

derived

P0 delay

n/a

lights-on → first shot

Substrate EC

faulted

channel pegged

VPD on / off

— / —

kPa

Air on / off

— / —

°F

RH lights-on

—

photoperiod 0 h

ET rate

n/a

pp VWC / h (P3 slope)

Zone spread

0.0 pp

1 substrate zone(s)

Effectively un-instrumented. This room has no climate telemetry at all and its only substrate probe has failed. Across the full 14-day window Bloom 3.4 returned zero air-temperature, RH, VPD and CO₂ samples — not a gap, not a dropout: none. And its single substrate zone (84071) shows VWC collapsing to 0.8% against a mean of 18.7%, with pore EC running a mean of 11.2 and topping out at the 15.0 ceiling.

There is nothing to grade here and nothing to automate against. The room is presumably in production — but as far as AROYA is concerned it is dark. Of everything in this report, this is the item with the highest ratio of impact to effort.

The mesh radios in this room are healthy and reporting. The fault is at the sensor layer.

Bloom 3.5 AUTOMAIT: READY

Bloom 3 · 3 substrate zone(s) · signal-inferred: ripening
Basis: substrate EC restacking 4.5 → 7.0 mS · dryback 31.6 pp · 8.7 shots/day · ET 1.38 pp/h

Dryback

31.6 pp

±3.2 pp · absolute pp

VWC peak → trough

66.3 → 34.8

Shots / day

8.7

derived

P0 delay

183 min

lights-on → first shot

Substrate EC

5.32

4.50 → 7.04 mS/cm

VPD on / off

1.28 / 1.33

kPa

Air on / off

78.3 / 77.4

°F

RH lights-on

61.4%

photoperiod 13 h

ET rate

1.38

pp VWC / h (P3 slope)

Zone spread

5.6 pp

3 substrate zone(s)

Steering intent: signal-inferred ripening — and the data agrees, emphatically. The most aggressive, most controlled program in the building. Overnight dryback averaged 31.6 absolute pp (66.3 peak → 34.8 trough) — inside the 25–35 pp ripening band — sustained on 8.7 shots/day, the highest shot count on the floor, with an ET rate of 1.38 pp VWC/hour. That combination (many shots, very deep dry-down) is what a deliberate ripening P3 looks like: keep the bag from crashing, then pull it hard overnight.

Substrate EC restacked 4.50 → 7.04 mS/cm — straight into the 6.0–8.0 ripening restack band. Unlike Bloom 1.3, this room built its stack and earned it with dryback. Lights-on VPD 1.28 kPa is slightly under the 1.4–1.6 ripening target — the one number that could be pushed.

Three substrate zones, 14 full days, dryback SD 3.2 pp on a 31.6 pp program. Deep, aggressive, and repeatable. A very strong AUTOMAIT room.

Bloom 3.6 AUTOMAIT: CONDITIONAL

Bloom 3 · 2 substrate zone(s) · signal-inferred: bulking
Basis: substrate EC stacking 5.5 → 7.9 mS · dryback 15.2 pp · zone spread 7.8 pp (widest on floor)

Dryback

15.2 pp

±1.6 pp · absolute pp

VWC peak → trough

48.1 → 32.9

Shots / day

3.3

derived

P0 delay

486 min

lights-on → first shot

Substrate EC

6.73

5.51 → 7.89 mS/cm

VPD on / off

1.34 / 1.22

kPa

Air on / off

78.5 / 78.0

°F

RH lights-on

59.9%

photoperiod 12 h

ET rate

0.34

pp VWC / h (P3 slope)

Zone spread

7.8 pp

2 substrate zone(s)

Steering intent: signal-inferred bulking — the room-level numbers are good, the zone-level numbers disagree. Substrate EC stacked 5.51 → 7.89 mS/cm, top of the 4.0–7.0 bulking band and pushing past it. Dryback averaged 15.2 absolute pp — mid-band for bulking (12–18 pp) — with a tight SD of 1.6 pp, on 3.3 shots/day. Lights-on VPD 1.34 kPa, inside the 1.2–1.4 bulking target. On paper this is a well-behaved room.

The catch is uniformity: zone-to-zone dryback spread is 7.8 pp, the widest of any room in the facility. The two substrate zones are not experiencing the same irrigation. That is either genuinely uneven emitter delivery across the room or a probe in a bag that is not properly on-line — and it matters disproportionately for automation, because AUTOMAIT would steer on the average of two probes that disagree by half the room's total dryback.

Ten minutes with a hand meter resolves it. Do that before this room goes on the automation list.

7 · Phase Biophysics7 · Biofísica de Fase

Daily aggregates for the selected room(s). Solid lines = lights-on; dashed = lights-off.

Air temperature — lights-on vs lights-off (°F)

Solid = lights-on mean · dashed = lights-off mean

VPD — lights-on vs lights-off (kPa)

Target bands: veg 0.8–1.1 · early flower 1.0–1.2 · bulking 1.2–1.4 · ripening 1.4–1.6

Relative humidity — lights-on vs lights-off (%)

Dehumidification headroom is what sets achievable VPD

CO₂ — lights-on (ppm)

Bloom 3 runs 1000–1700 ppm supplemented; the Bloom 2 wing reads 370–930 ppm

Substrate VWC — daily peak vs pre-irrigation trough (%)

Solid = peak (field capacity reached) · dashed = pre-irrigation trough

Substrate temperature — lights-on vs lights-off (°F)

Root-zone temp drives uptake; below 68 °F suppresses it

8 · Irrigation Shot-Cycle8 · Ciclo de Riego

The four-phase day. P0 activation (lights-on → first shot) · P1 ramp-up to first runoff · P2 maintenance pulses at field capacity · P3 dry-down, the primary steering lever. Dryback is reported in absolute pp (daily peak VWC minus the following pre-irrigation low) throughout.

Overnight dryback (absolute pp)

Peak VWC − pre-irrigation trough, in absolute percentage points. Targets: veg 5–15 · early flower 20–30 · bulking 12–18 · ripening 25–35

Shot count per day (derived)

Derived from VWC step-up signatures — see the derivation footnote in Data Quality

P0 delay — lights-on → first shot (minutes)

Early-flower generative target 120–240 min · veg 60–120 min. Unreliable where the photoperiod could not be resolved (Bloom 3.2 / 3.3 / 3.4)

Substrate EC (mS/cm) and day-over-day EC stack delta

Solid = substrate (pore) EC · dashed = day-over-day change. Substrate EC is the steering output, not the feed setting

Drip-position vs drain-position EC (mS/cm)

Only Bloom 1.6 has a live drip/drain pair. Drip < drain is NORMAL. Drip > drain sustained = probes reversed

ET rate — P3 dry-down slope (pp VWC / hour)

Measured from the no-shot decay after the last shot of the day

Phase-target reference. Root-in: dryback 10–15 pp · shot 1–3% · substrate EC ~3 mS/cm · VPD ~0.8 kPa. Veg: dryback 5–15 pp · shot 3–5% · EC 3.0–5.0 · VPD 0.8–1.1. Early flower / stretch: dryback 20–30 pp · shot 6–10% · P0 delay 120–240 min · EC stacking 6.0–12.0 · VPD 1.0–1.2. Bulking: dryback 12–18 pp · shot 3–5% · EC pulled back to 4.0–7.0 · VPD 1.2–1.4. Ripening: dryback 25–35 pp · shot 6–10% · EC restacked to 6.0–8.0 · VPD 1.4–1.6.

9 · Configuration9 · Configuración

Room	Room ID	Wing	Photoperiod (inferred)	Substrate zones	Substrate samples	Climate samples	Phase (signal-inferred)
Bloom 1.3	19823	Bloom 1	12 h	1	2,240	2,240	signal-inferred: late flower / ripening
Bloom 1.4	19824	Bloom 1	13 h	1	2,240	2,240	signal-inferred: bulking
Bloom 1.5	19825	Bloom 1	12 h	1	2,240	2,240	substrate-blind — phase not inferable
Bloom 1.6	19826	Bloom 1	17 h	1	2,240	2,240	signal-inferred: early flower / stretch
Bloom 2.1	20375	Bloom 2	20 h	2	938	605	new run — substrate live 07-08
Bloom 2.2	20376	Bloom 2	12 h	2	938	605	new run — substrate live 07-08
Bloom 2.3	20377	Bloom 2	12 h	2	938	605	new run — substrate live 07-08
Bloom 2.4	20378	Bloom 2	12 h	2	937	605	new run — substrate live 07-08
Bloom 2.5	20379	Bloom 2	11 h	2	937	605	new run — substrate live 07-08
Bloom 2.6	20380	Bloom 2	14 h	3	1,875	605	new run — substrate live 07-02
Bloom 3.1	20311	Bloom 3	12 h	3	2,207	2,198	signal-inferred: early flower / stretch
Bloom 3.2	20312	Bloom 3	6 h	1	2,208	486	substrate-blind + climate-dark — phase not inferable
Bloom 3.3	20313	Bloom 3	4 h	2	2,240	2,240	signal-inferred: bulking (erratic)
Bloom 3.4	20316	Bloom 3	0 h	1	2,205	0	substrate-blind + no climate — phase not inferable
Bloom 3.5	20317	Bloom 3	13 h	3	2,207	2,199	signal-inferred: ripening
Bloom 3.6	20318	Bloom 3	12 h	2	2,206	2,197	signal-inferred: bulking

Photoperiod is inferred from each room's diurnal air-temperature ramp, not read from a lighting schedule. Where the day–night temperature swing is very small (Bloom 3.2 at 0.6 °F, Bloom 3.3 at 1.6 °F) or absent entirely (Bloom 3.4, no climate), that inference is unreliable and the derived P0 delay for those rooms should not be trusted. A harvest-group roster pull would replace all of this with ground truth.

10 · Recommendations10 · Recomendaciones

Prioritised. Owner = facility lead · Grower = room operator · Process = AROYA / Addium follow-up.

Priority	Owner	Action	Why
P0	Owner	Replace or re-seat three substrate probes: Bloom 1.5 (zone 81548), Bloom 3.2 (zone 84065), Bloom 3.4 (zone 84071).	Each of these is the room's only substrate zone, and each is faulted — two pegged at the pore-EC channel ceiling (25.0 and 30.0 mS/cm), one with VWC collapsed to 0.8%. These three rooms are agronomically invisible today and permanently AUTOMAIT-blocked until fixed. The mesh radios serving them are healthy, so this is a probe swap, not a network job.
P0	Owner	Restore the Bloom 3.4 and Bloom 3.2 climate stations.	Bloom 3.4 returned zero air-temp / RH / VPD / CO₂ samples across the full 14-day window. Bloom 3.2 has been silent since 2026-07-03. Without a VPD input there is no transpiration model — for the grower or for AUTOMAIT.
P1	Grower	Bloom 1.3 — deepen P3 and stop the EC runaway. Target dryback 25–35 pp (currently 16.1 pp) and hold the stack at 6.0–8.0 mS/cm (currently 10.0 and climbing).	This was asked for in the 2026-06-26 audit and dryback has since moved the wrong way (17.7 → 16.1 pp). Substrate EC climbing without a matching dryback is textbook over-stacking for the phase: the salts are going in and nothing is pulling them through. Lengthen the dry-down and cut the shot count before adding any feed change.
P1	Grower	Bloom 1.4 — stop the substrate EC bleed. Currently 2.72 mS/cm (3.41 → 2.20) against a 4.0–7.0 bulking band.	The room is the wettest in the building (mean VWC 65.3%) and the stack is washing out. This is the mirror image of Bloom 1.3, on the same wing — and the 3.7× EC spread between two neighbouring rooms is the strongest argument that the floor needs a converged steering intent before it needs automation.
P1	Grower	Bloom 3.3 — bring lights-on VPD down from 1.73 to 1.2–1.4 kPa and stabilise the program.	Highest VPD in the facility and the least repeatable dryback (SD ±8.5 pp). The over-driven air is destabilising the dry-down. Substrate EC is also below band at 3.33 mS/cm. Stabilise here before considering automation — AUTOMAIT would chase an unstable signal.
P2	Owner	Pilot AUTOMAIT on Bloom 1.6 and Bloom 3.1.	Both have 14 clean days of substrate and climate, in-band dryback and EC, and repeatable day-over-day behaviour. Bloom 1.6 additionally has the facility's only working drip/drain EC pair. Bloom 3.5 is a strong third if a deep-generative ripening room is wanted in the trial.
P2	Grower	Resolve zone disagreement in Bloom 3.6 (7.8 pp) and Bloom 2.6 (6.9 pp).	The substrate zones in these rooms disagree with each other by roughly half the room's total dryback — either uneven emitter delivery or a probe in a bag that is not on-line. AUTOMAIT would steer on the average of that disagreement. Ten minutes with a hand meter resolves it.
P2	Owner	Check CO₂ delivery to the Bloom 2 wing.	Bloom 2.1 reads 368 ppm lights-on — ambient air — and the wing as a whole runs 368–935 ppm against 1000–1735 ppm in Bloom 3. Either enrichment is not plumbed to Bloom 2 or it is not running.
P3	Process	Re-login at app.aroya.io to restore the SPA session.	Two consecutive Insa audits have now shipped with signal-inferred phases because the harvest-group roster, cultivar list and AUTOMAIT configuration could not be read. This is a one-click fix on Insa's side and it materially tightens every target band in this report.

11 · Data Quality & Limitations11 · Calidad de Datos

Audit-level limitations — read these before quoting any number.

No harvest-group roster, no cultivars, no AUTOMAIT configuration. The AROYA SPA session is invalidated (401 — this token is no longer valid due to user token cutoff). Re-login at app.aroya.io restores it. Every phase label in this report is therefore signal-inferred (SPA roster unavailable) and is marked as such.
We do not claim AUTOMAIT is, or is not, currently enabled at Insa. That lives in the SPA configuration, which we could not read. Nothing in public_api proves it either way, so no assertion is made.
The Bloom 2 wing (2.1–2.6) does not have a 14-day window. Substrate telemetry begins 2026-07-08 (2.6: 2026-07-02) and climate telemetry begins 2026-07-10. Those rooms are reported on 4–11 days of real data and are explicitly not graded against a 14-day standard. No data has been synthesised, enveloped or extrapolated to fill the gap.
Shots and P0/P1/P2/P3 cadence are DERIVED from VWC step-up signatures (a rise of ≥ 2.0 absolute pp between consecutive native-cadence samples). There is no flow or valve telemetry in public_api. Reliable for rhythm, not for absolute shot volume.
No runoff series — runoff percentage and runoff EC cannot be graded. Only Bloom 1.6 exposes a drip/drain EC pair.
Overnight dryback is absolute pp throughout — daily peak VWC minus the following pre-irrigation low, zone-mean. It is never expressed as a proportion of field capacity.
Photoperiod and therefore P0 delay are inferred from each room's diurnal air-temperature ramp. In Bloom 3.2, 3.3 and 3.4 the swing is too small (or absent) for that inference to hold — treat their P0 figures as unusable.
Device last-seen is derived from telemetry timestamps, not from AROYA's live device-status endpoint (which is part of the SPA and therefore unavailable). See the companion device-health report.

Coverage = share of the expected 2,240 native-cadence samples over the 14-day window (2026-06-30 13:00 UTC → 2026-07-14 13:00 UTC).

Room	Substrate coverage	Climate coverage	Notes
Bloom 1.3	100%	100%	Full substrate + climate suite.
Bloom 1.4	100%	100%	Full substrate + climate suite.
Bloom 1.5	100%	100%	zone 81548: pore EC pegged (mean 12.4, max 25.0 mS)
Bloom 1.6	100%	100%	Full substrate + climate suite.
Bloom 2.1	42%	27%	Climate only 27% of window. Full substrate + climate suite.
Bloom 2.2	42%	27%	Climate only 27% of window. Full substrate + climate suite.
Bloom 2.3	42%	27%	Climate only 27% of window. Full substrate + climate suite.
Bloom 2.4	42%	27%	Climate only 27% of window. Full substrate + climate suite.
Bloom 2.5	42%	27%	Climate only 27% of window. Full substrate + climate suite.
Bloom 2.6	84%	27%	Climate only 27% of window. Full substrate + climate suite.
Bloom 3.1	99%	98%	Full substrate + climate suite.
Bloom 3.2	99%	22%	Climate only 22% of window. zone 84065: pore EC pegged (mean 20.6, max 30.0 mS); zone 84065: VWC collapsed to 0.0% (mean 15.4%) — probe out of media / dead
Bloom 3.3	100%	100%	Full substrate + climate suite.
Bloom 3.4	98%	0%	NO climate telemetry — 0 air-temp samples in 14 d. zone 84071: VWC collapsed to 0.8% (mean 18.7%) — probe out of media / dead
Bloom 3.5	99%	98%	Full substrate + climate suite.
Bloom 3.6	98%	98%	Full substrate + climate suite.