★ Scan miss, corrected 7/28 — Schmitt was approved while the board said “Rejected.” Cortney sent “4 Approved Permit” on Mon 7/27 at 3:21pm. The scan ran at 2:38pm and reported five days of silence — the approval landed 43 minutes later, and the board carried the wrong state for 20 hours, including through a production meeting. No search error this time; the scan was simply early. The lesson is different from the 7/21 and 7/27 misses: a scan result is true only at its timestamp, and on a permit with a booked inspection two days out, that is not good enough. Where a card asserts something urgent and negative — “no confirmation,” “still rejected” — it should carry the scan time next to it, and anything with an inspection inside 48 hours deserves a re-check rather than a daily one. Also note what the approval carried: the permit was not simply released — it came with an inspection hold, a CO hold and a fire-sprinkler condition. “Approved” is not “clear,” and the conditions are where the work now is.
Correction 7/27, from Karis's portal view — the NOC claim on Slaski was wrong. This board asserted for days that the Bonita NOC was unrecorded and blocking the first inspection. The portal workflow says Notice of Commencement — Passed 07/23/2026. Where it came from: Cortney's “This doesnt seem to be recorded” on 7/22, which was true when she wrote it and was never re-checked once the permit moved. The 7/23 ready-for-issuance notice arrived the same day the NOC passed and should have been read as evidence it had cleared — a permit does not go ready-for-issuance with an open NOC step. The lesson is the 7/21 one again in a new form: a blocker recorded from a third party's in-passing remark needs the same re-verification as a status, or it ages into a fact. Two of the four items on that card were stale — the NOC, and an issuance-fee question the issuance itself had already answered. Neither the inbox nor Albi would ever have caught this; only the portal shows the workflow steps. Worth pulling the portal view on each issued permit rather than inferring state from email.
Scan 7/27 2:38pm: one change — Slaski/COM26-134654-BOS issued at 1:25pm. Albi: nothing. All nine tracked jobs re-pulled by job number; every one unchanged — Schmitt, Heasley, La Peninsula, Slaski, Kelley, Sieber and Bovett still Work In Progress, both Shah jobs still Final File Review, and no new notes and no new tasks on any of them since 7/24. The only post-7/24 Albi item at all is the Bovett invoice-request task from 7/24 7:30am, which is now marked completed. Repeat of the 7/21 lesson, and it nearly repeated exactly: the keyword search (permit OR inspection OR issued…) did not return the issuance email, even though “issued” is in its subject line — Gmail truncated the result set at ten threads. It was found only by a second, unfiltered sweep of everything newer than a day. Do not trust the keyword search alone; always follow it with a plain recency sweep. Note also that both Bonita notices — ready-for-issuance and issued — went to KEN@aeretusa.com and reached office@ only because Ken forwarded them. Bonita has the wrong contact on this permit.
Permit Portal is live with Elite: the walkthrough happened Friday 7/17 (Zoom 11:00am, then a Teams session at 2:45pm). It took immediately — on 7/17 at 3:56pm Cortney sent the Naples plans answer with "I added this into the permit portal you created, but just wanted to send an email about it as well". Elite is now posting status into the portal rather than only emailing.
★ RESOLVED 7/27 — the missing eighth permit is 1500 Popham Dr #A38, and Karis named it, not the scan. Lee County RES2026-04564, job 26-0142-RES (Michael Greene), issued 6/18 and now in the inspection period. It has its own card. Why no scan would ever have found it: it was issued five weeks before the scan window opened, and a permit that is working generates no mail — the last correspondence was 6/23. The board's whole detection method is inbox deltas plus Albi status, and this permit is invisible to both by design. Standing lesson: a quiet permit is not an absent one. Any future “how many are in flight” question should be answered by enumerating Albi jobs with permit activity, not by counting threads. The Bovett theory was wrong and is retracted — 26-0197-RES had no permit trail because it has no permit, not because one was hiding. It is still worth one question to Zac about the completed plumbing rough-in, but it is not the eighth permit. Not it either: the “Sieber Residence” fire-inspection request that arrived 7/21 is Deer Crossing under the owner's name — same permit PRMFH20260624432, same unit. One alias, not a new permit.
Albi 7/24: all tracked jobs re-pulled by job number — every one still Work In Progress, no status change, no stage change, no new notes and no new tasks since 7/22. The single exception is 26-0197-RES Bovett: a new Invoice Request task at 7:30am on 7/24 — $7,406.77 progress draw, “plumbing rough in completed per contract.” That matters for the missing-permit question below: a job the 7/22 production list called “still working out scope details,” 36 days silent, has in fact completed plumbing rough-in. Rough-in is an inspected milestone. Either that work is riding on a plumbing sub's own permit, or there is a permit on this job that has never appeared in the inbox — worth one question to Zac, because it decides whether Bovett is the eighth permit or simply not a permit job at all.
Albi 7/22 (morning): all eight tracked jobs re-pulled by job number — no status change, no stage change, no new notes and no new tasks in the last 20 hours. Schmitt and Heasley remain Work In Progress, both Shah jobs remain Final File Review. Schmitt's latest note is still the 7/21 10:46am mitigation-complete entry. Everything that moved yesterday moved in email, not in Albi — which is the pattern on every permit here: the system of record lags the actual permit by days.
Scan miss, corrected 7/21: the automated inbox scan reported "no permit changes" on the morning of 7/21 and was wrong. Cortney emailed the county's structural rejection on 7/20 at 3:37pm, and there was a four-message exchange that day. The scan's search returned the Schmitt thread but listed only its messages through 7/16, and the thread was never opened — so a same-day reply from Elite was read as five days of silence. Fix: open the thread, never trust the search snippet's message list. Elite is communicating; the delay is in filing, not in reporting.