Situs phone wiring: how calls reach Clara

Fede's preference was (b), "forward the whole line with our own business-hours logic." It gives us control over hours, but daytime reliability then depends on us, Eileen's no-delay ask fails during the day, and it is the exact shape that looped at Western Slope. Gera's 5-second budget fits (a) comfortably.

Western Slope forwarded its whole RingCentral main line to Clara (Sep 7 decision). Clara's "something else" transfer pointed at that same main line. The main line forwarded straight back to Clara, so on Sep 22 three real calls bounced and dropped (calls CA765a4f88, CA14680d90).

• Fix 1 (Sep 22): a number the customer forwards into Clara is never a valid transfer target. Clara takes a message instead.
• Fix 2 (propflowai #18002, Oct 8): the call-center leg and ring-list saves run through the same guard. A teammate whose number loops back is refused by name when saved.
• Not built (part 2): the guard reads a hand-kept list of forwarded lines. It does not yet detect a forward from Twilio's ForwardedFrom. So for Situs, (303) 789-3030 and the two numbers that route to it must be entered by hand, and the office catch-all must be a direct staff number or "take a message", never the main line.

Under (a), Crexendo never sends a daytime call to Clara, so there's no "send it back" step. The only remaining loop is Clara dialing the main line herself, which the guard blocks once the number is registered.

Crexendo runs on NetSapiens, which it bought in 2021 (announcement). Its knowledge base is cxdo.helpjuice.com.

Not found publicly: whether a Ring Group or auto-attendant can target an outside number directly, and which Remote Storage providers are supported. Ask Dani. (Both resolved in the Oct 13 evening pass below.)

Recommended: record on Crexendo, download on our side. No change to how calls route, no delay, no loop risk. In order of preference:

1. API token (from Crexendo, not a portal login): list call records, fetch each recording's file-access-url, feed them into our call review for Eileen's quality readout. Can be automated.
2. Remote Storage to a bucket we control, if Crexendo supports one we can use.
3. Zapier "New Call Recording" → a shared folder we read. No code, works with only a portal login.
4. Manual portal download as the floor. Fine for a first batch of calls.

Pull within 90 days (Crexendo's media retention). These are calls staff answer today, so they show what callers ask and how the team responds. That is the training material.

Not recommended: passing calls through us to record them. That is option (b) for a week: daytime delay, the loop shape, and Situs's phones depend on us, all before Clara answers a single call. Our own recording starts the day Phase 1 goes live; Clara's calls are recorded end to end on our side.

Caller notice: Colorado only needs one party's consent, but check whether Crexendo plays a recording notice. If staff calls are recorded for training, Situs may want one.

Hugo wants Clara to catch calls while agents are out at showings (Gera and Fede are cooler on it). The likely Crexendo setup: daytime ring group → Ring Timeout (say 20–25 s) → a system user whose Answering Rule is "Call Forwarding, Always" → Clara's number. That is the same pattern as Night Mode's routing user. Verify it in Situs's portal once we have access.

Loop to prevent: a caller asks Clara for a person, Clara dials the ring group, nobody answers, and the ring group sends the call back to Clara. Build: (1) Clara's daytime "talk to a person" dials direct staff numbers that don't overflow, never the ring group or main line; (2) if a call reaches Clara a second time within a few minutes after Clara forwarded it, she takes a message instead of transferring again. This is part 2 of the loop guard, using Twilio's forwarded-from data.

Daytime rules Situs already gave: routine tickets go into AppFolio for Beryl, no tech is pulled off their planned day, and Clara never assigns.

1. Office hours, holidays, and when the weekend starts. These are the time frame. Our hours must match Crexendo's to the minute, Mountain time.
2. How does MCC get after-hours calls today? The main number's current after-hours slot answers this. Does any number still go straight to MCC?
3. What are the two other numbers that route to the main line, and should Clara cover them?
4. Who has the Crexendo admin role? It's an add-on seat feature. If nobody does, Dani makes the change.
5. Is the login Eileen is getting a portal login or an API token? Recording download is much easier with a token.
6. Can Crexendo swap the forward announcement for ringback? And can SIP trunk to an outside address be a later option?
7. Is automatic recording on for the seats that answer the main line? Do they want a recording notice?
8. Who does Clara transfer to after hours when a caller wants a person, if anyone? Must be a direct number, never the main line.
9. Phase 2: which ring group or queue answers in the day, and how long should it ring before Clara picks up?

Sources: Situs meeting and debrief notes (Oct 13); phone line and hand-off architecture; Western Slope call tree; WS emergency line hand-off.

A second read of the Oct 13 call transcript and Crexendo's public docs, done after the page above was published. Nothing here changes the recommendation. It fixes three "not found" gaps, adds the facts that matter for Friday, and settles the vendor-name question as far as public sources allow.

Two things Eileen said that shape the week: she will ask Dani for one extra login with no new DID, framed as maintenance-only; and she wants a readout of who on staff is using the new system at all, calls and texts, as her own management tool. Both are served by the same login or key. Transcript: #transcripts, "PropFlow Situs Follow up", Oct 13.