Anyone using Claude Scheduled Tasks with Hospitable MCP?

For what it's worth, both these use cases are supported in the product already, with relatively simple messaging rules (no code, no third-party products).

5 days before check-in, I send a message that says:

--------------------
Hey Laurie, I was looking over our calendar and noticed the night before your stay is still available. If you had any interest in arriving a day early, we could offer a pretty good discount, since you’re already booked with us.

The same offer actually applies to the night after your stay too, if you wanted to tack on an extra night.

No pressure either way — just wanted to offer in case you had flexible plans. If not, we'll see you as planned on Saturday, August 22 :-)
--------------------

Of course it’s infused with the appropriate short-codes for customization. The outer part sends only if “the night before is available” and the inner part is a custom code that only injects if “the night after is available” (and then technically I have a second permutation for situations where only the night after is available and not the night before). 

This message is highly successful, BTW. 


And then for booking inquiries where you respond with a pre-approval and they never accept it, there’s a messaging condition for “expired pre-approval,” and for that I say: 

--------------------
Hi %guest_first_name%, looks like your pre-approval expired... I'm guessing you've booked another place by now — happens all the time, no explanation needed! Hope you have an awesome time in %listing_city% :-) In case you're still looking, though, I figured I'd reach out to see if there are still any questions I can help with? -- Regards, Tom
--------------------

This one just got me a booking today, by nudging someone who wandered off mid-conversation after their initial inquiry :-)

Hi ​@Tom Beerley , this is super helpful. Thank you for sharing.
I’m new to Hospitable, please excuse questions but how to set the rule for “The outer part sends only if “the night before is available” and the inner part is a custom code that only injects if “the night after is available”?

Hi Ian,

I checked the github repo you posted. It looks like the automations are missing. All I see is a license file and a readme. 

how do you set it to determine if the night before or after are available?

In the “condition” section of a new messaging rule, you can specify that it should only send if “the night before check-in is available”. That messaging rule is the “outer part.” Write the words for offering the “night before,” and somewhere in that message you can slip in a “custom code” (a snippet of text that you define elsewhere under “Custom Codes”). The custom code (which is the “inner part”) can itself have a condition applied to it, so that it only gets injected when “the night after checkout is available”.

If neither night is available, can you stop the message from sending?

Thanks!

Each message (or custom code) can only support one condition, so to cover all the permutations, you’ll need a couple different messages.

Possible permutations:
Scenario A: “Before” only
Scenario B: “Before” and “After” 
Scenario C: “After” only
Scenario D: Neither

If the “night before” IS available, then send a message saying so, and inject a custom code into that message to mention that the “night after” is ALSO available. This covers scenarios A and B.

If the “night before” IS NOT available, then send a message that contains NOTHING except a custom code whose condition is “night after IS available”. This covers scenarios C and D. This message is the one that hurt my brain the most, LOL. The reason this message covers scenario D is because a blank message will not be sent, so if the night before is not available and the night after is also not available, then this will produce a blank message inside of a blank message, which will be suppressed.

Hello, ​@ian!

Really interesting workflow — especially the fact that you’re checking conversation context before sending the offer. The next layer we’ve found important is what happens after the guest says yes. For example, an early arrival can change housekeeping priority, access timing, or other operational tasks. With Una, we handle that handoff from guest conversation → decision → operational action → team/task → confirmation. Curious whether you’re planning to extend the workflow into execution as well, or keep Airtable mainly as the state layer? 

​@Polydom_Lana I checked on Una and it seems great, but $590/month for 3 STRs is really crazy high.

Wes

Hi ​@WesA - glad to hear you liked Una!

Just to clarify: $599/month is not for 3 STR listings — it covers up to 30 units.

Una also goes well beyond guest messaging. She handles guest communication, housekeeping and maintenance task dispatching and coordination — 24/7, in multiple languages, and across all your WhatsApp groups, email, inbox, CRM and etc. at the same time.

For all of that work, Una’s cost comes to about $0.82 per hour with responses typically within three seconds - not bad, I believe! 😎

For just 3 units, though, I completely understand that the standard package may not be the right fit as-is. Feel free to reach out, and we can discuss what might make sense for your setup.

lana.u@polydom.ai

I created a daily task that checks any guests checking out tomorrow and if all has gone well with the guest and it’s possible, it sends them an offer to extend at a small discount. 

That’s really interesting — so you’re already moving beyond analysis into actual execution.

How are you determining that “all has gone well” with the guest? Is the system evaluating the conversation/guest sentiment and operational signals automatically, or is that based on predefined rules?

And if the guest accepts the extension, does the workflow complete the reservation change automatically as well? This is exactly the execution layer I’m especially interested in.

So I just let the AI determine whether it thinks that there aren’t any issues. I likely could have it give me a confidence score. It will evaluate the conversation and determine if there was ever any issues that the guest reported, and if it seemed like those were resolved quickly and easily or if it seems like the guest might be a little unhappy.

It will actually send a notice In Slack with a list of the guests and the approximate Discount amount. It will wait 30 minutes and read any replies to that Slack message so I could reply. Would send all or send one or send two etc.

It will then send a message to that guest, and if the guest Replies our VA’s will just review that Slack post, which will have the numbers for them and they will alter the Reservation and submit an adjustment. 
 

I don’t let the system run 100% automated but I try to make it as smooth as possible.

That makes sense — you’ve already automated the analysis and outreach, while keeping the final reservation change human-controlled.

Out of curiosity, if that second part could also be automated — understanding the guest’s reply, checking availability and conditions, updating the reservation, and completing the adjustment — would that be something you’d be interested in exploring?

We’ve been working quite deeply in that execution layer, so I’d be very interested to compare notes.

I am pretty sure it can be done 100% but I prefer not to go that level until I am 100% sure, also in some cases we have to setup unique rules like don't go below w certain total, for instance if a property is already down to $100 a night, I won’t want to go below that so we will not include the discount just the offer.

That’s exactly the point — full automation only works if you can define very clear guardrails around it.

I think we’re working on a very similar problem from slightly different angles. We’re very focused on the execution layer, and our agent can operate within clearly defined business rules — including when it can act, when it should escalate, and what conditions or discount parameters should apply.

I’d be more than happy to share how we’re approaching this through our existing Hospitable integration. I think it could be a very interesting conversation.😎