From 10–20 minutes of digging to a reply ready in seconds.
Built for The Inn at Fox Briar Farm
10–20 min→Seconds
from research and drafting to review
10–15
emails handled per day
5
inquiry types automatically triaged
The full email agent workflow — from inbox trigger to drafted reply.
The email wasn't the hard part. Finding the answer was.
Donna owns The Inn at Fox Briar Farm, a boutique inn and wedding venue in Pennsylvania.
On a typical day, 10–15 inquiries land in her inbox. Some are questions about weddings. Others are private events, overnight stays, general property questions, or messages that don't need her attention at all.
Responding to them meant more than writing an email.
Before Donna could answer, she often had to figure out what the guest was asking, search across the tools where the relevant information lived, find the right reservation or policy, check calendars or contracts when needed, and only then write the response.
Every message started with the same thing: digging.
And because everything arrived in the same inbox, an urgent message could look exactly like a routine one until Donna stopped what she was doing and opened it herself.
The challenge wasn't getting AI to write an email.
That part is easy.
The harder problem was giving an AI system enough understanding of the business to prepare an answer Donna could actually use.
The right response might depend on a live reservation. Or a date on Google Calendar. Or something written in a wedding contract. Or a property policy. And different kinds of inquiries needed different information.
A useful system had to know what the guest was asking, know where to look for the answer, retrieve the right context, and use it to draft a response that sounded like Donna.
It also had to know its limits.
The goal was never to put customer communication on autopilot.
It was to remove everything that happened before Donna's judgment was needed.
Our approach
Let the agent do the digging. Keep Donna in control.
We designed the system around a simple division of responsibility.
The agent handles the repetitive work computers are good at: reading the inquiry, determining what it needs, searching the right business systems, gathering context, and preparing the response.
Donna keeps the part that should still belong to her: reviewing the message and deciding what gets sent to her guests.
That means she gets the leverage of automation without handing the customer relationship over to a black box.
What we built
It understands what came in.
Every new email is automatically classified into one of five paths:
Wedding inquiry · Private event · Inn stay · General question · Ignore
That first decision determines what happens next and what information the agent may need.
It works from Fox Briar's actual business information.
The agent isn't answering from generic AI knowledge.
It can pull context from the sources Donna already relies on to run the property, including property information and policies, live reservations, Google Calendar, and wedding contracts.
Instead of Donna moving from tool to tool looking for the answer, the system brings the relevant information to the email.
It prepares the response in Donna's voice.
Once the agent has the context it needs, it drafts the reply and saves it directly in Gmail.
Nothing is automatically sent.
When Donna opens the message, she isn't facing a blank screen or starting another round of research. The work is already there for her to review, adjust if needed, and send.
It knows when something needs attention now.
Critical messages trigger a Telegram notification so an important inquiry doesn't disappear into an otherwise routine inbox.
Automation handles the ordinary work without making the unusual work easier to miss.
Why we built it this way
The goal wasn't to remove Donna from the process.
For a hospitality business, the way you communicate with a guest is part of the product.
Fully automating the inbox might have saved another click, but it would also have removed Donna from a moment where her judgment and personal touch still matter.
So we automated around her instead.
The system removes the searching, sorting, context gathering, and first-draft writing that consumed her time while preserving the final human decision.
That distinction matters to us.
Good AI doesn't automate the most human part of a process just because it can. It takes responsibility for the work that keeps the human from getting to that part.
The result
10–20 minutes → seconds
The research and drafting that used to begin every inquiry is already done when Donna opens the message.
10–15 emails every day
The system works across the daily flow of incoming inquiries rather than solving one narrow use case.
5 inquiry types
Wedding inquiries, private events, inn stays, general questions, and messages that need no response are each handled through the appropriate path.
But the larger change is simpler:
Donna no longer has to become a researcher every time she opens her inbox.
The information she needs comes to her. The draft is waiting. Urgent messages stand out.
She gets to spend less of her day assembling answers and more of it making the decisions only she can make.
"It's like having someone who already knows everything about the farm sitting with me when I open my email."
— Donna, Owner
Have a process that looks something like this?
You don't need to know what kind of AI agent would solve it — or whether AI is even the right answer.
Tell us about the work that keeps repeating, the information your team keeps hunting down, or the process that takes longer than it should.
We'll help you figure out what belongs in the system and what should stay human.
Talk through your business →