EMAIL AGENTS
Agentic email for hotel back office.
Launching July 2026: D3x reads messy guest and finance threads, identifies the intent, checks PMS and policy context, drafts the full resolution, and asks staff to click Agree. No more Gmail buttons that only generate text.
AGENTIC EMAIL WORKSPACE
The agent drafts the whole resolution. Staff click Agree.
Instead of asking operators to open Outlook, Gmail, or Freshdesk and generate a reply, D3x prepares the decision: what changed, what policy applies, what system action is safe, and the exact email to send.
Inbound thread
Change invoice and extend booking — BK-884291
Sarah Chen · corporate guest · Staycity Dublin
Hi, can you add one extra night to my booking and update the invoice to Chen Consulting Ltd? The finance team also needs the VAT number on the PDF.
Detected intent
Booking + invoice update
Confidence
High · policy match
Context graph pulled
- Booking
- BK-884291
- Channel
- Corporate direct
- Payment
- Company card · paid
- Policy
- Modify allowed
- Invoice
- PDF can regenerate
- Risk
- Human approval
Decision tree
- Guest matches original booker and corporate domain
- Extra night available at contracted rate
- Invoice legal entity change requires approval
- No refund or payment reversal requested
Ready for approval
Hi Sarah, Done — I extended your stay by one night at the corporate rate and updated the invoice to Chen Consulting Ltd with the VAT number included. The revised PDF is attached. Please reply if your finance team needs a different billing address.
Audit trail
- reservation.matched → BK-884291
- rate.checked → corporate rate available
- invoice.regenerated → approval required
- operator.approval → pending
START WITH TWO WORKFLOWS
Do not boil the ocean. Solve bookings and invoices deeply.
The first release is intentionally focused. It starts with two email queues hotel teams see every day: booking changes and invoice requests. Each looks simple, but the agent checks policy, context, and approvals before anything is sent or updated.
FLOW 01
Booking changes
Change dates, guest details, add-ons, arrival notes, corporate rules, and cancellation edge cases with PMS context and policy checks.
- Is the sender allowed to change this booking?
- Is the booking prepaid, OTA, direct, or corporate?
- Does the date change sit inside the policy window?
- What downstream systems need updating?
Prepare safe lookups and email drafts automatically; ask for approval before changing bookings, applying fees, overriding policy, or handling exceptions.
FLOW 02
Invoice requests
Correct company details, VAT numbers, billing addresses, folio attachments, revised PDFs, and finance team handoffs.
- Is this invoice linked to the right booking?
- Can the requested billing name be changed?
- Does the VAT or company detail need validation?
- Should accounting, front desk, or finance review?
Prepare the corrected email, attachments, and required system updates; keep finance-sensitive steps approved by a human.
CHAT ON RAILS
Agentic, but not free-form chaos. The system carries the judgment.
Hospitality operations need reliability more than open-ended chat. D3x uses AI for reading, classification, and drafting, while decisions run through explicit workflows, policy gates, and approval states.
RAIL 01
Intent-first
The agent detects booking, invoice, refund, complaint, or escalation intent before asking the operator anything.
RAIL 02
Context graph
The system pulls only the facts needed for that decision: booking, guest, invoice, property, policy, channel, payment, and time.
RAIL 03
Approval boundary
Reads and low-risk drafts can be automatic. Write actions, refunds, legal invoice edits, and overrides pause for approval.
RAIL 04
Audit replay
Operators and managers can see what was retrieved, why it was recommended, who approved, and what changed.
WHY THIS MATTERS
Back-office automation is where the leverage is.
The product thesis is simple: make high-volume hotel email work decision-grade, not chatty. Industry examples show invoice and back-office automation can unlock capacity when the workflow is narrow, governed, and auditable.
50%
AP processing capacity unlocked in a hospitality automation case study.
PPHE / Rossum
70%
invoice automation reported for hotel finance workflows.
DocMX hotel finance
94%
error-rate reduction reported in governed enterprise process-agent research.
FinRobot, 2025
2–3 WKS
on-site decision mapping sprint to capture real Staycity edge cases.
D3x pilot plan
Research references include PPHE/Rossum hospitality AP automation, DocMX hotel finance automation, and 2025 enterprise process-agent research. These are market context and design inputs, not guaranteed D3x outcomes.
TALK TO US
30 minutes with the founder.
Walk through this channel on a comparable group's deployment.