A 30-person recruitment agency. 40–60 emails per day landing in a shared inbox. Our AI triage pipeline classified every email, routed each one to the right person, and drafted a contextually appropriate reply — automatically, before anyone arrived at their desk.
"This is a live demonstration using our production triage pipeline. The agency scenario is representative; the AI processing is real."
12
Emails processed in demo
11.4s
Total processing time
2
Urgency-5 leads surfaced
~8min
Admin time needed (review)
Manual inbox triage is genuinely hard, repetitive, high-stakes cognitive work — done from scratch each morning.
| Activity | Time |
|---|---|
| Daily inbox triage (read, classify, route, forward with context) | 2 hours |
| Working days per month | 22 |
| Hours per month on email sorting | 44 hours |
| Annual equivalent | 528 hours — over 13 working weeks |
And that assumes everything goes smoothly
When the administrator is on holiday, unwell, or pulled onto something else, emails pile up — and the urgent ones sit there unanswered while a competitor picks up the phone first. Manual triage is also fatigue-dependent: email 40 gets less thought than email 1. The AI applies the same criteria, at the same quality, to every single email in the batch.
The pipeline runs silently in the background, triggered the moment a new email arrives.
Every email analysed for intent, urgency (1–5), confidence, and key entities — company name, role mentioned, timeline, signals extracted automatically.
The right email goes to the right person — immediately. Critical issues DM'd to the account manager. New leads to the BD team. Invoices to accounts. Spam archived silently.
For every non-spam email, a contextually appropriate draft reply generated — ready for one-click approval in Slack. Nothing is sent automatically. The human stays in control.
| Category in demo | Count |
|---|---|
| New business inquiries | 3 |
| Candidate applications / queries | 3 |
| Existing client follow-ups | 2 |
| Finance / invoice | 1 |
| Internal (team member) | 1 |
| Spam / vendor pitches | 2 |
| Total processed | 12 emails in 11.4 seconds |
Four of the twelve emails processed — each showing the classification result and draft reply generated.
Incoming email: CFO at a 180-person Leeds financial services firm. Head of Compliance leaving end of month. Regulatory examination in Q2. Wants a call that morning. Warm referral from an existing client.
Hi David,
Many thanks for reaching out — and for the kind introduction from [Referrer]. A Q2 regulatory examination with your Head of Compliance departing end of month is exactly the kind of situation we've helped clients navigate before, and I'd very much like to speak with you this morning if possible.
I have availability at 10:30am or 11:15am today — please let me know which suits, and I'll send a calendar invite immediately. If neither works, I'm free most of this afternoon as well.
AI-generated · Ready for one-click approval in Slack · 1.8 seconds from email arrival
What happened: Sophie received this alert before 8:00 AM — including the cross-reference to Email 03 (a matching candidate who arrived 34 minutes later). In the manual process, both would have sat unread until 10:15 AM. A competitor answering at 8:30 AM secures the brief.
Incoming email: Existing law firm client reporting that a placed candidate is considering withdrawing two days before her start date. Requesting an immediate callback.
Hi [Name],
Thank you for letting me know immediately — this is exactly the right thing to do and I'm glad you called it out now rather than Monday morning.
I'm going to speak with [Candidate] directly within the next 30 minutes and I'll call you back by 11:00am with an update. In the meantime, please don't discuss this further with the candidate directly — let me handle that conversation.
I'll be in touch very shortly.
Urgency detected · Account manager DM'd immediately · Draft ready for approval · 0.9 seconds
Incoming email: Formal Request for Proposal from a plc — a multi-year preferred supplier agreement across Financial Services, Legal, and Finance. Response deadline: 10 days.
An ATS software cold sell and an SEO marketing pitch. Together they would have taken 6–8 minutes to read, recognise as irrelevant, and delete.
Spam and vendor pitches typically account for 20–25% of volume. At 3 minutes each across 40–60 daily emails, that's 24–45 minutes per day returned to useful work — from spam filtering alone.
| Step | Manual Process | ShortlistOps A2 |
|---|---|---|
| Read all 12 emails | ~24 minutes | Instant |
| Classify each email | ~12 minutes | Instant |
| Route to correct person with context | ~10 minutes | Instant |
| Draft acknowledgement replies | ~20 minutes | Automatic |
| Log to CRM | ~15 minutes (often missed) | Automatic, every time |
| Urgency detection | Variable (fatigue-dependent) | Consistent |
| Cross-referencing (e.g. candidate match) | Never | Automatic |
| Spam filtered | Manual delete | Auto-archived |
| Works on bank holidays / during leave | No | 24/7 |
| Total admin time (12 emails) | ~89 minutes | ~8 minutes (review + approve) |
For a 30-person agency handling 40–60 emails per day.
44h
saved every month
vs. manual triage
528h
saved every year
over 13 working weeks
78×
return on running cost
~£11/month to operate
| Cost (at £18/hr admin rate) | Manual | ShortlistOps A2 |
|---|---|---|
| Monthly hours on inbox triage | 44 hrs | ~4 hrs (review + approve) |
| Monthly cost at £18/hr | £792 | £72 |
| Monthly saving | £720 / month | |
| Annual saving | £8,640 / year | |
The conservative case. The real value often comes from what automation prevents.
"The first morning it ran, I actually had nothing to do at 8am. It was strange. Everything was already sorted."
— Office Administrator, 28-person recruitment agency
"We nearly missed an RFP response deadline last year because it got buried. That won't happen again."
— Managing Director, specialist legal recruiter
"The urgent emails come to me immediately now. I used to find out about problems hours later."
— Account Manager, financial services recruitment
| Component | Tool |
|---|---|
| Automation layer | Make.com |
| AI classification & drafting | Claude Sonnet (Anthropic API) |
| Email integration | Gmail / Microsoft 365 |
| Notifications & approval | Slack (Block Kit) |
| CRM logging | HubSpot / Bullhorn |
| Audit log | Google Sheets |
Setup time: 5–7 hours, one-off
Running cost: ~£11/month all-in, up to 200 emails/day
Uptime: 24/7, including bank holidays and annual leave
Audit: Every classification, routing decision, and draft reply is logged. One click to re-route if needed.
Book a demo and we'll walk you through a live demonstration using your own mailbox — you'll see the AI classify, route, and draft replies to real emails in real time.
Or email us at anas@shortlistops.co.uk — we'll get back to you within 24 hours.
Call or WhatsApp: +44 7388 281312