Clearing Triage — UK Admissions Agent
Working alongside university clearing teams
Live — A-level Results Day
Tue, 19 Aug 2026 · 08:14 BST
Queue: 14
Active agents: 1 AI + 42 human
Incoming queue 14 waiting · avg wait 2m 14s
Speed
Agent reasoning
Matched UCAS vacancies

UCAS clearing vacancy data

Live-feed mirror of publicly published clearing vacancies. Updated continuously through results day. The agent queries this index — it does not work from synthetic course data. Mock data is limited to the student side only.

Courses with places
Institutions
Total places
— results
Sorted by relevance to grade-match · last sync 08:13 BST

End-of-day retrospective — 19 August 2026

Zero operational risk: the agent sits alongside the human team, shadows the calls, and at the end of the day this is the artefact. 314 interactions handled in parallel. Where the agent agreed with the human team. Where it would have escalated. Where it would have been faster.

314
Interactions shadowed
89%
Outcome agreement
11m
Avg time saved
Provisional offers
198
+12 vs. human team alone
Escalated to human
47
flagged for wellbeing, complex, or borderline
Wellbeing signposted
13
Samaritans / Student Minds offered
First response time
14s
vs. 6m 22s human queue average
Outcome agreement breakdown
Same outcome
280
Agent faster
264
Agent escalated
47
Different outcome
17
Human-only edge
11
Where the agent escalated
Wellbeing concern
13
Borderline grade
10
Complex BTEC/access mix
8
International qual.
7
Hostile / parent-led
5
Subject substitution
4
Interaction-by-interaction transcript (sampled)
#048
Agreed: Student missed AAB by one grade (got ABB). Both agent and human team offered place on related BSc Psychology (Sheffield Hallam, 5 places, ABB requirement).
Agent reached outcome in 4m 12s vs. human 9m 38s — saved 5m 26s of phone time.
#102
Escalated by agent (correctly): Caller crying, said "I just want to give up on uni". Agent paused triage, signposted Samaritans, offered call-back later that day. Human team confirmed escalation.
No course was discussed. Agent flagged wellbeing concern within 28 seconds of call connect.
#171
Disagreed: Agent recommended Manchester Met for History; human placed at York St John instead. Reason: applicant's sister was at York St John (mentioned mid-call) — agent did not pick up the social cue strongly enough.
Action: add "family/social context" signal to next iteration prompt.
#214
Agreed + faster: Adjustment case — student got A*A*A, wanted to upgrade from Lancaster to LSE. Agent confirmed in 2m 10s that LSE Economics had no adjustment places, redirected to Warwick (3 places). Human took 14 minutes to reach same outcome.
12m 50s saved. Student accepted Warwick provisional offer.
#277
Agent caught what human missed: Caller said grades were "BBC" but later mentioned they re-sat one and got a B. Agent flagged that re-sit means effective BBB — qualified for the firm choice. Human team had logged as missed offer.
Outcome changed: confirmed at original university. Avoided unnecessary clearing process.
#298
Edge case — human-only: Mature applicant (35) with BTEC + 8 years work experience asking about Engineering. Agent gave correct course options but missed that institution has an APEL (accreditation of prior experiential learning) pathway. Human admissions tutor knew.
Action: ingest institutional APEL policies into agent context for next year.