Overview
Northline Group / last 30 days
Sample workspace. Every figure on this screen is illustrative and belongs to no client. This is the interface a Krateon client is given on day one.
Work completed
1,284
+18.4%vs prior 30d
Tasks the agents closed without a human touching them
Hours given back
96 h
+11.2%vs prior 30d
Measured against your own time per task, not an industry average
Cost of this work
$3.4k
$7.8k undera hire at the same volume
Retainer plus usage. A hire at this volume costs $11.2k loaded
Meetings booked
14
+4vs prior 30d
Accepted on the calendar, not replies
Problems surfaced
9
+3vs prior 30d
Issues found in your operation that nobody had reported
First time right
94.2%
from 81.0%at deployment
Tasks closed correctly with no retry and no human correction
Work volume
30 DAYS
JUL 18JUL 26AUG 03AUG 10AUG 17
Tasks closed per day
Average per day
43
Busiest day
92 Aug 17
Longest queue wait
4m 10s
What is worth solving next
ORCHESTRATOR
01
Quote follow up, day 2 and day 5
41% of quotes sent in the last 30 days got no second touch. The ones that did closed at 3.1x the rate. This is the largest gap the agents can see in your own data.
HIGHEST ROI~14h / MONTHREADY TO BUILD
02
Inbound triage before 9am
Median first response on overnight enquiries is 6h 40m. Everything arriving after 18:00 waits for the morning.
~9h / MONTHREADY TO BUILD
03
Supplier price drift watch
Three suppliers changed line pricing twice this quarter and neither change reached the quoting sheet.
NEEDS 1 INTEGRATION
Problems the agents found
9 OPENREADING CONTINUOUSLY
HIGH
Quotes above $8k stall at legal review.
Eleven of the fourteen quotes over $8,000 sat with legal for nine days or longer. Under that number the same step clears in two. Nobody reported it because each quote looks normal on its own.
EVIDENCE: 142 quote records, 30 days · CONFIDENCE: HIGH
$46,000
PIPELINE EXPOSED
HIGH
Two reps answer the same six questions all day.
Thirty eight percent of inbound support volume is six repeating questions. One help page and one agent reply template removes almost all of it.
EVIDENCE: 1,847 tickets clustered by intent · CONFIDENCE: HIGH
22 h
PER MONTH
MEDIUM
Renewal reminders fire after the renewal date.
Seven accounts were reminded three to eleven days late. The trigger reads contract start instead of contract end, so the error grows with contract length.
EVIDENCE: 7 of 34 renewals, CRM audit · CONFIDENCE: HIGH
$18,400
ARR AT RISK
MEDIUM
You are paying twice to enrich the same companies.
Two hundred and fourteen duplicate lookups across two enrichment tools in thirty days. Same domains, both billed.
EVIDENCE: usage export, both vendors · CONFIDENCE: HIGH
$310
PER MONTH
LOW
Nobody opens the Friday report.
The weekly PDF has been opened twice in nine weeks. It takes three hours a month to produce.
EVIDENCE: document access log · CONFIDENCE: MEDIUM
3 h
PER MONTH
Found by reading your own records. Nothing here was reported by a person.Export all 9
How the system is getting better
SINCE DEPLOYMENT, 94 DAYS
Every task the agents close is logged, scored and fed back. These four numbers are the ones that move when a system is actually learning rather than just running.
First time right
94.2%
81.0% at deployment
Escalated to a human
4.1%
19.0% at deployment
Cost per closed task
$2.65
$6.10 at deployment
Retry rate
2.3%
8.7% at deployment
Every change to agent behaviour was approved by a person before it took effect.
Agents
3 LIVE / 1 IN BUILD
| Agent | Throughput | Status |
|---|---|---|
|
Client acquisition desk find, qualify, research, write, sequence |
612 |
RUNNING |
|
Inbound triage read, classify, answer or escalate |
438 |
RUNNING |
|
Call scoring and coaching score every call, write the report |
234 |
RUNNING |
|
Quote follow up recommended by the orchestrator |
not deployed |
IN BUILD |
Throughput is tasks closed in the last 30 days.Open agents
Handed to a human
7 THIS WEEK
Pricing outside the agent's range
Enquiry asked for volume terms above the ceiling you set. Nothing was quoted. Full thread attached.
14m
Complaint, second contact
Same account wrote twice in 48 hours. Routed straight to you, no auto reply sent.
2h
Approved and sent
You cleared the revised quote for Harbour Freight. Agent resumed the thread.
5h
Contract language request
Prospect asked to change a liability clause. Legal wording is never generated.
1d
Every escalation carries the reason and the full thread.All 7
Where the money landed this month
AUG 2026
Cost of the system
$3,400
Retainer plus usage, no setup this month
Same work, one hire
$11,200
Salary, tax and tooling, loaded, at this volume
Difference
$7,800
Saved this month on the work itself
Value of problems found
$64,700
Identified this month, not yet banked. Acting on them is your call
Hours returned
96 h
Measured against your own time per task
Cost per closed task
$2.65
Down from $6.10 at deployment as the agents learned