One source of truth for the fund's deals and portfolio companies.
3.6 hours back today
Received files in by 7 AM
Mapped by known layouts and by the intake agent
Checked checks with flagged
Tied out control totals tied
PublishedEvery figure traced
Try a new layout
A file in a new layout goes through the same intake as the morning's files. The intake agent proposes a mapping, and code accepts it only when every check and the control total pass.
Twelve layouts the agent had never seen
Pick one to see what the intake agent did with it in a recorded run.
The intake agent's mapping
V01_2026-10-06.xlsx
This file's column
Schema field
Load Number
load_id
Ship Date
date
Customer
customer
Lane
lane
Miles
miles
Empty Miles
empty_miles
Linehaul
linehaul_rev
Fuel Surcharge
fuel_surcharge
Accessorials
accessorials
Carrier Cost
carrier_cost
On Time
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$93,248 = $93,248
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 7 seconds.
The intake agent's mapping
V02_2026-10-07.xlsx
This file's column
Schema field
Shipment ID
load_id
Date Shipped
date
Shipper
customer
Route
lane
Loaded Mi
miles
Empty Mi
empty_miles
LH Revenue
linehaul_rev
FSC
fuel_surcharge
Other Charges
accessorials
Carrier $
carrier_cost
OTD
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$109,965 = $109,965
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 7 seconds.
The intake agent's mapping
V03_2026-10-08.xlsx
This file's column
Schema field
Client
customer
Load ID
load_id
Load Date
date
Lane Code
lane
Freight Charges
linehaul_rev
Fuel
fuel_surcharge
Accessorial $
accessorials
Purchased Transportation
carrier_cost
Trip Miles
miles
Deadhead
empty_miles
On-Time Delivery
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$96,192 = $96,192
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 15 seconds.
The intake agent's mapping
V04_2026-10-09.xlsx
This file's column
Schema field
Pro #
load_id
PU Date
date
Bill To
customer
O/D
lane
Billed Miles
miles
Unloaded Miles
empty_miles
Base Rate
linehaul_rev
Fuel Adj
fuel_surcharge
Extras
accessorials
Buy Rate
carrier_cost
Service Met
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$108,680 = $108,680
Accepted. Every check passed.
The mapping differs from the answer key in a detail the sample rows didn't show, and it still parses every row the same way.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 13 seconds.
The intake agent's mapping
V05_2026-10-10.xlsx
This file's column
Schema field
Order No
load_id
Ship Date
date
Dispatcher
null
Account
customer
Lane
lane
Equipment ID
null
Miles
miles
Empty Miles
empty_miles
Linehaul $
linehaul_rev
FSC $
fuel_surcharge
Accessorials
accessorials
Carrier Pay
carrier_cost
On Time
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$97,783 = $97,783
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 9 seconds.
The intake agent's mapping
V06_2026-10-13.xlsx
This file's column
Schema field
BOL #
load_id
Pickup Date
date
Shipper
customer
Origin-Destination
lane
Commodity
null
Weight (lbs)
null
Loaded Miles
miles
Deadhead Mi
empty_miles
Line Haul Revenue
linehaul_rev
Fuel Surcharge
fuel_surcharge
Accessorial Charges
accessorials
Carrier Cost
carrier_cost
OTD
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$130,128 = $130,128
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 10 seconds.
The intake agent's mapping
V07_2026-10-21.xlsx
This file's column
Schema field
Load #
load_id
Date
date
Customer
customer
Lane
lane
PO Number
null
Loaded Miles
miles
Empty Miles
empty_miles
Linehaul Rev
linehaul_rev
Fuel Surcharge
fuel_surcharge
Accessorials
accessorials
Carrier Cost
carrier_cost
On Time
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$126,593 = $126,593
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 10 seconds.
The intake agent's mapping
V08_2026-10-14.xlsx
This file's column
Schema field
Ref
load_id
Ship Date
date
Customer Name
customer
Lane
lane
Miles Loaded
miles
Miles Empty
empty_miles
Revenue - Linehaul
linehaul_rev
Revenue - Fuel
fuel_surcharge
Revenue - Accessorial
accessorials
Cost - Carrier
carrier_cost
On Time Y/N
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$110,749 = $110,749
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 8 seconds.
The intake agent's mapping
V09_2026-10-15.xlsx
This file's column
Schema field
Load ID
load_id
Ship Date
date
Shipper
customer
Lane
lane
Miles
miles
Deadhead
empty_miles
Linehaul
linehaul_rev
Fuel Index
null
FSC
fuel_surcharge
Accessorials
accessorials
Carrier Cost
carrier_cost
On Time
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$115,951 = $115,951
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 16 seconds.
The intake agent's mapping
V10_2026-10-16.xlsx
This file's column
Schema field
Load Number
load_id
Pickup Date
date
Customer
customer
Route
lane
Loaded Miles
miles
Empty Miles
empty_miles
Rate per Mile
null
Linehaul Revenue
linehaul_rev
Fuel Surcharge
fuel_surcharge
Accessorials
accessorials
Total Revenue
null
Carrier Pay
carrier_cost
OTD
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$102,886 = $102,886
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 9 seconds.
The intake agent's mapping
V11_2026-10-17.xlsx
This file's column
Schema field
Pro #
load_id
Ship Date
date
Bill To
customer
O/D
lane
Trip Miles
miles
Deadhead Miles
empty_miles
LH $
linehaul_rev
Fuel $
fuel_surcharge
Acc $
accessorials
Carrier $
carrier_cost
Late?
on_time
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$109,285 = $109,285
Accepted. Every check passed.
The mapping differs from the answer key in a detail the sample rows didn't show, and it still parses every row the same way.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 8 seconds.
The intake agent's mapping
V12_2026-10-20.xlsx
This file's column
Schema field
LOAD
load_id
DT
date
CUST
customer
LANE
lane
LD MI
miles
MT MI
empty_miles
LH
linehaul_rev
FSC
fuel_surcharge
ACC
accessorials
PT
carrier_cost
OT
on_time
EQUIP
null
DISP
null
What the mapping had to pass
Every row's date is the business day
Every load ID is unique
The totals row equals the rows above it
Miles and charges sit in their normal ranges
Revenue ties to the invoicing batch
$101,394 = $101,394
Accepted. Every check passed.
The mapping matches the answer key exactly.
Mapped by the intake agent in a run recorded on October 27 with Claude Opus 5.5 in 12 seconds.
Eleven files are waiting
This is one demo morning at a fund with three synthetic portfolio companies. Their systems sent eleven overnight files before 6 AM. Run the morning to receive, map, check, and tie them out, or press the right arrow for a guided tour.
A carve-out of a commercial solar operations and maintenance business from a larger electrical contractor
The data room
The deal analyst's assumptions
The agent read three years of history and proposed each operating assumption with a rationale that cites it. Code accepted each value only inside a range the history sets, and the deal team set the price and the debt.
Assumption
FY2026
Proposed
Used
Source
Rationale
Preventive maintenance growth
Agent
Preventive maintenance growth was after and I use a lower rate to avoid extending the recent pace.
Corrective repair growth
Agent
Corrective repairs growth eased from to and I use a lower rate to reflect the recent slowdown.
Module cleaning growth
Agent
Module cleaning growth eased from to and I use a lower rate to avoid projecting the latest result forward.
Monitoring subscription growth
Agent
Monitoring subscriptions growth eased from to and I use a lower rate as growth continues to moderate.
Gross margin
Agent
Gross margin rose from through to and I use a rate below the latest result to avoid extending the annual gains.
SG&A share before standalone cost
Agent
SG&A share declined from through to and I hold near the earlier level instead of extending the gradual improvement.
D&A share of revenue
Agent
D&A share moved from to and returned to so I use the latest level.
Capex share of revenue
Agent
Capital spending share moved from to and remained at so I use the recent level.
NWC share of revenue
Agent
Net working capital share rose from through to and I use a slightly higher rate to reflect the continuing increase.
Proposed by the deal analyst agent from a run recorded on October 28, 2026.
Entry enterprise value
Equity at close
MOIC
IRR
Net leverage at close
Covenant headroom
The model
Ten tabs of live formulas from the data room and the ledger. Blue is an input, black a formula, and green a link between tabs. Click a cell for its formula.
fxClick a cell to see its formula.
A
1
Brightrow Field Services first-draft model
2
3
A first-draft model of a carve-out built from the data room and the fund's ledger.
4
Sources holds the data room's values with the file and cells each came from and they are the only hard-coded history.
5
Historicals and Adjustments rebuild three fiscal years and the EBITDA bridge from Sources with formulas.
6
The deal analyst agent proposed the operating assumptions and each cites the history it rests on.
7
The deal team set the price and the debt terms and Checks turns red if any tie-out breaks.
8
Every company and number here is synthetic.
9
10
How to read the colors
11
Blue cells are inputs
12
Black cells are formulas
13
Green cells link to another tab
14
15
Version one built for the November 4 2026 review
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
X
Y
Z
AA
AB
AC
AD
AE
AF
AG
AH
AI
AJ
AK
AL
1
Data room values
2
Values from the data room with the file and rows they came from. These are the only hard-coded history in the model.
3
4
Line
From
Oct 2023
Nov 2023
Dec 2023
Jan 2024
Feb 2024
Mar 2024
Apr 2024
May 2024
Jun 2024
Jul 2024
Aug 2024
Sep 2024
Oct 2024
Nov 2024
Dec 2024
Jan 2025
Feb 2025
Mar 2025
Apr 2025
May 2025
Jun 2025
Jul 2025
Aug 2025
Sep 2025
Oct 2025
Nov 2025
Dec 2025
Jan 2026
Feb 2026
Mar 2026
Apr 2026
May 2026
Jun 2026
Jul 2026
Aug 2026
Sep 2026
5
Preventive maintenance contracts revenue
financials.xlsx P&L row 6
1,020,151
1,028,290
1,060,390
1,064,073
1,078,895
1,100,226
1,092,232
1,145,805
1,134,043
1,189,739
1,193,759
1,192,396
1,147,926
1,195,103
1,182,062
1,177,509
1,180,584
1,206,158
1,213,170
1,257,548
1,257,734
1,266,574
1,251,476
1,264,157
1,226,750
1,277,317
1,278,106
1,288,043
1,311,339
1,338,557
1,321,884
1,364,045
1,367,831
1,391,780
1,421,493
1,412,855
6
Corrective repairs revenue as reported
financials.xlsx P&L row 7
781,820
740,246
695,591
712,654
785,311
932,224
1,033,872
1,130,532
1,169,193
1,174,523
1,127,087
1,016,948
887,394
809,723
778,255
787,147
857,109
1,005,055
800,681
862,890
856,376
850,216
836,472
819,860
678,152
682,836
645,242
676,836
697,527
820,359
875,904
906,731
939,035
932,405
892,472
852,502
7
Module cleaning revenue
financials.xlsx P&L row 8
361,426
427,126
433,713
416,963
392,631
316,962
259,048
216,970
187,144
183,438
240,675
339,133
412,479
481,087
486,428
480,060
444,158
369,380
8
Monitoring subscriptions revenue
financials.xlsx P&L row 9
163,194
168,659
163,857
174,561
172,156
177,019
174,490
175,555
180,923
181,514
180,505
187,566
210,784
216,839
218,314
222,364
225,552
218,984
231,165
227,314
225,158
226,356
238,155
239,016
262,250
257,399
267,495
260,046
275,857
275,353
279,978
271,839
280,252
290,115
289,338
290,078
9
Total revenue
financials.xlsx P&L row 10
1,965,166
1,937,196
1,919,839
1,951,288
2,036,362
2,209,468
2,300,594
2,451,892
2,484,159
2,545,776
2,501,351
2,396,910
2,246,104
2,221,664
2,178,631
2,187,020
2,263,245
2,430,197
2,606,442
2,774,879
2,772,981
2,760,108
2,718,734
2,639,996
2,426,200
2,434,523
2,377,986
2,408,363
2,525,397
2,773,402
2,890,244
3,023,701
3,073,546
3,094,361
3,047,460
2,924,815
10
Gross profit
financials.xlsx P&L row 28
902,720
930,105
896,410
917,715
926,867
1,007,402
1,038,233
1,104,586
1,136,476
1,104,817
1,095,959
1,116,605
1,056,944
1,058,495
1,060,718
1,026,740
1,064,542
1,108,690
1,163,005
1,256,803
1,286,965
1,236,516
1,272,676
1,173,205
1,142,014
1,170,154
1,124,646
1,132,261
1,183,291
1,272,451
1,360,551
1,381,820
1,420,296
1,450,966
1,363,975
1,366,849
11
Salaries and benefits
financials.xlsx P&L row 31
371,835
364,300
361,055
370,411
383,424
418,345
433,980
458,971
470,555
482,385
468,711
449,990
424,803
421,087
412,746
409,471
429,862
460,354
493,171
523,095
523,993
520,331
509,499
501,316
460,973
467,033
458,490
464,483
485,529
532,342
549,711
581,474
581,951
591,335
577,804
561,658
12
Facilities
financials.xlsx P&L row 32
66,838
68,257
69,071
66,770
68,114
69,640
68,796
68,097
67,817
69,604
69,255
69,440
69,081
72,033
71,326
68,654
71,366
70,642
71,521
70,565
70,757
71,965
71,126
70,832
74,040
73,263
72,057
72,665
70,394
71,794
71,130
73,879
73,125
71,906
72,312
74,126
13
IT and software
financials.xlsx P&L row 33
31,749
32,590
32,370
34,313
35,015
36,492
39,446
40,509
41,855
43,260
43,212
41,610
39,456
36,287
36,371
38,471
37,794
40,673
45,093
46,252
49,278
46,549
48,334
44,073
41,247
41,279
40,640
39,126
42,328
48,448
50,765
50,341
53,502
54,047
51,431
49,648
14
Insurance
financials.xlsx P&L row 34
41,000
41,000
41,000
41,000
41,000
41,000
41,000
41,000
41,000
41,000
41,000
41,000
42,500
42,500
42,500
42,500
42,500
42,500
42,500
42,500
42,500
42,500
42,500
42,500
44,000
44,000
44,000
44,000
44,000
44,000
44,000
44,000
44,000
44,000
44,000
44,000
15
Corporate overhead allocation from the parent
financials.xlsx P&L row 35
115,000
115,000
115,000
115,000
115,000
115,000
115,000
115,000
115,000
115,000
115,000
115,000
120,000
120,000
120,000
120,000
120,000
120,000
120,000
120,000
120,000
120,000
120,000
120,000
123,750
123,750
123,750
123,750
123,750
123,750
123,750
123,750
123,750
123,750
123,750
123,750
16
Legal settlement
financials.xlsx P&L row 37
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
412,500
0
0
0
0
0
0
0
0
0
0
0
0
0
17
Reported EBITDA
financials.xlsx P&L row 38
276,298
308,957
277,914
290,221
284,314
326,924
340,012
381,010
400,248
353,568
358,782
399,565
361,104
366,588
377,775
347,645
363,020
374,522
390,720
454,391
480,438
435,170
68,717
394,485
398,004
420,829
385,709
388,238
417,290
452,117
521,195
508,377
543,968
565,927
494,678
513,666
18
Module cleaning booked inside corrective repairs
financials.xlsx P&L row 41
199,998
162,129
141,193
140,575
186,006
247,087
308,623
368,686
370,039
364,356
336,276
275,031
236,024
184,828
163,080
163,094
212,240
291,913
19
20
Annual summary from the data room
FY2024
FY2025
FY2026
21
Total revenue in the annual summary
financials.xlsx Annual summary
26,700,000
29,800,000
33,000,000
22
Gross profit in the annual summary
financials.xlsx Annual summary
12,177,894
13,765,301
15,369,274
23
Reported EBITDA in the annual summary
financials.xlsx Annual summary
3,997,813
4,414,574
5,610,000
24
25
Standalone corporate cost a month from Notes item four
financials.xlsx Notes item 4
85,250
26
27
Revenue by customer
FY2024
FY2025
FY2026
28
All customers
customers.xlsx
26,700,000
29,800,000
33,000,000
29
The largest customer
customers.xlsx
2,883,600
3,546,200
4,191,000
30
31
Working capital and capital spending
FY2024
FY2025
FY2026
32
Accounts receivable at year end
financials.xlsx Working capital and capex
3,669,481
4,074,144
4,579,274
33
Unbilled services at year end
financials.xlsx Working capital and capex
577,915
627,074
669,173
34
Accounts payable at year end
financials.xlsx Working capital and capex
1,206,091
1,260,214
1,354,915
35
Accrued liabilities at year end
financials.xlsx Working capital and capex
874,569
912,939
965,760
36
Capital expenditures in the year
financials.xlsx Working capital and capex
591,530
691,519
774,513
37
Depreciation and amortization in the year
financials.xlsx Working capital and capex
555,294
558,426
692,178
A
B
C
D
E
1
Three fiscal years ending September 30
2
3
4
FY2024
FY2025
FY2026
5
Preventive maintenance contracts
13,300,000
14,600,000
16,000,000
6
Corrective repairs as reported
11,300,000
10,151,178
9,600,000
7
Module cleaning as reported
0
2,348,822
4,100,000
8
Monitoring subscriptions
2,100,000
2,700,000
3,300,000
9
Total revenue
26,700,000
29,800,000
33,000,000
10
Module cleaning booked inside repairs
3,100,000
1,251,178
0
11
Corrective repairs restated
8,200,000
8,900,000
9,600,000
12
Module cleaning restated
3,100,000
3,600,000
4,100,000
13
Total revenue restated
26,700,000
29,800,000
33,000,000
14
Gross profit
12,177,894
13,765,301
15,369,274
15
Gross margin
45.6%
46.2%
46.6%
16
SG&A before the parent's allocation
6,800,081
7,498,227
8,274,274
17
Overhead allocation from the parent
1,380,000
1,440,000
1,485,000
18
Legal settlement
0
412,500
0
19
Reported EBITDA
3,997,813
4,414,574
5,610,000
20
Reported EBITDA rebuilt from its lines
3,997,813
4,414,574
5,610,000
21
Reported EBITDA margin
15.0%
14.8%
17.0%
22
SG&A share of revenue
25.5%
25.2%
25.1%
23
24
Growth in preventive maintenance
9.8%
9.6%
25
Growth in corrective repairs restated
8.5%
7.9%
26
Growth in module cleaning restated
16.1%
13.9%
27
Growth in monitoring
28.6%
22.2%
28
Net working capital at year end
2,166,736
2,528,066
2,927,773
29
Net working capital share of revenue
8.1%
8.5%
8.9%
30
Capital spending share of revenue
2.2%
2.3%
2.3%
31
Depreciation and amortization share of revenue
2.1%
1.9%
2.1%
A
B
C
D
E
F
1
From reported to adjusted EBITDA
2
3
4
FY2024
FY2025
FY2026
Why
5
Reported EBITDA
3,997,813
4,414,574
5,610,000
6
Add back the legal settlement
0
412,500
0
One-time and resolved in August 2025 as Notes item two says
7
Add back the parent's overhead allocation
1,380,000
1,440,000
1,485,000
The parent's charge ends at the carve-out as Notes item three says
8
Less the standalone corporate cost
(1,023,000)
(1,023,000)
(1,023,000)
Management's estimate of what a standalone Brightrow spends from Notes item four
9
Adjusted EBITDA
4,354,813
5,244,074
6,072,000
10
Adjusted EBITDA margin
16.3%
17.6%
18.4%
A
B
C
D
E
F
1
Inputs
2
3
4
Input
FY2026 history
Value
Source
Rationale
5
Operating assumptions for every projection year
6
Preventive maintenance growth
9.6%
9.0%
Deal analyst agent recorded 2026-10-28
Preventive maintenance growth was 9.6% after 9.8% and I use a lower rate to avoid extending the recent pace.
7
Corrective repair growth
7.9%
7.0%
Deal analyst agent recorded 2026-10-28
Corrective repairs growth eased from 8.5% to 7.9% and I use a lower rate to reflect the recent slowdown.
8
Module cleaning growth
13.9%
12.0%
Deal analyst agent recorded 2026-10-28
Module cleaning growth eased from 16.1% to 13.9% and I use a lower rate to avoid projecting the latest result forward.
9
Monitoring subscription growth
22.2%
18.0%
Deal analyst agent recorded 2026-10-28
Monitoring subscriptions growth eased from 28.6% to 22.2% and I use a lower rate as growth continues to moderate.
10
Gross margin
46.6%
46.0%
Deal analyst agent recorded 2026-10-28
Gross margin rose from 45.6% through 46.2% to 46.6% and I use a rate below the latest result to avoid extending the annual gains.
11
SG&A share before standalone cost
25.1%
25.5%
Deal analyst agent recorded 2026-10-28
SG&A share declined from 25.5% through 25.2% to 25.1% and I hold near the earlier level instead of extending the gradual improvement.
12
D&A share of revenue
2.1%
2.1%
Deal analyst agent recorded 2026-10-28
D&A share moved from 2.1% to 1.9% and returned to 2.1% so I use the latest level.
13
Capex share of revenue
2.3%
2.3%
Deal analyst agent recorded 2026-10-28
Capital spending share moved from 2.2% to 2.3% and remained at 2.3% so I use the recent level.
14
NWC share of revenue
8.9%
9.0%
Deal analyst agent recorded 2026-10-28
Net working capital share rose from 8.1% through 8.5% to 8.9% and I use a slightly higher rate to reflect the continuing increase.
15
16
Deal and debt terms
17
Cash tax rate
25.0%
Deal team
18
Yearly increase in the standalone cost
3.0%
Deal team
19
Entry multiple of FY2026 adjusted EBITDA
8.0x
Deal team
20
Transaction fees as a share of enterprise value
2.5%
Deal team
21
Exit multiple of FY2031 EBITDA
8.0x
Deal team
22
Senior debt at close as a multiple of EBITDA
3.5x
Deal team
23
Interest rate on the term loan
9.5%
Deal team
24
Yearly amortization as a share of the original loan
5.0%
Deal team
25
Share of excess cash swept to the loan
50.0%
Deal team
A
B
C
D
E
F
G
H
1
Base case projections
2
3
4
FY2026A
FY2027E
FY2028E
FY2029E
FY2030E
FY2031E
5
Revenue by service line
6
Preventive maintenance contracts
16,000,000
17,440,000
19,009,600
20,720,464
22,585,306
24,617,983
7
Corrective repairs
9,600,000
10,272,000
10,991,040
11,760,413
12,583,642
13,464,497
8
Module cleaning
4,100,000
4,592,000
5,143,040
5,760,205
6,451,429
7,225,601
9
Monitoring subscriptions
3,300,000
3,894,000
4,594,920
5,422,006
6,397,967
7,549,601
10
Total revenue
33,000,000
36,198,000
39,738,600
43,663,087
48,018,343
52,857,681
11
Revenue growth
9.7%
9.8%
9.9%
10.0%
10.1%
12
Gross profit
15,369,274
16,651,080
18,279,756
20,085,020
22,088,438
24,314,533
13
SG&A before the standalone cost
8,274,274
9,230,490
10,133,343
11,134,087
12,244,678
13,478,709
14
Standalone corporate cost
1,023,000
1,053,690
1,085,301
1,117,860
1,151,396
1,185,937
15
EBITDA
6,072,000
6,366,900
7,061,112
7,833,073
8,692,365
9,649,887
16
EBITDA margin
18.4%
17.6%
17.8%
17.9%
18.1%
18.3%
17
Depreciation and amortization
760,158
834,511
916,925
1,008,385
1,110,011
18
EBIT
5,606,742
6,226,602
6,916,148
7,683,980
8,539,876
19
Interest
2,018,940
1,857,761
1,670,559
1,454,137
1,204,905
20
Cash taxes
896,950
1,092,210
1,311,397
1,557,461
1,833,743
21
Capital expenditures
832,554
913,988
1,004,251
1,104,422
1,215,727
22
Net working capital
2,970,000
3,257,820
3,576,474
3,929,678
4,321,651
4,757,191
23
Change in net working capital
287,820
318,654
353,204
391,973
435,540
24
Cash flow before debt service
4,349,575
4,736,260
5,164,221
5,638,509
6,164,878
A
B
C
D
E
F
G
H
1
Term loan and leverage
2
3
4
At close
FY2027E
FY2028E
FY2029E
FY2030E
FY2031E
5
Opening balance
21,252,000
19,555,382
17,584,833
15,306,702
12,683,215
6
Interest on the opening balance
2,018,940
1,857,761
1,670,559
1,454,137
1,204,905
7
Mandatory amortization
1,062,600
1,062,600
1,062,600
1,062,600
1,062,600
8
Cash after interest and amortization
1,268,036
1,815,899
2,431,062
3,121,773
3,897,372
9
Cash sweep
634,018
907,950
1,215,531
1,560,886
1,948,686
10
Closing balance
21,252,000
19,555,382
17,584,833
15,306,702
12,683,215
9,671,929
11
Cash balance
0
634,018
1,541,967
2,757,498
4,318,384
6,267,071
12
Net debt
21,252,000
18,921,364
16,042,865
12,549,203
8,364,831
3,404,859
13
EBITDA
6,072,000
6,366,900
7,061,112
7,833,073
8,692,365
9,649,887
14
Net leverage
3.50x
2.97x
2.27x
1.60x
0.96x
0.35x
15
Interest coverage
3.15x
3.80x
4.69x
5.98x
8.01x
16
Illustrative leverage covenant
4.50x
4.25x
4.00x
3.75x
3.50x
17
Headroom to the leverage covenant
1.53x
1.98x
2.40x
2.79x
3.15x
18
Illustrative coverage covenant
2.50x
2.50x
2.50x
2.50x
2.50x
19
Headroom to the coverage covenant
0.65x
1.30x
2.19x
3.48x
5.51x
20
21
The covenants are illustrative for a first look and come from no term sheet.
A
B
C
D
E
F
G
H
1
Returns in the base case
2
3
4
FY2026 adjusted EBITDA
6,072,000
5
Entry multiple
8.0x
6
Entry enterprise value
48,576,000
7
Transaction fees
1,214,400
8
Total uses
49,790,400
9
Senior debt at close
21,252,000
10
Equity at close
28,538,400
11
FY2031 EBITDA
9,649,887
12
Exit multiple
8.0x
13
Exit enterprise value
77,199,098
14
Net debt at exit
3,404,859
15
Equity at exit
73,794,239
16
MOIC
2.59x
17
18
Equity cash flows by year
At close
FY2027E
FY2028E
FY2029E
FY2030E
FY2031E
19
Equity in and out
(28,538,400)
0
0
0
0
73,794,239
20
IRR
20.9%
21
IRR from MOIC to the one-fifth less one
20.9%
A
B
C
D
E
F
G
1
MOIC by entry and exit multiple
2
3
Entry multiples run down the side and exit multiples run across the top.
4
MOIC by entry and exit multiple
5
7.0x
7.5x
8.0x
8.5x
9.0x
6
7.0x
2.87x
3.09x
3.31x
3.52x
3.74x
7
7.5x
2.52x
2.71x
2.90x
3.09x
3.28x
8
8.0x
2.25x
2.42x
2.59x
2.75x
2.92x
9
8.5x
2.03x
2.18x
2.33x
2.48x
2.64x
10
9.0x
1.85x
1.98x
2.12x
2.26x
2.40x
11
12
IRR by entry and exit multiple
13
7.0x
7.5x
8.0x
8.5x
9.0x
14
7.0x
23.5%
25.3%
27.0%
28.6%
30.2%
15
7.5x
20.3%
22.1%
23.8%
25.3%
26.8%
16
8.0x
17.6%
19.3%
20.9%
22.5%
23.9%
17
8.5x
15.2%
16.9%
18.4%
20.0%
21.4%
18
9.0x
13.0%
14.7%
16.2%
17.7%
19.1%
A
B
C
D
E
F
1
Checks
2
3
Every check together
OK
4
What it checks
Model
Against
Difference
Status
5
FY2026 revenue equals the data room's annual summary
33,000,000
33,000,000
0
OK
6
Restated revenue equals reported revenue in FY2026
33,000,000
33,000,000
0
OK
7
Reported EBITDA rebuilt from its lines equals the data room's
5,610,000
5,610,000
0
OK
8
Adjusted EBITDA equals the fund's ledger
6,072,000
6,072,000
0
OK
9
Revenue by customer adds up to FY2026 revenue
33,000,000
33,000,000
0
OK
10
Sources equal uses at close
49,790,400
49,790,400
0
OK
11
The loan rolls forward every year
0
0
0
OK
12
The loan never goes negative
9,671,929
0
9,671,929
OK
13
MOIC equals the Python engine
2.59x
2.59x
-0.000000
OK
14
IRR equals the Python engine
20.9%
20.9%
0.000000
OK
15
IRR equals MOIC to the one-fifth less one
20.9%
20.9%
0.000000
OK
The teaser's claims against the ledger
The seller's teaser makes three claims with numbers in them. Each one is checked against the ledger and cited.
“Recurring, contracted revenue. Preventive maintenance contracts and monitoring subscriptions provide a predictable base, and contract renewal rates have stayed above 90% through the last twelve months.”
Supported. The teaser puts renewals above and the ledger's FY2026 renewal rate of supports it.
“Attractive margins. Brightrow has delivered EBITDA margins above 20% over the last twelve months, supported by route density and disciplined pricing.”
Contradicted. The teaser puts the margin above and the ledger's FY2026 adjusted EBITDA margin is . The teaser adds back the parent's overhead without the standalone cost that replaces it, and that basis comes to .
“Diversified customer base. No customer accounts for more than 10% of revenue, and the Company serves commercial owners, community solar funds and school districts across Texas and Oklahoma.”
Contradicted. The teaser caps every customer at of revenue. The largest customer brought in in FY2026 and that is of revenue.
First draft of the investment committee memo
Brightrow Field Services
Drafted by the memo agent from recorded run 71e4c32d and tied out by the gate. It's a first draft for a person to review.
This is a test. It types one figure into the projections the way a careless edit would.
Tied out by the gate. All 85 figures in the agent's text came from placeholders and none was typed.
The gate failed the memo. The figure $61.2 million in section 7 was typed by hand and no placeholder supplied it.
1. The recommendation
We recommend that the committee approve pursuit of Brightrow Field Services into the next round on the conditions listed at the end of this memo. The base case values the business at at an entry multiple of on adjusted EBITDA of and returns a MOIC of and an IRR of over five years.
The case rests on adjusted EBITDA that is lower than the seller presents. It also rests on a customer base that is more concentrated than the teaser claims. Our support depends on a quality of earnings review that confirms the standalone cost and on contract diligence for the largest customers.
Conditions for the next round
A quality of earnings review confirms adjusted EBITDA of with a supported standalone corporate cost near .
Contract review of the largest customer and the ten largest finds no change of control rights or near term expiries that put revenue at risk at close.
The bid stays at or near the entry multiple of because much of the return depends on entry price.
Separation costs and transition services from the parent are scoped and priced before the second round.
2. The business and the carve-out
Brightrow provides operations and maintenance services for commercial solar assets through preventive maintenance contracts, corrective repairs, module cleaning, and monitoring subscriptions. In September 2026 the business had under contract and field technicians, with an average response time of . The seller describes its customers as commercial owners, community solar funds, and school districts in Texas and Oklahoma. We have not yet verified that mix.
Today the business sits inside a larger electrical contractor that charges it an overhead allocation of a year. The separation removes that charge and replaces it with standalone corporate functions estimated at a year. The input does not cover the cost of building those functions or the terms of any transition services from the parent.
3. Why now
The parent is selling a business that sits outside its core contracting work, and the first round of bids is expected in early 2027. A separation of this kind requires a buyer that can stand up corporate functions. That requirement may limit the field of bidders.
Brightrow grew contracted capacity from in October 2023 to in September 2026 and renewed of its contracts in FY2026. These figures point to a growing installed base that needs ongoing service. The data does not show the pipeline of new contracts or the pace of new solar installations in the region.
4. Historical financials
Revenue rose from in FY2024 to in FY2025 and in FY2026. Gross margin was in FY2026, but the input does not give gross or EBITDA margins for the earlier years. We cannot yet say how margins have moved over the period.
Preventive maintenance contracts are the largest line at in FY2026 and grew . Corrective repairs contributed with growth of . Module cleaning reached with growth of and monitoring subscriptions reached with growth of . The segment figures are restated and we need the basis for that restatement before relying on them.
Fiscal years ending September 30
FY2024
FY2025
FY2026
Preventive maintenance contracts
Corrective repairs restated
Module cleaning restated
Monitoring subscriptions
Total revenue
Gross margin
Reported EBITDA
Adjusted EBITDA
Adjusted EBITDA margin
5. EBITDA adjustments
Reported EBITDA for FY2026 was at a margin of . We add back the parent overhead allocation of and deduct the standalone corporate cost of that replaces it. That gives adjusted EBITDA of at a margin of .
The seller's margin of adds back the allocation without any standalone cost. We do not accept that basis because the business will carry its own corporate functions once it is separated. The standalone cost estimate is the most sensitive item in the bridge and needs independent support.
The legal settlement of falls in FY2025 and does not enter the FY2026 bridge. It would matter only if the committee looks at a multiyear average. Diligence should confirm that no similar matters are pending.
From reported to adjusted
FY2024
FY2025
FY2026
Reported EBITDA
Add back the legal settlement
Add back the parent's overhead allocation
Less the standalone corporate cost
Adjusted EBITDA
6. Customers and concentration
The largest customer generated in FY2026 and accounted for of revenue. The ten largest customers together accounted for . This is more concentrated than the seller describes.
The loss of the largest customer would cut revenue by and would test leverage in the early years of the hold. The input does not show contract terms, renewal dates, or change of control provisions for these customers, and we need them before a second round bid.
Customers in FY2026
Revenue
Share of revenue
The largest customer
The ten largest customers
7. Base-case projections
The base case takes revenue to in FY2031 at annual growth of and EBITDA to at a margin of . That margin is close to the FY2026 adjusted margin of and the case does not depend on margin expansion.
Growth by line is set at for preventive maintenance, for corrective repairs, for module cleaning, and for monitoring subscriptions. Each rate is lower than the FY2026 growth for that line and so builds in some slowing from recent performance.
Gross margin is held at against in FY2026. SG&A is set at of revenue before the standalone corporate cost, and depreciation and amortization runs at of revenue. Capital spending is of revenue and net working capital is of revenue.
Fiscal year
FY2026A
FY2027E
FY2028E
FY2029E
FY2030E
FY2031E
Revenue
EBITDA
8. Returns and sensitivity
At close the deal funds an enterprise value of and fees of with senior debt of and equity of . At an exit multiple of the FY2031 enterprise value is and the equity is worth after net debt of . That yields a MOIC of and an IRR of .
Paying the highest entry multiple in the grid and exiting at the lowest gives a MOIC of and an IRR of . The lowest entry with the highest exit gives and an IRR of . The lowest entry with the lowest exit still gives . Entry price therefore carries much of the return and the grid argues for discipline in the bid.
Base case
Value
Entry enterprise value
Equity at close
Equity at exit
MOIC
IRR
MOIC by entry multiple down the side and exit multiple across the top
Entry and exit
7.0x
7.5x
8.0x
8.5x
9.0x
7.0x
7.5x
8.0x
8.5x
9.0x
9. Debt capacity
Senior debt at close is on a term loan priced at . That sets net leverage at against a sizing target of of adjusted EBITDA. Interest coverage in FY2027 is .
Net leverage falls to in FY2027 and in FY2031 as the business repays over five years. The smallest headroom to the illustrative leverage covenant is . The covenants are illustrative and we do not yet have lender terms.
Fiscal year
FY2027E
FY2028E
FY2029E
FY2030E
FY2031E
Net leverage
Headroom to the illustrative covenant
10. Key risks and diligence questions
The teaser's claim of recurring contracted revenue with renewal rates above its stated threshold is , and the FY2026 renewal rate was .
The claim of EBITDA margins above the teaser's threshold is . Reported margin was and adjusted margin was . Only the seller's basis of clears the threshold because it adds back the parent allocation without any standalone cost.
The claim that no single customer passes the teaser's threshold is . The largest customer accounted for of FY2026 revenue and the ten largest accounted for . We have not verified the customer types or the Texas and Oklahoma footprint that the teaser describes.
Other risks include the cost and timing of separation, the terms of transition services, and whether the parent refers work to Brightrow that would stop after the sale. Diligence should test the standalone cost, the restated segment figures, labor costs for the technician base, and the margin history before FY2026.
The teaser's claim
Result
The ledger
Recurring, contracted revenue. Preventive maintenance contracts and monitoring subscriptions provide a predictable base, and contract renewal rates have stayed above 90% through the last twelve months.
Attractive margins. Brightrow has delivered EBITDA margins above 20% over the last twelve months, supported by route density and disciplined pricing.
Diversified customer base. No customer accounts for more than 10% of revenue, and the Company serves commercial owners, community solar funds and school districts across Texas and Oklahoma.
11. Next steps
The deal team will request data room access to contracts with the ten largest customers, the parent allocation method, and the basis for the restated segment figures. The deal team will also engage an accounting adviser for a quality of earnings review focused on the standalone cost and the margin history.
The deal team will open early talks with lenders to test the debt package and covenant terms. It will return to the committee with an updated model and a proposed bid range ahead of first round bids in early 2027.
1. The recommendation
We recommend that the committee approve pursuit of Brightrow Field Services into the next round on the conditions listed at the end of this memo. The base case values the business at at an entry multiple of on adjusted EBITDA of and returns a MOIC of and an IRR of over five years.
The case rests on adjusted EBITDA that is lower than the seller presents. It also rests on a customer base that is more concentrated than the teaser claims. Our support depends on a quality of earnings review that confirms the standalone cost and on contract diligence for the largest customers.
Conditions for the next round
A quality of earnings review confirms adjusted EBITDA of with a supported standalone corporate cost near .
Contract review of the largest customer and the ten largest finds no change of control rights or near term expiries that put revenue at risk at close.
The bid stays at or near the entry multiple of because much of the return depends on entry price.
Separation costs and transition services from the parent are scoped and priced before the second round.
2. The business and the carve-out
Brightrow provides operations and maintenance services for commercial solar assets through preventive maintenance contracts, corrective repairs, module cleaning, and monitoring subscriptions. In September 2026 the business had under contract and field technicians, with an average response time of . The seller describes its customers as commercial owners, community solar funds, and school districts in Texas and Oklahoma. We have not yet verified that mix.
Today the business sits inside a larger electrical contractor that charges it an overhead allocation of a year. The separation removes that charge and replaces it with standalone corporate functions estimated at a year. The input does not cover the cost of building those functions or the terms of any transition services from the parent.
3. Why now
The parent is selling a business that sits outside its core contracting work, and the first round of bids is expected in early 2027. A separation of this kind requires a buyer that can stand up corporate functions. That requirement may limit the field of bidders.
Brightrow grew contracted capacity from in October 2023 to in September 2026 and renewed of its contracts in FY2026. These figures point to a growing installed base that needs ongoing service. The data does not show the pipeline of new contracts or the pace of new solar installations in the region.
4. Historical financials
Revenue rose from in FY2024 to in FY2025 and in FY2026. Gross margin was in FY2026, but the input does not give gross or EBITDA margins for the earlier years. We cannot yet say how margins have moved over the period.
Preventive maintenance contracts are the largest line at in FY2026 and grew . Corrective repairs contributed with growth of . Module cleaning reached with growth of and monitoring subscriptions reached with growth of . The segment figures are restated and we need the basis for that restatement before relying on them.
Fiscal years ending September 30
FY2024
FY2025
FY2026
Preventive maintenance contracts
Corrective repairs restated
Module cleaning restated
Monitoring subscriptions
Total revenue
Gross margin
Reported EBITDA
Adjusted EBITDA
Adjusted EBITDA margin
5. EBITDA adjustments
Reported EBITDA for FY2026 was at a margin of . We add back the parent overhead allocation of and deduct the standalone corporate cost of that replaces it. That gives adjusted EBITDA of at a margin of .
The seller's margin of adds back the allocation without any standalone cost. We do not accept that basis because the business will carry its own corporate functions once it is separated. The standalone cost estimate is the most sensitive item in the bridge and needs independent support.
The legal settlement of falls in FY2025 and does not enter the FY2026 bridge. It would matter only if the committee looks at a multiyear average. Diligence should confirm that no similar matters are pending.
From reported to adjusted
FY2024
FY2025
FY2026
Reported EBITDA
Add back the legal settlement
Add back the parent's overhead allocation
Less the standalone corporate cost
Adjusted EBITDA
6. Customers and concentration
The largest customer generated in FY2026 and accounted for of revenue. The ten largest customers together accounted for . This is more concentrated than the seller describes.
The loss of the largest customer would cut revenue by and would test leverage in the early years of the hold. The input does not show contract terms, renewal dates, or change of control provisions for these customers, and we need them before a second round bid.
Customers in FY2026
Revenue
Share of revenue
The largest customer
The ten largest customers
7. Base-case projections
The base case takes revenue to in FY2031 at annual growth of and EBITDA to at a margin of . That margin is close to the FY2026 adjusted margin of and the case does not depend on margin expansion. Revenue reaches $61.2 million by FY2031.
Growth by line is set at for preventive maintenance, for corrective repairs, for module cleaning, and for monitoring subscriptions. Each rate is lower than the FY2026 growth for that line and so builds in some slowing from recent performance.
Gross margin is held at against in FY2026. SG&A is set at of revenue before the standalone corporate cost, and depreciation and amortization runs at of revenue. Capital spending is of revenue and net working capital is of revenue.
Fiscal year
FY2026A
FY2027E
FY2028E
FY2029E
FY2030E
FY2031E
Revenue
EBITDA
8. Returns and sensitivity
At close the deal funds an enterprise value of and fees of with senior debt of and equity of . At an exit multiple of the FY2031 enterprise value is and the equity is worth after net debt of . That yields a MOIC of and an IRR of .
Paying the highest entry multiple in the grid and exiting at the lowest gives a MOIC of and an IRR of . The lowest entry with the highest exit gives and an IRR of . The lowest entry with the lowest exit still gives . Entry price therefore carries much of the return and the grid argues for discipline in the bid.
Base case
Value
Entry enterprise value
Equity at close
Equity at exit
MOIC
IRR
MOIC by entry multiple down the side and exit multiple across the top
Entry and exit
7.0x
7.5x
8.0x
8.5x
9.0x
7.0x
7.5x
8.0x
8.5x
9.0x
9. Debt capacity
Senior debt at close is on a term loan priced at . That sets net leverage at against a sizing target of of adjusted EBITDA. Interest coverage in FY2027 is .
Net leverage falls to in FY2027 and in FY2031 as the business repays over five years. The smallest headroom to the illustrative leverage covenant is . The covenants are illustrative and we do not yet have lender terms.
Fiscal year
FY2027E
FY2028E
FY2029E
FY2030E
FY2031E
Net leverage
Headroom to the illustrative covenant
10. Key risks and diligence questions
The teaser's claim of recurring contracted revenue with renewal rates above its stated threshold is , and the FY2026 renewal rate was .
The claim of EBITDA margins above the teaser's threshold is . Reported margin was and adjusted margin was . Only the seller's basis of clears the threshold because it adds back the parent allocation without any standalone cost.
The claim that no single customer passes the teaser's threshold is . The largest customer accounted for of FY2026 revenue and the ten largest accounted for . We have not verified the customer types or the Texas and Oklahoma footprint that the teaser describes.
Other risks include the cost and timing of separation, the terms of transition services, and whether the parent refers work to Brightrow that would stop after the sale. Diligence should test the standalone cost, the restated segment figures, labor costs for the technician base, and the margin history before FY2026.
The teaser's claim
Result
The ledger
Recurring, contracted revenue. Preventive maintenance contracts and monitoring subscriptions provide a predictable base, and contract renewal rates have stayed above 90% through the last twelve months.
Attractive margins. Brightrow has delivered EBITDA margins above 20% over the last twelve months, supported by route density and disciplined pricing.
Diversified customer base. No customer accounts for more than 10% of revenue, and the Company serves commercial owners, community solar funds and school districts across Texas and Oklahoma.
11. Next steps
The deal team will request data room access to contracts with the ten largest customers, the parent allocation method, and the basis for the restated segment figures. The deal team will also engage an accounting adviser for a quality of earnings review focused on the standalone cost and the margin history.
The deal team will open early talks with lenders to test the debt package and covenant terms. It will return to the committee with an updated model and a proposed bid range ahead of first round bids in early 2027.
Kiskadee Freight
Board meeting for October 2026
Headlines and bullets drafted by the board deck agent from a run recorded on October 29, 2026. The charts come from the ledger.
Kiskadee FreightBoard meeting for October 20261 / 5
October finished better than budget on loads, revenue, and the EBITDA estimate.
Loads totaled and finished .
Revenue was and finished .
Gross margin reached or of revenue, and the EBITDA estimate was at .
Loads
Revenue
Gross margin
Gross margin share
EBITDA estimate
On time
Kiskadee FreightBoard meeting for October 20262 / 5
Gross margin finished as gains from loads and revenue per load outweighed the effect of costs.
Gross margin moved from a budget of to an actual of .
Loads added , and revenue per load added .
The effect of costs was .
Gross margin from budget to actual
Budget margin
Volume
Rate
Cost
Actual margin
Each step is a ledger value and the steps add up to the actual margin.
Table view
Gross margin from budget to actual
Step
Amount
Budget margin
Volume
Rate
Cost
Actual margin
Kiskadee FreightBoard meeting for October 20263 / 5
Operations delivered of loads on time in October with an empty miles share of .
On-time delivery was for the month.
The empty miles share was .
Loads totaled , and revenue per load was .
On-time share by week
Oct 1 to 7Oct 8 to 14Oct 15 to 21Oct 22 to 28
The partial last week of October appears only in the table view.
Table view
On-time share by week
Week
Value
Oct 1 to 7
Oct 8 to 14
Oct 15 to 21
Oct 22 to 28
Oct 29 to 31 partial
Kiskadee FreightBoard meeting for October 20264 / 5
Cash collected of exceeded invoicing of in October.
The company invoiced and collected .
Receivables were at month end, and days sales outstanding was .
Invoiced and collected by week
Oct 1 to 7Oct 8 to 14Oct 15 to 21Oct 22 to 28
Invoiced
Collected
The partial last week of October appears only in the table view.
Table view
Invoiced and collected by week
Week
Invoiced
Collected
Oct 1 to 7
Oct 8 to 14
Oct 15 to 21
Oct 22 to 28
Oct 29 to 31 partial
Kiskadee FreightBoard meeting for October 20265 / 5
The board may decide whether to set targets for on-time delivery, empty miles, and days sales outstanding.
On-time delivery was , and the empty miles share was .
Days sales outstanding was .
Gross margin finished .
Hours back
Every workflow the desk runs is timed against the hours it takes by hand. Run times are measured, and the hours by hand and the review minutes are estimates you can change in the table.
Hours back this week
25.8
Hours by hand less review and run time
Hours back today
3.6
The morning report and the new layout
Hours by hand this week
30.0
What the same runs take when they're built by hand
Hours of review kept
4.2
A person still reviews every draft
Workflow
Runs this week
Measured run time
Estimated hours by hand
Estimated review minutes
Hours back
The daily portfolio report
3
17 seconds
6.7
A new file layout mapped
1
10 seconds
1.3
The deal model's first draft
1
14 seconds
8.5
The memo's first draft
1
72 seconds
5.0
The board deck
1
61 seconds
4.2
This week
25.8
Hours back is the runs times the hours by hand less the review and the run time. Blue marks an estimate as it does in the deal model, and totals update as you change them.
Hours by hand against hours with the desk
The daily portfolio report
6.7 h back
A new file layout mapped
1.3 h back
The deal model's first draft
8.5 h back
The memo's first draft
5.0 h back
The board deck
4.2 h back
By hand
With the desk and a review
Each pair compares the hours a workflow's runs take by hand with the hours they take with the desk and a review.
How it works
One ledger sits under every deliverable. Agents do the reading and the drafting, code checks every number, and nothing reaches a person without passing the gate.
The architecture
Portfolio files
Eleven files a morning
The data room
The seller's financials, customers, and teaser
Intake
Parsers for known layouts and an agent for new ones
Checks
Ranges, keys, and tie-outs
The ledger
Every file, fact, and derived value with its lineage
1,065 files, 7,555 facts, and 1,902 derived values
The gate
No typed numbers
Daily report
Commentary by an agent
Deal model
Assumptions by an agent
Investment memo
Drafted by an agent
Board deck
Headlines by an agent
The spark marks a step where an agent works, and code bounds every agent.
The agents and their limits
Each agent has one job, a fixed input, and an output schema, and code enforces its limits. None of them can publish anything on its own.
Agent
What it sees
What it returns
Limits enforced in code
Intake mapper
A new file's header and sample rows
Mappings and formats
Accepted only when every check and the control total pass, and saved as a layout after that
Daily commentary
Each company's KPIs, variances, and checks
Two to four sentences each
Only placeholders get past the gate
Deal analyst
Three years of data room history
Cited operating assumptions
Each value must sit in a band set by history, and manual overrides win
Memo writer
The model's outputs, the adjustments, the customers, and the claims check
Eleven cited sections
The gate checks every figure, and the risks section must cite every claim
Board deck writer
Kiskadee's month and its variances
Five headlines with two or three bullets each
Charts come only from the ledger
Recorded once
The morning's five agents ran on the models routing picked for them, and every reply passed the same validation before it was saved.
In the recorded runs a morning's five agent calls came to $0.52 at API prices.
Replayed here
The page replays those runs and checks each one again on every build. Nothing on it calls a model or waits on the network, and a visit costs nothing. The invented company names changed after the runs were recorded, and the swap touched nothing else in them.
Model routing
Every model here took the same tests with the same prompts and the same checks. Each agent runs on the cheapest model that clears its quality bar, and the next one that clears it takes over when a call fails. The daily commentary, the board deck writer, and the memo writer stay on Claude Opus 5.5 because passing validation isn't the same as reading well.
The blended morning
$0.51
27% below Claude Opus 5.5 on every agent
Every agent on Claude Opus 5.5
$0.70
The same five agent calls on one model
Intake on its route
12 of 12
Layouts mapped exactly by Claude Haiku 5.5
How the blend adds up
Model
Runs
Model time a morning
Cost a morning
Claude Opus 5.5
Commentary, board deck, and memo
2.4 min
$0.51
Claude Haiku 5.5
Intake
12 s
$0.0030
GPT-6 Luna
Deal analyst
18 s
$0.0013
The blended morning
2.9 min
$0.51
Claude Opus 5.5 on every agent
2.8 min
$0.70
Claude Haiku 5.5 on every agent
7.3 min
$0.065
Intake quality against cost
Layouts mapped exactly
89101112
Within one of the best
Claude Haiku 5.5Gemini 3.1 ProClaude Opus 5.5Gemini 3.8 FlashGPT-6 AstraClaude Sonnet 5.5GPT-6 Luna
$0.001$0.01$0.10
Cost of one mapping on a log scale
Anthropic
OpenAI
Google
The ringed dot is the intake agent's model
2 of the 7 models cleared the intake bar, and Claude Haiku 5.5 did it for the least at $0.0030 a mapping.
Table view
Intake quality against cost
Model
Vendor
Intake exact
Cost of one mapping
GPT-6 Luna
OpenAI
9 of 12
$0.00055
Claude Haiku 5.5
Anthropic
12 of 12
$0.0030
Gemini 3.8 Flash
Google
10 of 12
$0.010
Gemini 3.1 Pro
Google
12 of 12
$0.030
Claude Sonnet 5.5
Anthropic
9 of 12
$0.041
GPT-6 Astra
OpenAI
9 of 12
$0.043
Claude Opus 5.5
Anthropic
10 of 12
$0.075
Where each agent runs
Agent
Runs on
Why
If a call fails
Intake mapper
Claude Haiku 5.5
The cheapest of the 2 models that cleared its bar
Gemini 3.1 Pro
Daily commentary
Claude Opus 5.5
Held for writing quality because a person reads it
GPT-6 Luna and 5 more
Deal analyst
GPT-6 Luna
The cheapest of the 7 models that cleared its bar
Claude Haiku 5.5 and 5 more
Board deck writer
Claude Opus 5.5
Held for writing quality because a person reads it
GPT-6 Luna and 5 more
Memo writer
Claude Opus 5.5
Held for writing quality because a person reads it
Claude Haiku 5.5 and 4 more
How each model did
Model
Vendor
Intake exact
Prose valid first try
Model time a morning
Cost a morning
Routed agents
Gemini 3.1 Progemini-3.1-pro-preview
Google
12 of 12
4 of 4
3.6 min
$0.52
None
Claude Haiku 5.5claude-haiku-5-5
Anthropic
12 of 12
3 of 4
7.3 min
$0.065
Intake
Gemini 3.8 Flashgemini-3.8-flash
Google
10 of 12
4 of 4
5.8 min
$0.20
None
Claude Opus 5.5claude-opus-5-5
Anthropic
10 of 12
4 of 4
2.8 min
$0.70
Commentary, board deck, and memo
GPT-6 Lunagpt-6-luna
OpenAI
9 of 12
3 of 4
2.3 min
$0.0086
Deal analyst
GPT-6 Astragpt-6-astra
OpenAI
9 of 12
3 of 4
2.1 min
$0.55
None
Claude Sonnet 5.5claude-sonnet-5-5
Anthropic
9 of 12
2 of 4
4.7 min
$0.65
None
The intake agent has to land within one layout of the best model's score, and none of its misses can be a mapping the checks would have let through. Each prose agent has to pass validation on every run within the one repair. The comparison ran on October 28. Costs use each vendor's list prices that day, and Claude's come from what the CLI reported.
Code excerpts
These excerpts of the desk's code are copies made for this page. The page reads only the copies and never the code that runs the desk. Open one below or click a box in the diagram above to see its excerpt.
Results from the tests and evals
Each result here is read from a file the tests, the evals, or the recordings wrote. None of them is typed.
Planted errors caught
18 of 18
On held-out seeds 3119, 6427, and 8831 after the checks were frozen
False alarms
0
Across the same held-out seeds
Look-alikes left alone
9 of 9
Normal days that only look like errors
New layouts mapped exactly
10 of 12
Held-out layouts the agent had never seen, and the other 2 still parsed every row like the answer key
Tests passing
148 of 148
In the last full run on November 20
Workbook against Python
74 values
Each formula's value matches the Python model to the cent
LibreOffice recalculation
383 formulas
Recalculated from scratch, and each one matched to the cent
Agent runs recorded
17
With Claude Haiku 5.5, Claude Opus 5.5, and GPT-6 Luna from October 27 to 29
How I'd roll it out at a fund
The first thirty days
One portfolio company's daily report on its real files with every tie-out in place.
The first sixty days
The model and memo template on the fund's real deal process with the gate on every figure.