Cubyts-AI-Cost-Estimate-Workbook1.xlsx

	A	B	C	D	E
1	CUBYTS - AI COST ESTIMATE
2	Modelled AI cost for a stated engagement profile. All figures are median (p50) estimates.
3	Every input is listed below. Change an input and the cost changes; nothing is hidden.
4
5	1. GIVEN VALUES
6	Supplied by the customer.
7	Initial story backlog	50	stories
8	New stories per month	50	stories		Growing 50 -> 100 -> 200+
9	Confluence pages	30	pages		Technical design, architecture, HLD, LLD
10	SharePoint / Teams folders	6	folders		5-7 stated
11	SharePoint / Teams total size	100	MB
12	Engagement duration	6	months
13
14	2. ASSUMED VALUES
15	Not supplied. Each one changes the cost. Confirm before this estimate is relied on.
16	Engineers	8	people		Confirm team size
17	Sprint length	2	weeks		Confirm cadence
18	Repositories	3	repos		Confirm repo count
19	Tracked branches per repository	2	branches		Each branch is indexed separately
20	Pull requests per sprint	30	PRs		Largest single driver of cost
21	Files changed per pull request	4	files		Cost is per file, not per PR
22	Pushes per pull request	2	pushes		Each push re-reviews the changed files
23	Commits per sprint, all branches	450	commits		Includes branches that never open a PR
24	Edits per new story per month	3	edits		Available from the Jira changelog
25	Existing stories edited per month	30%	of backlog		Available from the Jira changelog
26	Stories long enough to analyse	70%	of edits		Stories under 50 words are not analysed at all
27	Confluence pages revised per month	30%	of pages
28	Pages of text in SharePoint / Teams	6,000	pages		Most of the 100 MB is images and media, not text
29
30	3. RECURRING COST, PER SPRINT
31	Capability	Google Cloud	Microsoft Azure	AWS
32	Code review	$122.40	$68.66	$121.97	Every changed file, on every push
33	PR documentation, dependency risk & drift checks	$186.90	$143.10	$186.90	Four capabilities, all on by default
34	Engineering agents	$56.00	$40.00	$56.00	Sprint readiness, PR comments, triage
35	Code indexing	$36.00	$32.00	$36.00	Runs on every commit, any branch
36	Requirement drift analysis	$4.45	$2.75	$4.45	Only on stories that changed
37	Work-item processing	$0.15	$0.11	$0.15
38	Document processing	$0.10	$0.08	$0.10	Ongoing revisions only
39	Platform services	$18.00	$14.00	$18.00	Linking, metadata, rules
40	Total per sprint	$424.00	$300.70	$423.57
41
42	4. ONE-OFF SET-UP COST
43	Repository onboarding	$79.80	$74.28	$79.20	Per repository, per tracked branch
44	Story backlog ingestion	$0.48	$0.36	$0.48
45	Confluence ingestion	$1.19	$0.90	$1.19
46	SharePoint / Teams ingestion	$29.70	$22.50	$29.70	Scales with connected content, not with team activity
47	Total one-off	$111.17	$98.04	$110.57
48
49	5. TOTAL FOR THE ENGAGEMENT
50	Sprints in the engagement	13	13	13	6 months at 2-week sprints
51	TOTAL, 6 MONTHS	$5,623	$4,007	$5,617
52	Per engineer, per month	$117.15	$83.48	$117.02
53
54	6. HOW TO READ THIS
55	- Accuracy is roughly plus or minus 50 per cent. Prices and call counts are confirmed against production usage. Token volumes scale with repository, file and commit patch sizes, so the figures move up or down with those.
56	- Median values throughout. No maximum is presented as a baseline.
57	- Google Cloud and AWS are within one per cent of each other. Azure is lower because it uses a single model tier where the others use two.
58	- There is no fixed monthly infrastructure charge on any of the three providers. The cost is usage only.
59	- Story and document volumes are a very small part of the total. Pull request volume and team activity drive almost all of it.
60	- SharePoint and Teams ingestion scales with whatever content is connected, not with engineering activity. Agree the folder scope up front.
61	- Costs shown are the AI model charges only. Hosting, storage and database costs are covered separately.
62
63	Prepared 20 August 2026. Confidential - Cubyts & Deloitte only.
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100