Every number below is computed from the 2754 cases in this directory when the site was built (2026-11-14). It describes what I have collected, not the whole world, and my collection is biased toward things people wrote about publicly. Read it as a sighting log, not a census.
The headline numbers
2754cases
19topics
209named agents
130source types
1637 cases (59%) come from vendor case studies, press releases or news about vendor pilots. Their results are the vendor's own, and each case says so.
Cases by topic
Show as a table
Topic
Cases
Sales & Marketing
613
Customer Support
554
Productivity
225
Software Development
221
Research & Data
201
Finance
156
DevOps & Infra
130
Email & Inbox
112
Business & Startups
90
Shopping
75
Travel & Booking
72
Media & Content
70
Automation
65
Knowledge & Notes
60
Personal Assistant
57
Home & Family
25
Design & Creative
22
Developer Tools
3
Personal Files
3
Most-named agents
Show as a table
Agent
Cases
Custom agent
266
Claude Code
255
Agentforce
219
Claude
213
ChatGPT
145
Microsoft 365 Copilot
122
Copilot Studio
116
Grok Bot
86
Amazon Bedrock
81
Harvey
52
ChatGPT Agent
49
GitHub Copilot
47
Claude Cowork
42
OpenClaw
41
Microsoft Copilot
39
Tools and builders mentioned
Counts include mentions in case text, so they can overlap. See the tools directory.
Show as a table
Tool
Cases
Claude Code
258
ChatGPT
230
Salesforce Agentforce
227
Microsoft Copilot Studio
139
Amazon Bedrock AgentCore
96
Grok Bot
86
GitHub Copilot
47
OpenClaw
44
n8n
43
Claude Cowork
42
Gemini
36
Meta Muse
35
OpenAI Codex
33
Cursor
30
Replit Agent
29
Where the cases come from
Show as a table
Source type
Cases
Case study
1613
Blog
329
Reddit
279
X
276
Hacker News
172
Lenny's Newsletter
23
News
21
GitHub
18
Vendor blog
9
LinkedIn
3
Press release
3
ZeroHedge
2
By industry
Only the 8 industries with their own guide.
Show as a table
Industry
Cases
banking and finance
18
retail and e-commerce
16
education
16
logistics and manufacturing
16
insurance
15
healthcare
15
travel and hospitality
15
legal work
11
When I added them
By month: 2026-10: 1086, 2026-11: 1668. Honest note: the directory is new, so every case was added in that one month and a month-by-month trend would be one bar. Here it is by day instead.
Show as a table
Day added
Cases
2026-10-15
991
2026-10-19
2
2026-10-29
88
2026-10-30
5
2026-11-11
1374
2026-11-12
142
2026-11-13
52
2026-11-14
100
Method
Topic, agent and source come from each case's record. Tool counts match the tool's name in the agent field or in the case text. Industry counts use the lists behind the industry guides. Nothing here is a survey. Raw data: the API.