Original library data · October 2026
What are people using AI agents for?
In The Daily Prompt’s October 21 snapshot of 1,462 published agent-use entries, the largest categories were work and operations (253), coding and DevOps (224), home administration (191), and sales and marketing (190). These counts describe our library’s coverage; they do not estimate how all AI users spend their time.
Our Agent Uses library collects reported workflows, sources and prompts. We froze its category and agent filters to make the collection easier to inspect and cite. The four largest categories contain 858 entries, or 58.7% of this snapshot.
Which tasks appear in the library?
Percentages use the full 1,462-entry library as their denominator. The eleven category counts sum to 1,462; percentages are rounded to one decimal place.
| Category | Entries | Share of library |
|---|---|---|
| Work & Operations | 253 | 17.3% |
| Coding & DevOps | 224 | 15.3% |
| Home & Life Admin | 191 | 13.1% |
| Sales & Marketing | 190 | 13.0% |
| Money & Bills | 122 | 8.3% |
| Travel & Bookings | 103 | 7.0% |
| Research & Learning | 102 | 7.0% |
| Shopping & Errands | 98 | 6.7% |
| Content & Creative | 96 | 6.6% |
| Games & Simulations | 49 | 3.4% |
| Science, Security & Hardware | 34 | 2.3% |
| Total | 1,462 | 100% |
Category links open the current collection, which may have grown since this capture. A larger category does not establish greater real-world adoption or better results.
Which agent labels appear?
The snapshot includes 1,575 label assignments across 28 labels. Entries can carry multiple labels, so these counts overlap. They are not market share. “Custom agent” is a grouping; labels include products, assistants and automation tools classified by the library.
| Library label | Entries with label |
|---|---|
| Custom agent | 252 |
| Claude Code | 194 |
| Grok Bot | 109 |
| Claude | 104 |
| OpenClaw | 78 |
| n8n | 76 |
| ChatGPT Agent | 66 |
| Codex | 58 |
| Claude Cowork | 58 |
| Meta Muse | 53 |
| ChatGPT | 52 |
| Manus | 50 |
| Gemini | 48 |
| Cursor | 43 |
| Devin | 40 |
| Perplexity Comet | 38 |
| Claude in Chrome | 37 |
| Replit Agent | 35 |
| Hermes Agent | 31 |
| Dots | 24 |
| Lindy | 24 |
| Operator | 23 |
| Claude computer use | 18 |
| Make | 17 |
| Zapier Agents | 16 |
| Alexa+ | 14 |
| Microsoft Scout | 10 |
| GitHub Copilot | 7 |
| Label assignments | 1,575 |
The 113 assignments above the entry total do not tell us how many entries involve multiple tools. Tool labels also do not establish capability, quality or successful completion.
Three reported examples to explore
These recent entries illustrate different tasks. They are not a representative sample.
- Let an agent read your notes and log a nightly entry
- Let OpenClaw coordinate email, meetings, and research
- File pet food reimbursement claims in one batch
Read each entry’s underlying source and context before adapting its prompt. The reports are not independently verified experiments, and this analysis does not measure success rates, time saved or money saved.
Data, method and citation
Download all aggregate counts as CSV. The file contains the library total, every category and every agent label, with the snapshot date and denominator. It includes no personal information.
We captured the public library filter counts on October 21, 2026 and checked their sums against the reported total. The unit is a published library entry, not a unique person, organization or experiment. We did not deduplicate underlying reports or audit their outcomes. Selection, classification and publication choices shape this collection.
Suggested citation: The Daily Prompt (2026). AI Agent Uses Library: October 21, 2026 Aggregate Snapshot. Cite this dated analysis and its CSV so later library updates do not change the cited figures.
Find your next useful AI workflow
Get the daily AI briefing and browse practical AI answers.