AI Agent Use Cases
AI Agent Use Cases
Escribano is a local work memory for AI agents. It captures, indexes, and retrieves everything you do — files edited, commands run, pages visited, messages sent — so your agent can answer questions about your work with actual evidence.
How it works
Escribano runs entirely on your Mac. It records your screen, transcribes audio, and indexes every observation into a local SQLite database. When an agent needs context, it queries Escribano via a simple CLI or API call. All processing happens on-device — no data leaves your machine.
Use cases
1. Research Summary
"Summarize my research on vector databases from this afternoon."
After hours of scattered exploration — docs, blog posts, Slack threads, terminal experiments — your agent can reconstruct what you learned and what you concluded. Escribano correlates browser history, file edits, and terminal output into a coherent summary with sources.
Example output:
pgvector: tested locally, works for current scale
Pinecone: noted as "overkill for v1"
Weaviate: closed tab after 5 minutes
2. Debug Context Recovery
"What was I debugging yesterday around 3pm?"
Overnight context loss is real. Escribano reconstructs your debugging session from file edits, test failures, documentation visits, and Slack messages — so your agent can pick up where you left off.
3. Incident Runbook Reconstruction
"Reproduce the steps we took for last week's database incident."
Incidents generate scattered signals: PagerDuty alerts, terminal commands, Slack threads, log grepping. Escribano correlates these by timestamp and reconstructs a step-by-step timeline. Your agent turns this into a structured runbook or post-mortem.
4. Morning Pickup
"What should I pick back up?"
After a weekend or a context switch, your agent checks Escribano for unfinished work: open branches, failing tests, unanswered Slack threads, tabs you never closed. It suggests the highest-priority item to resume.
Privacy
🔒 Local only. Escribano stores everything in ~/.escribano/ on your Mac. The database is SQLite with WAL mode. No cloud service, no API calls, no telemetry. Your work memory stays yours.
Try it
If you have Escribano installed, ask your agent:
escribano-query search "your topic here" -d 24h