PraxisAgent: Autonomous Open-AI Operations Worker Built for My Friend Sarah
说明与用途
PraxisAgent 自动处理发票,防止重复付款和输入错误,并在最终操作前要求用户确认,节省工作时间。
问题
Sarah, who handles billing and clinic operations, repeatedly spends Friday afternoons manually processing scattered vendor invoices, scanning PDFs with scrambled filenames, and entering data into outdated ERP/healthcare portals, risking duplicate payments, typos, and wasted hours.
结果
PraxisAgent achieved a 9/9 verified pass rate across 9 end-to-end chaos engineering scenarios, successfully filtering duplicate invoices, preventing typos, and requiring Sarah’s approval before any irreversible action.
工作流
- 1. Receive plain-English request (e.g., "Find Acme's latest bill in the invoices folder and enter it into the ERP").
- 2. Locate and parse unstructured documents (PDFs, JSONs, text tickets) using pdf-parse and custom extraction logic.
- 3. Distill live browser pages into compact numbered element references ([ref] selectors).
- 4. Navigate the ERP form using the distilled DOM references.
- 5. Intercept state-mutating actions (submit, approve, pay) via the deterministic policy safety gate.
- 6. Prompt Sarah for interactive approval before executing irreversible actions.
- 7. Verify submission by querying the portal’s database endpoint (/__state) and cryptographically proving record creation without duplicates.
证据
PraxisAgent is an autonomous, open-source operations agent that takes plain-English operational requests (e.g., "Find Acme's latest bill in the invoices folder and enter it into the ERP"), autonomously locates and parses unstructured documents (PDFs, JSONs, text tickets), drives a real browser session to populate the portal, and stops at an interactive Human-in-the-Loop policy gate before submitting irreversible actions.
PraxisAgent was designed, tested, and benchmarked across 9 end-to-end chaos engineering scenarios (achieving a 9/9 verified pass rate across validation errors, session dropouts, and duplicate invoice filters).
- 记录日期: 2026-10-04T22:56:07.444Z