> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fetchlake.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Review and export

> Check the AI's work, override where needed, and export your results.

## Reviewing decisions

The results view brings together every decision with progress and accept-rate figures. Filter it to focus your checking:

| Filter | Use it to |
| - | - |
| **AI Decisions** / **Manual Decisions** | See who decided what |
| **Low Confidence AI** | Check the AI's least certain calls first |
| **Accepted Only** / **Rejected Only** | Work through one outcome at a time |

Sort by **confidence**, **title** or **year**. Select a record to see its abstract and the AI's reasoning and confidence, and change the decision if you disagree.

### A checking routine

1. Review **every** low-confidence AI decision.
2. Resolve **every** Maybe: include it for full-text review, or exclude it.
3. Spot-check a random sample (for example 10%) of confident AI exclusions. If you disagree with more than a few, adjust your criteria and re-run.

## Exporting

Choose **Export** to download your **accepted** (included) articles as an **RIS** file you can import into EndNote, Mendeley, Zotero or RefWorks.

* The export includes title, authors, abstract, journal, year and DOI.
* The original RIS details are kept where available.

> [!TIP]
>
> Export your included records and add them to a [Fetchlake Research](https://docs.fetchlake.com/research/overview.md) collection for full-text reading and data extraction.

## For your PRISMA diagram

Note the counts from the project overview at each stage: records imported, excluded at title/abstract, and sent on to full text.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.