Creating a corpus is only the beginning. As your content grows and your needs change, the corpus Settings tab lets you tune how the corpus is indexed, keep web content fresh automatically, organize it with tags, and protect personal data, all from one place.Why corpus settings matter#
Answer quality vs. cost. The indexing mode decides how thoroughly content is processed for search. Moving a key corpus to a higher mode can noticeably improve answers; keeping simple corpora on Basic keeps costs down.
Fresh answers without manual work. Websites change constantly. Scheduled Refresh re-indexes your web links on a schedule, so assistants don't quote last month's prices or an outdated policy.
Better organization. Corpus tags label everything in the corpus at once, which helps search and filtering.
Privacy. PII Redaction masks personal information before assistants use your documents.
What you can change#
| Section | What it does | When to change it |
|---|
| 1. Basic Settings | The corpus Name and Description | When the corpus's purpose changes, or to make it easier for others to recognize |
| 2. Indexing Settings | Basic, Moderate, or Exhaustive indexing | When answers aren't precise enough (move up) or costs matter more (move down). Choose Exhaustive to upload video files |
| 3. Scheduled Refresh | Automatic re-indexing of web links | When the corpus includes web pages that change |
| 4. Tags | Labels applied to every document in the corpus | To improve search relevance and organize corpora |
| 5. PII Settings | Enable PII Redaction | When documents contain personal information |
Choosing an indexing mode#
Indexing is how Ejento AI processes your content so assistants can search it. The indexing mode controls how thoroughly that happens, which affects both how relevant search results are and how much indexing costs.| Mode | What it means | Best for | Trade-off |
|---|
| Basic | The most cost-effective option, for when indexing quality isn't the highest priority | Small or simple corpora, such as FAQs, short policies, or test corpora | Lowest cost, but search results may be less precise |
| Moderate | A balance of performance and cost: efficient indexing without the need for maximum accuracy | Everyday business content, such as help centers, handbooks, and product guides | A middle ground between cost and accuracy |
| Exhaustive | The highest-performance option, with the most comprehensive indexing for the best search accuracy and relevance | Large, detailed, or high-stakes content, such as contracts, technical manuals, or research; also the only mode that supports video files | Highest cost |
Start with Basic for a new corpus, and test your assistant with real questions.
Move up to Moderate or Exhaustive if answers miss details that are in your documents, or if the corpus is large and complex.
Use Exhaustive when accuracy really matters, for example legal, compliance, or medical content, or when you need to upload videos.
Stay on Basic for content that rarely changes and gets simple questions, to keep costs down.
You can change the indexing mode at any time from the Settings tab.
Understanding Scheduled Refresh#
Scheduled Refresh keeps web content current without you re-uploading anything. Keep these points in mind:It applies only to web links. Uploaded files and text blobs aren't refreshed; update those by uploading a new version.
Incremental Refresh refreshes only the links that have changed since the last run. It's faster and cheaper, but it relies on ETags from the website. If the site doesn't send ETags, the refresh won't work as expected. See ETag Setup for Corpus Incremental Refresh. Full Refresh refreshes all links from scratch on every run. It works with any website, but takes longer and costs more for large sites.
Pick a schedule. Set the Frequency (Daily, Weekly, Monthly, or Yearly) and the Time (Local), in your local time zone. A summary such as "Incremental refresh runs daily at 2:30 AM" confirms your choice.
Choose a quiet time. Running refreshes outside business hours, such as overnight, means content is updated before people start asking questions.
Match the frequency to the content. A pricing page that changes weekly may need a daily refresh; an annual policy page only needs a monthly or yearly one.
Use cases#
Help center that changes daily: A support team indexes its public help center and sets an Incremental Refresh to run Daily at 2:30 AM, so new articles are searchable by morning.High-stakes contracts: A legal team moves its contracts corpus from Basic to Exhaustive, because precise answers matter more than cost.Employee records: An HR team turns on PII Redaction before uploading employee records to an HR assistant's corpus.
1. Open the corpus Settings tab#
Open your corpus and click the Settings tab. Update the Name and Description under Basic Settings if needed, and choose an indexing mode under Indexing Settings.2. Set up Scheduled Refresh#
Turn on Enable Scheduled Refresh. Choose a Refresh Type (Incremental Refresh or Full Refresh), then set the Frequency and Time (Local).Use Incremental Refresh only if the websites you index send ETags. Otherwise choose Full Refresh.
To undo changes you haven't saved yet, click Reset.
