Looping Over Data
Process a dataset step by step
You'll be able to
- Drive an agent to iterate over a list, applying the same steps to each item
- Bound a loop so a large or surprising dataset can't run away
Read first
Day 5 Lab
**Status: Generally available — Glean’s Agent Builder and its core features (scheduled triggers, versioning, native actions) are GA. Only the *content triggers* trigger type remains Beta; confirm availability in your tenant.**
- Pick a dataset your chore repeats over — open Jira tickets, new leads, stale docs.
- Have the agent fetch that list as the loop's input, and look at what one row contains.
- Define the per-item steps once: the agent applies them to every row.
- Add a cap — process at most N items per run — so a huge list can't run away.
- Decide what a per-item failure does: skip and log, or stop the whole loop.
Working Example: Iterating Over a List
Per Glean's Fall '25 launch announcement, an agent can loop over a dataset, applying the same steps to each item. Verified against fall25_launch.
Input list: open Jira tickets with no owner -> [PROJ-12, PROJ-19, PROJ-23, ...]
For each ticket in the list (cap: 25 per run):
Step 1 read the ticket
Step 2 find a likely owner from the related component
Step 3 post a suggestion in #triage (read-style, no auto-assign)
On item error: log it, skip to the next ticket
Same steps, every row. The cap and skip-on-error keep one bad row from
breaking — or runaway-looping — the whole job.Looping is leverage: one reviewed step list, applied across a whole dataset. That makes *shaping the input* the critical move — what's in the list, how big it is, what one row looks like — which is MAP work. The cap and the per-item error rule keep an unexpected dataset from turning leverage into damage. Looping is announced capability, not confirmed GA: verify it behaves this way in your tenant before you point it at real data.
Check for understanding
Why should you cap the number of items a loop will process on a single run?
Check yourself
1. In Glean's vibe-code agent builder, how do you start building an agent?
2. The vibe-code builder, 100+ native actions, scheduled triggers, looping, and versioning are best described as...
3. Before adding a native WRITE action (e.g., update a Salesforce case) to an agent step, what should you check?
4. An agent runs unattended on a daily schedule. Which guard most directly keeps a bad run from compounding?
5. Why does rolling an agent back to a known-good version count as a governance control?
In 10 seconds
“Glean's Fall '25 launch announced looping: hand an agent a list and it runs your steps over each row. The skill is shaping the input and bounding the loop so it can't run away.”
Versioning and Lifecycle