Configuring Prompts & Scheduling

This guide walks through the key fields in the prompt editor so you can design reliable monitoring cadences and rich downstream analysis.

Prompt essentials

Label – Appears in the dashboard and Sources filters. Use a descriptive, action-oriented label (e.g., “Enterprise positioning audit”).
Excerpt – Optional helper text that surfaces in prompt tables and filters.
Prompt text – The full instruction sent to each AI platform. Include context hints (“Act as a CMO evaluating…”) to keep answers consistent.
Visibility – Private prompts are only visible to their creator and owners/admins. Team prompts are shared with every member.

Target platforms & regions

Select one or more AI platforms per prompt. Supported options today are ChatGPT, Perplexity, and Gemini.
Regions let you localize prompts. Leave blank for a global audience.

Scheduling options

Option	Description
Daily	Runs once per day at the scheduled time
Weekdays	Monday–Friday cadence
Weekly	Pick the day of week that matters most
Custom days	Choose specific days (e.g., Mon/Wed/Fri)
Interval	Runs every n minutes (minimum depends on plan)
Manual	No automation; run from the table or via API when needed

Time – Use the team’s timezone or override it per prompt.
Interval – Required only when Interval cadence is selected.
Custom days – Appears when the cadence is set to custom_days.

Updating the schedule pushes changes to the next automation window; in-flight runs are not cancelled.

Analysis policy

Analysis policies control which enrichment jobs run after the raw AI answer is captured. Current toggles:

Ranking – Extract competitors and normalize share-of-voice positions.
Sentiment – Score the response on a velocity scale.

Modes:

Auto – Chatobserver decides based on your plan and toggles in the prompt.
Always – Force the job to run even if the default policy would skip it.
Never – Skip the job to conserve credits.

Personas & groups

Personas represent archetypes you want to compare (e.g., “VP of Demand Gen”). They’re created in Personas and attached in the prompt editor.
Groups cluster prompts by campaign, product line, or ownership. Groups also power filters in the Prompts tab.

Prompts can belong to multiple personas and groups.

Result sync status

The prompt list shows the last known status for each prompt:

Pending/Queued – Awaiting the next trigger window
Running – Actively collecting answers
Complete – Latest batch processed successfully
Failed – Automation was unable to fetch answers; check the run details
Credit hold – Workspace has exhausted monthly analysis credits

Statuses update in real time as Chatobserver processes runs. Open the run drawer for detailed timestamps, the rendered response, and any error message.

Best practices

Keep prompts short and consistent across platforms; you can tailor phrasing inside the same prompt when necessary.
Use interval or manual cadences for high-frequency monitoring paired with credit top-ups.
Archive prompts instead of deleting them when you want to pause reporting but preserve historical context.