Workspace
How Chat Works
Chat is the primary interface for interacting with ASAPilot's AI agents. Ask questions in natural language, get instant analysis, and take action on your Apple Search Ads β all through guided conversations.
The Chat Interface
Every conversation starts with three controls at the top that define the scope of your query. These ensure the AI agents always work within the right context.
Mode Selector
Switch between AI Insights and other analysis modes
Time Range
Yesterday, 7D, 30D or custom date range for your query
App Filter
Scope the conversation to all apps or a specific one
Thread History
Resume any previous conversation from the sidebar
Default scope: When "All Apps" is selected, chat analyzes all selected apps from your active connections. When you pick a specific app, all analysis and actions are scoped to that single app. Read-only connections can be analyzed but not modified.
Guided Prompts
Don't know where to start? The chat homepage surfaces four categories of guided questions that map directly to common daily workflows. Each prompt is routed to the most relevant AI agent automatically.
Diagnose Issues
β Analytics Agent
Why is my CPI increasing this week?
Which campaigns are wasting budget?
Daily Check-In
β Analytics Agent
Give me an overview of my account performance
Which campaigns are performing best?
Optimization
β AutoPilot Agent
What should I optimize first?
How can I reduce my CPI?
Scale Strategy
β Budget Guard Agent
Which campaigns can I scale?
Where should I increase budget?
Quick Actions
The sidebar also provides one-click quick actions for common operational tasks β these bypass the need to type a question and jump straight to execution.
Run Account Audit
Full scan of account health
Check Budgets
Budget utilization and risk
Create Automation
Set up smart optimization rules
Account Snapshot
At the top of every new chat session, a real-time snapshot displays your key metrics β CPI, Installs, and CVR β for the selected time period compared to the previous period. This gives you immediate context before asking any question.
Account Snapshot
Last 7 Days Β· vs prev periodCPI
$2.34
β 12% vs prev
Installs
1,847
β 8% vs prev
CVR
42.6%
β 3% vs prev
How Responses Are Built
When you send a message, ASAPilot executes a 5-step pipeline to deliver a contextualized, real-time response.
Session Init
A new conversation starts with the selected mode, time range, and app filter.
Context Resolution
ASAPilot loads your access config and resolves the actual allowed app scope for this interaction.
Agent Routing
Based on your question, the system routes to the right specialist agent β Analytics, Campaign, Budget Guard, or Audit.
Streaming Response
Text, tool calls, and agent transfers stream back to the dashboard in real time.
Thread Persistence
The final response is stored so follow-up turns continue from the same context.
Chat β Agent Routing
You don't need to know which agent to use. Just ask your question β ASAPilot automatically routes it to the right specialist.
"Why did my CPI spike yesterday?"
insights Analytics Agent
"Launch campaigns for my new app"
rocket_launch Campaign Agent
"Which campaigns are burning budget?"
shield Budget Guard
"Run a full health check on my account"
health_and_safety Audit Agent
Usage & Threads
AI Query Quota
Each plan includes a monthly AI query allowance. The sidebar displays your current usage and reset date. When the quota is reached, you can upgrade to continue.
Conversation Threads
Every chat is saved as a thread. Resume any previous conversation from the sidebar to continue where you left off β full context is preserved across turns.
Frequently asked questions
What account and app scope does Chat read by default?
Chat reads the selected apps from active, non-paused connections unless the operator explicitly narrows to one account.
Does Chat keep thread history?
Yes. ASAPilot saves the conversation thread so follow-up questions can continue the same analysis.
Is Chat meant for silent writes?
No. The public posture is read-only, so Chat is positioned as an analysis and explanation surface rather than a silent write layer.