When an agent needs fresh information from the web, you no longer stare at a blank thread wondering whether anything is happening. MorningAI shows live research activity while the turn is in progress — so you can see what the agent is searching, what it is reading, and which sources it is pulling from.
Why live research activity matters
Agent turns that use web research can take a few seconds longer than a quick reply. Visible status reduces guesswork: you know the agent is working, you can see the query it is exploring, and you get an early signal about which sites will inform the answer.
What you will see during a turn
While research is running, the thread shows status such as:
- Searching "…" with the query the agent is running
- Reading followed by the site's domain as the agent opens a source
- Searched the web when the research step completes, with a compact stack of source favicons that grows as more pages are consulted
This also works for pages you bring yourself: paste any URL into the composer and the agent reads that page as part of its answer — you'll see the same Reading status for the domain you pasted.
These indicators live in the progress chip for the in-flight turn. When the agent finishes, the chip collapses to a summary you can expand to review the step history — the final answer stays front and center.
Step-by-step walkthrough
- Open an agent thread and ask a question that benefits from current web information — market trends, competitor pages, recent news — or paste a URL you want read in context.
- Watch the thread while the turn runs. If web research kicks in, you will see Searching or Reading status replace the idle waiting state.
- Scan the favicon stack to preview which domains the agent consulted before the answer lands.
- Review the completed response. Use the answer as you normally would — cite it, edit it, or send follow-up questions in the same thread.
Real-world use cases
Competitive checks before a campaign brief. Ask an agent to compare positioning on a competitor site. The live status confirms it is reading the right pages, not guessing from memory.
Timely social copy. When a trend is moving fast, web research activity shows the agent is pulling current context before drafting caption options.
Pro tips
Be specific in your prompt. Named companies, product categories, or URLs help the agent research efficiently.
Do not refresh mid-turn. The status clears when the turn completes — interrupting can restart the work.
Frequently asked questions
Does every agent turn use web research? No. Simple questions that your brand context already covers may answer without leaving the thread.
Can I click the favicons? The stack is a progress indicator during the turn. Review sources in the completed answer or ask a follow-up if you need a specific citation.
Next steps
Open an agent thread and ask a question that needs current web context — any of the seven experts can research the live web. Open MorningAI →