The classic time-sink: forty open tabs comparing two products, half of them ads. Your Muse reads the reviews, forums, and official pages, then hands you one table with sources linked. Perfect for big purchases, hiring a contractor, or any decision with too many options. Nothing is bought or booked here — this is pure research.
Connect these first
Connect a docs app so the agent can save briefs where you'll find them. Everything else it needs is the built-in web browser — no purchases involved.
Frame the question
Tell Muse what you’re deciding, your budget, and what matters most — vague questions get vague tables. ‘Best laptop under $1,000 for video editing, battery matters most’ beats ‘good laptop?’ every time. The quality of the whole answer starts right here.
Paste into Muse
I'm deciding between [product A] and [product B] for [use case]. My budget is $[X]. Research both: current prices, real owner complaints from forums and reviews (not marketing copy), and warranty or return terms. Summarize in a comparison table with a source link for every claim.
Pressure-test the winner
Ask Muse to argue against its own recommendation — to find the flaws, the bad reviews, the reasons to pick the runner-up. This catches the hype before your money does. A recommendation that survives its own counter-argument is one you can trust.
Paste into Muse
Now argue against your own recommendation. What are the three strongest reasons someone would regret buying [the winner]? Be specific and cite sources.
Save the brief
Muse saves one document with the comparison table, the counter-arguments, and the links — filed where you’ll find it when you’re ready to buy. A month later you won’t remember why you picked it; the brief will. Decisions with receipts beat decisions from memory.
Paste into Muse
Save the comparison table and the counter-arguments as a Google Doc titled "[Product] research — [date]". Put the source links at the bottom and share the doc link with me.
Where approvals sit
No approvals needed — nothing is sent, bought, or booked. The one rule: treat every price and claim as a draft until you verify it (see below).
What to verify
Click the source links. Open the store page and check the price yourself — agents quoting stale catalog prices and expired discounts is the single most-reported agent failure. If a claim has no link, ask the agent where it came from before trusting it.
What can go wrong
Accuracy is the number one complaint about AI agents: they state wrong prices and confident-sounding facts with equal confidence. They can also lean on thin sources — three blog posts that all quote each other. Demand links, verify prices on the live page, and never let a research brief turn directly into a purchase without a human look.
Research in minutes, not tabs: questions
Research with an agent is fast — which is exactly why the questions matter: where do the sources come from, how does the pressure-test catch hype, and where does the brief live afterward? These answers cover the three steps and what makes the output trustworthy.
- How is this different from just asking a chatbot?
- A chatbot answers from memory. This workflow sends the agent to browse live pages, forums, and reviews, then structures it into a table with sources — and saves it as a document you keep.
- Can it research services too, like contractors or insurers?
- Yes — same steps. Ask for licensed/insured status, real customer complaints, and quote ranges. For anything regulated, verify credentials on the official registry yourself.
- What if the two options I name aren't actually the best?
- Add this line to step 1: "If there's a clearly better option I'm missing under $[X], include it as a third row." Agents are good at surfacing the alternative you didn't know existed.