Skip to content
Submit a tool
SPONSOR SLOT · OPENPut your tool in front of builders choosing one right now. One slot, every page, $50 a week.Your tool here, $50 a weekClaim it

HQ / Use cases

Research

These are ways people use AI agents to find, read and sum up information, from comparing sources to building research notes. Each entry sums up a public post on X in our own words and links back to it. These are reports from the people who did the work, not tests we ran.

5reports

3agents

Car insurance quotes compared, with a follow-up email

After a pushy sales call, they asked Grok Bot to find and compare car insurance quotes. They say it visited insurer websites, compared prices and listed several offers. It also emailed one insurer to ask whether EV battery coverage was complete, and the insurer sent a detailed reply.

New hereGrok BotPosted 12 Oct 2026

Reported by @worawisut, not tested by us.

Our prompt to try this

Find car insurance quotes for [car, location, coverage] from at least four insurers' websites. Put them in a table with price and key coverage terms. Draft an email to the best option asking about [specific coverage question] and show it to me before sending.

View the original post on X

Picking a safe soap from the manual

They asked Grok Bot to find a car wash soap for their Cybertruck. They say it checked Tesla's service manual first, picked a pH-neutral soap marked safe for stainless steel, and compared prices across several stores with shipping included. It also pointed out where they could pick it up the same day.

New hereGrok BotPosted 11 Oct 2026

Reported by @Texas_Twins_Dad, not tested by us.

Our prompt to try this

I need a [product] for my [item]. Check the manufacturer's manual for any requirements first, then suggest two or three products that meet them. Compare total price including shipping to [zip code], and tell me if any are available nearby today.

View the original post on X

Seven agents asked the same support question

To compare agents on a very recent problem, they asked several of them whether a sign-in option worked on a business account, where the answer depended on two support docs updated the day before. They report that some agents got it wrong or made up a URL, while ChatGPT Work and Hermes Agent gave correct answers. Their takeaway is that the setup around a model matters as much as the model.

New hereHermes AgentPosted 30 Sep 2026

Reported by @viticci, not tested by us.

Our prompt to try this

Answer this question using only the vendor's current official docs: [question]. Link each doc you used and its last-updated date, and say plainly if the feature isn't available for my plan.

View the original post on X

Weekly AI-tools brief, then a cull

Two scheduled Hermes Agent jobs run their AI news watch. On Mondays one scans GitHub trending, X and YouTube and saves a research brief in Notion. On Fridays the other goes through their bookmarks, tries the tools itself and drops the ones that don't hold up.

Hermes AgentPosted 25 Sep 2026

Reported by @jeliladss, not tested by us.

Our prompt to try this

Every Monday, check GitHub trending, X and YouTube for new tools about [topic] and save a one-page brief to my Notion page [page]. Every Friday, go through the links I saved that week, try any that are free to try, and mark each keep or drop with one line on why.

View the original post on X

Claude writes the brief for Gemini

They had Claude write a deep research brief for Gemini, giving Claude some prompting best practices first. The topic was a newly announced documentary, and they picked Gemini because it can watch YouTube clips. Claude then turns the finished research into a media dashboard for them.

Claude CodePosted 6 Sep 2026

Reported by @BretKerr, not tested by us.

Our prompt to try this

I want to run a deep research task on [topic] in another tool. Write the research brief for it: the goal, key questions, sources to prioritize, what to leave out and the output format. Keep it under one page.

View the original post on X