From Prompt to Live Directory in 10 Minutes
A minute-by-minute account of building a real directory through conversation, what the agent gets right, what needs correcting, and what ten minutes genuinely does and does not produce.
Every tool in this category claims some version of "live in minutes". Most of them are technically telling the truth and practically misleading, because what exists after those minutes is a shell.
Here is an honest minute-by-minute account, including what ten minutes does not give you.
Minute 0: the input that decides everything
Before anything, write your description properly. This is the highest-leverage thing in the whole process and it is the step people rush.
Weak: "a directory of gunsmiths"
Strong: "a directory of gunsmiths in France, for people who need work on a specific regulatory weapon category and want to know whether the shop can deliver to a partner armoury. Buyers care about which calibres are serviced, certifications held, and turnaround time."
The difference is not length. It is that the second one states what a buyer decides on, which is what the agent needs to design a data model. A vague description produces a generic schema, and a generic schema produces a directory that cannot filter.
Spend five minutes here. It is not part of the ten.
Minutes 1 to 2: connect and create
You authorise the MCP connection in your AI client, once, over OAuth. The client discovers the available tools.
You paste your description. The directory is created and a subdomain is minted.
Nothing interesting has happened yet.
Minutes 2 to 4: the schema
The agent designs the typed attributes for your niche and shows them to you.
For the gunsmith example that produces something like: regulatory weapon category as a constrained set of values, partner-armoury delivery as a boolean, calibres serviced as a multi-value list, certifications held, turnaround in working days.
This is the step to slow down on. Read the proposal properly and ask one question: could a buyer filter on these and reach a shortlist of three?
On a common niche the first proposal is usually close. On an unusual one it will miss something. Say what is missing: "add whether they work on antique firearms, and split turnaround into standard and expedited."
One round of correction is normal. Do it now, because retrofitting a schema at three hundred entries is close to impossible and most people never finish.
Minutes 4 to 7: entries
You ask the agent to research the niche and insert listings. It runs a background job that discovers entries, extracts their data against your schema, and inserts them.
What you get back is a working set that shows you the shape of the market. Categories become obvious. Gaps in your schema become obvious.
What this is not: publishable data. Anything a visitor would act on needs verifying, contact details, certifications, regulatory categories, hours. Publishing unverified research is how a directory acquires a reputation for being wrong, and accuracy is the entire product.
Treat the first pass as a draft with a useful shape.
Minutes 7 to 9: structure
Categories, and collections for the comparative queries that still earn clicks after AI Overviews. "Gunsmiths offering partner-armoury delivery" is a collection and it is exactly the kind of page an answer box cannot replace.
One rule to hold: do not create a category or collection with fewer than five to ten real entries behind it. Thin pages are a liability now, not a placeholder. Google's March 2026 enforcement was specifically about pages that exist without substance.
Minute 10: publish
One instruction and the directory is live.
And here is the part the marketing usually skips. That live site is on a free subdomain, which is noindex by design so a shared domain does not become an SEO spam magnet. It cannot rank. If organic search is your plan, you need the Pro plan and a custom domain, and you want to attach it now rather than after three months of work.
What ten minutes actually gave you
A real data model designed for your niche. A working set of entries showing you the shape of the market. A structure. A live URL you can show someone.
That is genuinely useful. It is a validated starting point rather than three days of configuration.
What it did not give you
Verified data. The research pass is a draft.
Depth. You need thirty to fifty complete entries in one tight category before the directory is worth visiting. That is 10 to 20 minutes per entry, researched properly. Call it a weekend.
Category introductions. 150 to 300 words per category, written by someone who understands the niche. The most undervalued real estate on any directory and the cheapest differentiation available.
Indexation. Custom domain required.
Traffic. Three to six months on a new domain before meaningful organic traffic, regardless of how good the setup is.
A business. Listers do not know you exist yet. Telling every listed business that they are listed is your first distribution and your best early backlink source, and it is not automated.
The honest framing
Ten minutes replaces the part that used to take three days: setup, configuration, schema decisions, deployment.
It does not touch the part that takes months: sourcing real data, verifying it, writing the category content, doing outreach, and keeping the whole thing accurate for years.
That is fine. The setup was never the reason directories fail. Abandonment is, and the friction of maintenance is what causes abandonment, which is why the ten minutes matters less than what happens in month fourteen when a listing has a wrong price and fixing it is a sentence rather than a deploy.
FAQ
Is ten minutes realistic?
For a live directory with a reviewed schema and a first pass of entries, yes. For something worth sending to a stranger, a weekend.
What if the schema is wrong?
Correct it immediately, in conversation. Correcting at minute three is free. At entry three hundred it is a project.
Can I import my own data instead of researching?
Yes. Bulk operations match on slug and update rather than duplicate, so imports are safe to re-run.
Do I need to verify everything?
Anything a visitor would act on. Contact details, certifications, regulatory categories, hours. Descriptions matter less.
What is the single most important minute?
Minute zero. The quality of your niche description determines the quality of the data model, and the data model determines everything else.
Write your description and see what comes back →