For most of the history of software, the hard part was building it. A program needed a developer, a budget and months of meetings. So there were few programs, and the ones that existed were big: a CRM, an accounting package, an app on your phone. Each came with a way to get it to you, a store, an installer, a person from IT.
That part is over. An AI writes the code now, and a tool for one job takes an afternoon. The people making these tools aren’t developers. They’re the ones who do the job: the salesperson who wrote a skill that drafts follow-up emails, the planner with a script that merges ten spreadsheets into one, the manager who saved a set of instructions that checks a contract the way the company wants. Most of these tools will never be products. They’re small and handy, and the person next to you would use them tomorrow.
What hasn’t changed is how they get from one person to another.

A skill that never reached the person who needed it
Take a colleague of ours. In a meeting last week we suggested he try Grilling. It’s a skill, a saved set of instructions that teaches an AI assistant to do one job a certain way. This one has the assistant question you about a plan, one question at a time, until the weak spots show. He isn’t a developer. He knows the business well, and he has plans worth questioning every day.
The skill sits on GitHub, a website where developers keep and share code. A project there is a folder of files with instructions for setting it up on your own computer. He has no account on GitHub and no idea how to install what he’d find. A developer gets through those steps without thinking. Our colleague stops at the first one. So he plans without the skill, and the skill stays where it is.
Multiply that by every small tool in every company. The people who’d use them sit one desk away, and the only way to hand them over was built for developers.
Why we didn’t start with a store
We’re building IntoChat, a chat that works on top of your software, and in the same meeting we argued about this problem. The first idea was a marketplace, a store where people publish what they’ve made and others browse and install it. The case for it was strong: sharing is the biggest problem for anyone who builds tools this way, so solve it from day one.
We decided against it, for now. Someone has to run a store and check what goes into it. Without that, the first bad upload makes the whole store look like junk, and the good tools get lost in it. Our conclusion was to start with a link. In the version we’re building, every app will get its own address you can paste anywhere, in a message or on a forum. If people later ask for a store, that’s the sign to build one.

What a link has to carry
A link only helps if the person on the other end can use the tool straight away, so that’s what we’re designing for. The link will be the install: you paste it into the chat, say “install this app”, and use it in the same conversation. The app will list the accounts it connects to, like Google Drive or Salesforce, and you’ll sign in to those with your own details, so nobody’s password travels with the link.
We also cut what there is to share down to one kind of thing. A few weeks ago we wrote about three sizes of tool. In that meeting we dropped them for one: an app. Whoever receives a link shouldn’t have to know what size they got.

Share one tool properly this week
None of this exists yet, and you don’t need it to share better this week. Pick one small tool you made for your own job, a skill, a script or a spreadsheet, that a colleague would use. Before you send it, write two lines: what it does, and which accounts it needs. Send both in one message.
If your colleague still can’t get it running, ask where they got stuck. That’s the step a link has to remove.