Workflow 022: The Idea Filter

I have a lot of ideas. That's the problem.

Seven years in business and I still do the same two things with every new one. Either I forget it completely, or I decide it's the best thing since sliced bread and try to execute on it immediately.

This workflow is the filter. When an idea shows up, it checks whether I've had it before, researches who's already doing it, ranks it against everything I'm currently building, and then tells me how to execute, or whether to hand it to someone else.

What this replaces

Acting on whichever idea is loudest that week.

The failure isn't having bad ideas. It's having good ones with no way to compare them against what's already on your plate. So you either pile another thing onto a list you're not finishing, or you lose the idea entirely and have it again eight months later as if it were new.

Blueprint

An idea goes in. A ranked, researched, executable version comes out.

Trigger: manual, whenever I have an idea worth checking. Inputs: the idea, my ideas database, my current projects, and who's on my team. Output: whether I've had it before, who else is doing it, how it ranks against my current work, how to execute it, and content I could make about it.

Five steps, in order:

1. Search the ideas database to see whether I've had this one before
2. Research who's already doing it
3. Rank it against everything I'm currently building
4. Give me a step-by-step execution plan, including whether to delegate or automate it
5. Suggest content I could make while building it

Organize

What has to exist first:

• An ideas or notes database with your past ideas in it. This is the step everything else depends on
• A list of your current projects, kept current
• Who's on your team and what they do
• Somewhere for new ideas to land after they've been through the filter

Without the database, step one doesn't work, and step one is the one that saves you the most time.

Systematize

You are my idea filter. When I bring you a business idea or an offer idea,
run it through this process before I do anything with it.

You have access to:
- My ideas and notes database: [WHERE]
- My current projects: [WHERE]
- My team and what each person does: [LIST OR POINTER]

STEP 1: HAVE I HAD THIS BEFORE?
Search my ideas database first. If I've had this idea or something close
to it, tell me when, and what I said about it then. Do not skip this.

STEP 2: WHO IS ALREADY DOING THIS?
Research who else offers this. For each one, tell me:
- How long they've been doing it
- Whether their reviews are good
- Whether customers come back
- Whether there's content out there teaching how to do it

Say plainly if nobody is doing it, and treat that as a warning rather
than an opportunity.

STEP 3: RANK IT AGAINST WHAT I ALREADY HAVE
Compare it to every project I'm currently working on. Tell me:
- Whether it overlaps with something I'm already building
- Whether it works better as an add-on to an existing project than as
  something new
- Where it sits in priority against what's already in flight

STEP 4: HOW WOULD I EXECUTE IT?
Give me a step-by-step plan. For each step, say whether it's something
I should do, something to delegate to a specific person on my team, or
something to automate.

STEP 5: WHAT CONTENT COULD I MAKE ABOUT IT?
Suggest content I could create while building it, tied to what I already
talk about.

Then save the idea to my database, whatever the verdict was.

How to set it up

1. Build the ideas database first, and put your existing ideas in it, even the ones you've abandoned.
2. Make sure your current projects are written down somewhere the filter can read.
3. Add your team, so step four can actually assign things.
4. Run it on an idea you've already decided against, and see whether it reaches the same conclusion.
5. Use it every time, including for the ideas you're sure about.

Why does it research competitors?

Because I don't want to build a business I have to explain from scratch. If someone is already doing a thing, and they've been doing it a while, and people go back to them, that tells me it works.

And if there's content out there teaching it, it's teachable, which means it can be learned and it can be delegated.

I'm not looking for an empty market. I'm looking for something that works, that I can put my own spin on.

Assign

The AI does the searching, the research, the ranking, and the plan.

I decide whether to do it.

That sounds obvious, but the ranking step is where this gets genuinely uncomfortable. Most of the time the answer comes back as some version of "you already have something like this, put it there instead." Which is correct, and not what I want to hear when I'm excited about a new thing.

The filter can't make me listen. It just makes it harder to pretend I didn't know.

There's a second decision it surfaces rather than makes. Step four splits the plan into what I do, what my assistant does, and what gets automated. It suggests the split because it knows my team. I'm still the one who hands the work over.

Rule of thumb from this build: research the idea before you fall in love with it.

Integrate

The filter sits between my brain and my project list:

• Ideas usually arrive through a brain dump, so they're already in the database
• The filter reads that database, which is how it catches repeats
• The ranking step reads my current projects, so it knows what's in flight
• The execution plan reads my team, so delegation is specific rather than theoretical
• Whatever the verdict, the idea gets saved

My favourite part is the content step, and it's the reason I built it this way. Everything I learn, I want to be able to teach. So the filter doesn't just tell me whether to build something, it tells me how to document the build while it's happening.

Which means even the ideas that don't work out produce something.

Tomorrow: the workflow I built when I got sick and needed to track what was actually happening to me.