So many options, so little time

StrategyCanvas is becoming a labyrinth.

Not because the architecture is particularly complicated, but because every feature seems to open three doors to another potentially useful feature.

The original idea was relatively simple: build a node-based environment for strategic thinking. Connect facts, goals, audiences, insights and decisions into systems where relationships are visible instead of disappearing into PowerPoints.

That basic idea now works.

There is a library of more than 200 nodes covering different strategic models and planning needs. I have built reference canvases for strategy, digital marketing, communications, campaigns and concepts. Nodes can contain structured information, connect to each other and ultimately produce actual plans and other outputs.

And every time something starts working, I discover ten things I could build next.

Workshop spaces. Freeform notes that become Facts, Opinions, Ideas or Policies. Hubs. Prioritisation. Wizards that construct a starting canvas through questions. Marketing automation planning. More models. Better editors. Collaboration. New outputs.

This is the dangerous part.

Getting lost in features

I could probably spend the next year making StrategyCanvas increasingly elaborate without ever making it substantially more useful.

So I am trying to narrow the next phase down to four things:

Security. Nodes that actually do something. Useful outputs. Simulation.

Security comes first because these canvases may contain exactly the information companies don't want leaking out: unfinished strategies, assumptions, competitive thinking and uncomfortable internal facts. Private canvas content needs to remain private — including from routine platform administration.

The second priority is making the node system deeper rather than wider. I don't particularly need another hundred node types. I need the existing nodes and their relationships to become genuinely useful instruments for reasoning.

Third, StrategyCanvas needs to escape StrategyCanvas.

A beautiful graph isn't particularly useful when someone needs to present the strategy on Monday morning. PDF and PowerPoint outputs therefore matter much more to me now than another collection of clever features.

And then there is the part I find most interesting: simulation space.

If a strategy is represented as a connected system rather than a static document, you should be able to change an assumption and ask what happens.

What if the budget drops by 30%?

What if this audience behaves differently?

What if this competitor enters?

What if this assumption is simply wrong?

That starts turning the canvas from a place where strategy is documented into a place where strategy can be tested.

AI as a plugin, not the product

AI will probably become part of this, but I increasingly don't want StrategyCanvas to be an AI product.

I want AI to be optional infrastructure.

Turn it on and it can help construct node systems, challenge assumptions, create analyses and scenarios, discover missing connections and improve written outputs.

Turn it off and StrategyCanvas still works.

That distinction is important for information security. Some organisations should be able to use the system without strategic information ever being sent to an external model.

I'm currently leaning towards Mistral for the AI layer. Partly technically, but also because its European roots and greater distance from the largest US technology platforms fit better with what I want the project to become.

The important architectural idea, however, is that the model should be replaceable.

AI is a plugin.

The strategy belongs to the user.

September

My current target is much less glamorous than another giant feature list.

I want to have StrategyCanvas ready for a small beta with a couple of companies by the end of September.

That means resisting the labyrinth.

Security. Better node functionality. PDF and PowerPoint outputs. Simulation. Optional AI where it genuinely improves the work.

Everything else can wait.

There are now enough doors.

The job is to find out which ones are actually worth opening.