The Hex Constraint: Why Six Tools, Not Twelve
You have too many AI tools. You're paying for subscriptions you opened in a moment of enthusiasm three months ago, and the last time you logged into half of them was the day you signed up. The hex constraint is a simple rule that fixes this: pick six AI tools. Wire them together. Use nothing else. That number is not arbitrary, and the constraint is not punishment — it's the thing that makes the tools actually work.
The hex is CustomClanker's core framework, built on a structural observation about how AI tools produce value. The observation is this: every additional tool you add to your stack has a diminishing return on output and an increasing cost in attention, context-switching, and integration friction. Somewhere around six tools — give or take one, depending on your work — you hit the point where adding a seventh makes you slower, not faster. The hex is a constraint that keeps you on the right side of that curve.
Where the Number Comes From
Six is not a magic number pulled from a productivity blog. It comes from three places.
First, the practical reality of MCP — Anthropic's Model Context Protocol. MCP is the architecture that lets Claude connect to your actual tools: your email, your calendar, your files, your CMS, your database. Each connection is a connector. In practice, a Claude session with six well-configured MCP connectors runs smoothly. A session with twelve starts to degrade — not because of a hard technical limit, but because the model's attention to each tool gets thinner as the number grows. The system prompt gets longer. The tool-selection decisions get noisier. The model starts picking the wrong tool for the job more often because there are too many options to evaluate on every turn. Six keeps the agent sharp. Twelve makes it hesitant.
Second, the cognitive science of working memory. George Miller's famous "seven plus or minus two" paper from 1956 established that humans hold roughly five to nine chunks in working memory at once. Subsequent research has revised that number downward — Nelson Cowan's work suggests the real number is closer to four. Six tools is right at the boundary of what you can hold in your head as a complete system. You can remember what each tool does, when to use it, and how it connects to the others. At twelve tools, you can't. You start forgetting what you have, re-discovering tools you already own, and defaulting to whichever one you opened most recently rather than whichever one is best for the task.
Third, the observed behavior of people who actually ship. This is the least scientific and most convincing source. Go look at the setups of people who produce real output with AI tools — not the people who post elaborate architecture diagrams on Twitter, but the ones who publish, build, ship, and invoice. Their stacks are small. Three to seven tools, wired tightly, used daily. The people with fourteen subscriptions and a Notion page tracking their "AI stack" are the ones who produce the least. The constraint isn't holding back the productive people. It's what makes them productive.
What Twelve Gets You That Six Doesn't
The obvious objection: more tools means more capability. If six is good, twelve is twice as good. This feels intuitively true and is practically false.
What twelve gets you is coverage. You have a tool for every possible task. Image generation, video generation, three different writing assistants, two automation platforms, a transcription service, a code assistant, a presentation tool. Your stack handles everything. In theory.
What twelve costs you is depth. You never get past the surface of any tool because you're spreading your attention across all of them. You know how to start a task in Midjourney but not how to build a consistent style. You know how to create an n8n workflow but not how to debug one that fails at 3am. You know how to prompt Claude for a first draft but not how to architect a multi-step process that produces a finished deliverable. Your twelve tools each operate at maybe 20% of their potential. Six tools at 80% of their potential produce more than twelve tools at 20%.
There's a hidden cost beyond depth, and it's the one that actually kills you: decision overhead. Every time you sit down to do a task, you have to choose which tool to use. With twelve options, that choice is non-trivial. Should I draft this in Claude or GPT? Should I generate this image in Midjourney or DALL-E or Flux? Should I automate this in n8n or Make or Zapier? Each decision takes time and energy, and worse, it opens the door to second-guessing. You start the task in Claude, wonder if GPT would have been better, and lose focus. With six tools, the decisions are already made. You know which tool does what in your stack because you chose deliberately and committed. The choosing happened once, not every time you sit down.
Why Constraint Produces More Than Freedom
This is counterintuitive enough that it's worth sitting with. We are trained to believe that more options lead to better outcomes. The opposite is reliably true in creative and productive work.
Constraints force depth. When you only have one image generation tool, you learn it well. You learn its strengths, its failure modes, its prompt language, its settings. You develop craft with it. When you have four image generation tools, you learn none of them well. You hop between them based on which one handled the last prompt best, never building the deep fluency that produces consistently good output. The constraint isn't limiting your capability — it's concentrating your learning.
Constraints force integration. When you only have six tools, they have to work together. You can't afford a tool that sits in isolation, producing output you then manually copy into another tool. Every tool in the hex needs to talk to the others — through MCP, through APIs, through some form of automation. This integration requirement is the filter that separates genuinely useful tools from impressive-looking tools that don't connect to anything. A tool that can't integrate with your hex is a toy, regardless of how good its demo looks.
Constraints force evaluation. The hex makes you choose, and choosing forces you to evaluate. You can't keep three automation platforms "just in case." You have to pick one. That picking process — which one actually runs reliably, which one has MCP support, which one I'll actually use daily — is the evaluation most people skip when they have unlimited slots. They never pick because they never have to. The hex makes you have to.
What the Hex Is Not
The hex is not a productivity system. There is no morning routine, no weekly review template, no journaling practice. It's a constraint applied to your tool selection. Full stop. What you do with those tools is your business.
The hex is not a specific set of tools. "The hex" doesn't mean Claude plus Ghost plus n8n plus Midjourney plus ElevenLabs plus Gmail. Your hex is whatever six tools serve your actual work. A developer's hex looks nothing like a publisher's hex. A consultant's hex looks nothing like a YouTuber's hex. The framework is the constraint. The contents are yours.
The hex is not permanent. Tools improve, new ones ship, your work changes. The hex gets reviewed quarterly — what stays, what goes, what's new. The constraint is the number, not the specific tools filling the slots. You're not married to your current six. You're committed to only having six at any given time.
The hex is not about suffering. If six tools genuinely aren't enough for your work, the constraint might not be right for you. The section on breaking the hex exists for a reason. But the honest truth is that most people who think they need twelve tools actually need six tools and more practice with each one. The urge to add tool number seven is almost always a signal that you haven't gone deep enough with the six you have — not that you've outgrown them.
How the Hex Works in Practice
The hex operates on a simple daily reality: when you sit down to work, you know exactly what tools you have and what each one does. There's no browsing, no evaluating, no "let me try this other thing." The decision tree is short and clear. Need to write? That's tool one. Need an image? That's tool three. Need to automate a step? That's tool five. The speed of starting — the time between "I need to do this" and "I'm doing this" — is where the hex pays for itself.
Over weeks, the hex compounds. Your system prompts get better because you're using the same tools every day and learning what instructions produce the best output. Your workflows get smoother because the integration points between your six tools are well-tested. Your output quality improves because you're building deep fluency instead of shallow familiarity. The gap between you and someone with twelve tools widens over time, not because you have better tools, but because you know yours better.
The hex also changes your relationship with AI tool news. When a new tool launches — and something launches every week — you don't need to evaluate it immediately. You have your six. The new thing is interesting, maybe, but it has to be dramatically better than what it would replace AND have MCP support AND pass a real-world test before it earns a slot. Most new tools don't clear that bar. The hex turns the firehose of AI tool launches from an anxiety-producing obligation into background noise you can safely ignore until your next quarterly review.
The Honest Limitation
The hex assumes you're willing to go deep rather than wide. If your actual need is breadth — if you're a reviewer who has to test fifty tools, or an agency that serves clients with incompatible requirements — the hex constraint may genuinely not fit. It's designed for the person who wants to produce output, not the person who needs to survey the landscape. Those are different jobs, and they need different approaches.
But if you're reading this and your first reaction is "I'm the exception, I really do need twelve tools" — sit with that for a minute. Check whether that reaction is coming from an honest assessment of your work or from the part of your brain that doesn't want to give up the dopamine of new tool discovery. The hex works precisely because it's uncomfortable. The discomfort is the constraint doing its job.
This article is part of The Hex System series at CustomClanker.
Next in series: The Audit — How to Inventory What You Actually Use