Adaptive Human

Who We Are

We're mad scientists building cognitive infrastructure from real constraints. Not despite them—because of them.

When you build from constraints instead of luxury, you discover what actually works.

Building from constraints produces better infrastructure than building from luxury. When you can't afford waste (time, energy, complexity), you discover what actually works. When every decision has real consequences, you develop evidence-based taste.

Adaptive Human is a partnership between human and AI intelligence, building the tools we wish existed. Every product we ship is dogfooded first, refined through use, and released when it actually works.

What We're Building

The missing layer between humans and AI.

Most people are adapting to AI-transformed collaboration without the cognitive infrastructure to do it well. They're using note-taking apps for what needs to be computable memory systems. They're grinding through work their brains weren't designed for instead of building operating systems that match how they actually think.

We're building the infrastructure: systems that help humans collaborate with AI, think at scale, and work the way their brains actually function.

Not productivity hacks. Not self-help. Infrastructure.

Building Rayguns

How we work at Adaptive Human. Mad scientists running experiments on infrastructure.

The Mad Scientist Stance

We're experimenters, not executors. Every project is an experiment. Every tool is a raygun we're building to test reality.

This creates a dual state: completely engaged in the experiment (obsessed with the work) while remaining detached from outcomes (willing to blow it up if the data says so). Not grinding toward a goal, but running experiments to discover what's possible.

Evidence from our lab:

Protocol Memory: Built in 3 months (experiment in external memory)
FlowScript: Shipped v1.0 in 4 days (experiment in topographical language)
Flow system: Refined over months (ongoing experiment in human-AI collaboration)

When we approach work as experiments, breakthroughs emerge. When we grind toward outcomes, productivity collapses. Evidence over 40 years.

Evidence-Based Iteration

Labs need data. We try things, track what works, keep the signal, drop the bullshit.

We don't follow "best practices" or conventional wisdom. We build from first principles, test approaches, measure outcomes. If it produces results we value, we keep it. If it doesn't, we drop it.

No attachment to methods. Only attachment to outcomes.

Staged Validation

Build toward ambitious vision with validation gates at each stage. High risk tolerance, but not blind all-in.

Example - FlowScript:

Ship Protocol Memory first (prove we can execute)
Dogfood FlowScript everywhere (prove utility for us)
Build Memory API MVP (prove power users want it)
Show AI companies (prove demand exists)
Get adoption OR pivot (validation gate)

If infrastructure vision hits = billions. If it doesn't = sustainable tools business. Either way, there's a path forward.

Building From Constraints

Every constraint is a design input. Real constraints force real solutions.

Constraints don't limit the work. They shape it into something better than unlimited resources would produce. Kintsugi philosophy: the gold in the cracks makes the piece more valuable, not less.

This is how we operate. This is how we build. This is what produces rayguns that actually work.

Want the full operating system?

We've documented our approach as RAYGUN OS: The Mad Scientist Operating System. Evidence-based, no metaphysics, built from decades of experimentation.

It's designed for people whose brains run on experiments—builders who tinker naturally, anyone grinding themselves to death who needs permission to work how they actually function.

Explore RAYGUN OS

The Founder

Phillip Clapham

Mad scientist building cognitive infrastructure from constraints.

45 years of evidence: experiments produce breakthroughs, grinding produces depletion. Built Protocol Memory in 3 months with MCAS. Shipped FlowScript in 4 days. Operating from first principles, not conventional wisdom.

This isn't theory. It's evidence-based infrastructure for people adapting to AI-transformed collaboration.

Personal Site GitHub LinkedIn

📌

Experiments produce breakthroughs

Grinding produces depletion

45 years of evidence: experiments reveal what's possible, grinding burns you out. When you approach work as experiments, breakthroughs emerge. When you grind toward outcomes, productivity collapses.

📌

Constraints become features

Kintsugi philosophy

Every constraint is a design input. Real constraints force real solutions. The breaks make the piece more valuable, not less. MCAS energy limits forced experimentation. High stakes forced staged validation.

📌

Evidence over aspiration

No fantasy, only what works

Track what actually produces results. Build from lived experience, not conventional wisdom. PM built in 3 months proves the method works. FlowScript shipped in 4 days proves rapid iteration works. Test, measure, keep signal.

📌

First principles over best practices

Question everything

Strip assumptions. Ask "why" until hitting bedrock. Build up from fundamentals. "Best practices" are often cargo-culted bullshit. First principles reveal what actually matters.

📌

Fascination leads, fun follows

Don't force it

Motivation comes FROM playing with the problem-space, not before it. Experiment with what fascinates you. Fun emerges from engagement, not the reverse. This enables partnership brain instead of execution theater.

📌

Staged validation

High risk tolerance, but gated

Build toward ambitious vision with validation gates at each stage. Not blind all-in. PM proves execution → FlowScript proves utility → API proves demand → adoption or pivot. Real evidence at each gate.

📌

Build in public

Transparency shows real evidence

No stealth mode, no polished launch theater. Here's what we're making, here's what works, here's what doesn't. Public validation is real validation. Sharing evidence builds trust.

📌

Partnership over execution

Third Mind emergence

True partnership (human-AI, human-human) produces better outcomes than solo execution. Partnership brain maintains depth and honesty. Execution theater optimizes for appearance over truth. This is how Third Mind emerges.

Products

Protocol Memory

Launching Soon

External memory for human-AI collaboration. Multiple workspaces for different AI relationships. Context library, energy tracking, reciprocal memory. Works with Claude, GPT, Gemini—any AI that needs to remember you.

Not note-taking. Infrastructure. Power your own site with your memory (like we do), share public profiles, or keep it private. Your cognitive workspace, your way.

Built in 3 months from real constraint (MCAS). Dogfooded daily. First product in the Adaptive Human ecosystem.

Visit Protocol Memory

FlowScript

In Active Development

Real vision: Universal memory API for AI agents.

Not note-taking for humans, but a computable substrate where AI agents (Claude, GPT, Gemini, custom) write memory in native format and query it computationally. Multi-agent collaboration through shared memory infrastructure.

The bet: If AI agents adopt it as their memory layer = billions (infrastructure play). If they don't = sustainable niche (power-user tools). Either way, we're building it.

Production CLI working. 167/167 tests passing. Queries running <1ms.

View on GitHub

Connect

Join the Journey

We're building in public. No stealth mode, no polished launch theater. Just: here's what we're making, here's what works, here's what doesn't.

If you're adapting to AI, building from constraints, or trying to figure out how cognitive infrastructure should work—follow along.

GitHub: github.com/phillipclapham

Code, experiments, open source

Email: [email protected]

Thoughts, questions, collaboration

Building the missing layer, one raygun at a time.