/ writings ·

The lab is the spec

Shipping an interactive simulator is the new pitch deck. Whatever the lab computes is what the system claims. The discipline is wiring the lab to the same engine production runs.

By Abhishek Krishna

The deck told a story you couldn't argue with because the numbers were frozen. The lab makes the same numbers live, and live numbers are arguable. That's the upgrade. The conversation moves from 'do I believe this slide' to 'what happens in scene four when the spread widens.'

There used to be a twenty-slide deck for this. You opened with the market, you closed with the team, and somewhere around slide nine you had a chart that was either ambitious or fabricated, depending on who was reading it.

The deck is no longer the artifact. The lab is.

What a lab looks like

Over the last two months I have shipped three of them and watched a fourth take shape:

  • Meridian, an RWA market-making lab. Pick an asset class, pick a strategy, watch quotes, fills, inventory, and PnL evolve scene by scene. The same engine that runs the lab runs the on-chain executor. Eight scenes.
  • A CR8-USD + MUSD lifecycle lab in the stablecoin toolkit. Mint, redeem, depeg, recover. The reserve manager and the depeg guard you see in the browser are the same Solidity behind it. Eight scenes.
  • An IPTO economics simulator in ipto-dashboards. Per-license value growth, demand curves, payout flows. The numbers that come out of the dashboard are the numbers in the next investor conversation, because they are the same model.
  • A kcolbchain brand playground for color, type, and motion tokens, so anyone shipping a kcolbchain surface can verify the system in a tab.

The pattern is consistent. Eight scenes, each one a state, each one a button-press, each one a number you can argue with.

Why this is the new artifact

Three things happened at once.

Browser compute got cheap enough to run real Monte Carlo and full lifecycle simulation in a tab, without a backend, on a phone. Frontier coding models got good enough that one person can ship the math, the front end, and the writeup in a week. And the audience changed. Investors, partners, regulators, prospective hires all want to touch the system before they form an opinion of it.

A static slide cannot survive that demand. A live lab can.

Fig. 01 · Two artifacts, on the same axes DECK LAB frozen live Audience can argue with it Edge cases discoverable by user Tied to production engine Surfaces internal disagreement Survives a hostile read (absent) ● first-class
A deck holds the same five properties on the audience side, but resolves none of them on the artifact side. The lab does, by being live, by being executable, by being the same code that runs in production. The deck told a story you couldn't argue with. The lab tells a story you can.

What changes when the lab is your default artifact

Specifications become executable. Whatever the lab does is what the system claims to do. There is no second document to maintain, no PDF that says one thing while the contract says another. Argue with the lab and you are arguing with the spec.

Sales conversations get shorter. You stop defending a spreadsheet. The spreadsheet defends itself. The partner clicks scene five, asks one question, and you are already past the part where decks usually stall.

Investors stop needing you to interpret. They click through, find the edge case, and tell you about it. That is the moment a check moves.

Internal disagreement collapses to data. If a teammate thinks the slippage cap is too tight, they open the lab, set scene four to the harder parameters, and either prove their point or change their mind. The argument resolves in minutes, not slack-threads.

The discipline

The risk is obvious. A lab can become marketing, not truth. If the simulator shows the bull case and the production code does something else, you have made a slicker version of the same fabricated chart.

The discipline is non-negotiable: every lab has to be wired to the same engine the production code uses, or the lab is worthless. In Meridian, the lab’s quoting logic is the same Python that the live agent runs. In the stablecoin lab, the depeg guard you watch trip in the browser is the same Solidity that pauses mints on chain. The dashboard’s economics model and the model the founder shows investors are one file.

Anything less and the lab is just a deck in a costume.

A small claim

If you are working on infrastructure right now, especially financial or licensing infrastructure, the single highest-leverage week you can spend is the one where you stop writing the deck and ship the lab.

The lab is the spec. The spec is the system. The system is what people can argue with.

Sources

  • Meridian, RWA market-making stack and browser lab. Open source: github.com/kcolbchain/meridian.
  • Stablecoin Toolkit, CR8-USD + MUSD lifecycle lab. Open source: github.com/kcolbchain/stablecoin-toolkit.
  • IPTO Dashboards, per-license value-growth and demand-based revenue projection. Internal as of writing.
  • kcolbchain brand kit, interactive token + motion playground. Open source.
  • All four ship with their engines pinned to the production code that runs underneath. That is the part that makes any of this honest.