Operating system

Four moves, found in twenty ventures and run the same way in every sector.

We went back through more than twenty of our own ventures and innovation projects looking for what the wins had in common. The opportunity DNA turned out to be identical every time. This is the system that came out of that work.

Signals

Every business records its own behavior, all day, every day: quotes, invoices, contracts, redemptions, routing. When the operating model goes unquestioned for decades, those records start to disagree with the assumptions it runs on. We read them and find where they disagree.

Alphas

A signal becomes an alpha candidate when it can be sized and given a likely explanation. Sized means a dollar range drawn from your own records, an evidence profile, and the condition that would kill it, written down first. The explanation is our best reading of why the gap survived: the incentive that kept everyone playing the old game, or the habit that was rational for each person and expensive for the business. History can find the gap; only a controlled test proves it. A candidate becomes a proven alpha when that test holds, and that is when it earns a build.

Shipping

We don’t hand over findings and wish you luck. We build the mechanism that captures the alpha, whether that is a pricing engine, a market, a buyer, or a product, and it is working inside your operation within weeks.

Portfolio

For owners and investors who want more than one bet at a time. A managed portfolio of opportunities, each one gated, each one killed or scaled on evidence, run like a production line rather than a series of projects. When one of them outgrows the business that found it, we help it become a company.

Two inputs

The records, and everything the records don’t know.

The first input is what the business has recorded about itself. The second is everything it never wrote down: the workaround on the floor, the customer who always calls, the price nobody has touched since 2011. That lives in the heads of the people who run the operation, and half of what we find comes from there. Both inputs land on one map of assumptions, the gaps get sized, and nothing that isn’t on that map gets built.

The workshop is an afternoon with your people, not a program. Afterward we run the same assumptions through a second, virtual room: synthetic panels of buyers, operators, and skeptics, built from your records and your people’s own words and built to argue. The panels don’t tell us what customers will do; nothing simulated can. They tell us where an idea is weakest, so we know exactly what to test with real customers and what to drop before it costs anything.

The records

Quotes, invoices, routing, contracts, redemptions. Twelve to twenty-four months of them, as a structured export. We never ask for access to your systems.

The room

The owner and the operators, for one afternoon. Every instinct, gripe, and workaround in the room becomes an assumption with a test attached, and most of them were never in a database.

The panels

Simulated buyers, operators, and skeptics, built to disagree. They stress-test each assumption and decide what goes into the real test. They are never counted as evidence.

Signals, live

Recorded. Expected. The gap between them.

Our agents read what actually happened against what the operating model says should have happened. Wherever the two diverge, an alpha gets sized and a stop condition gets written, and only then does anyone build.

Signals — extraction run
READING EXPORT · 5 TABLES
extract.pyassumptions.pysources.sqlMODVEN / SIGNALS
1# Read what the business already records. Write nothing back.
2from modven.sources import Export, Ledger
3from modven.models import SignalExtractor, AssumptionMap
4from modven.alpha import size, grade, StopCondition
5
6sources = Export.load("client_2024-2026.zip")   # structured export, nothing connected
7tables  = ["quotes", "invoice_lines", "shipments", "contracts", "returns"]
8
9extractor = SignalExtractor(window="24M", grain="line")
10ledger    = Ledger()
11assumed   = AssumptionMap.from_operating_model(sources)
12
13for t in tables:
14    recorded = extractor.read(t)            # what actually happened
15    expected = assumed.project(t)            # what the model says should
16    gap      = extractor.diverge(recorded, expected)
17
18    for signal in gap.signals(min_strength=0.6):
19        alpha = size(signal, horizon="90d")
20        alpha.stop = StopCondition.write(signal)   # before anyone builds
21        ledger.add(alpha, grade(alpha))
› export  24M window · 5 tables · 2.3M rows
› quotes  2.1M lines  ·  assumption: price follows list  ·  divergence 0.71
› shipments  184K loads  ·  assumption: freight is billed  ·  divergence 0.83
› contracts  ·  assumption: resets track the index  ·  divergence 0.64
› panel  buyers ×40 · operators ×20 · skeptics ×20 · 3 assumptions failed
› 3 signals above threshold · stop conditions written · next: assumptions.py

The clock

Ten days to know what’s worth testing. Ten weeks to have it running.

Day 0

The NDA is signed. Nothing moves before it.

Day 1

You choose the export: twelve to twenty-four months of the records your team already keeps. Nothing is installed on your side and nothing is connected.

Days 2 to 9

Agents read the export against your operating model and surface the places where behavior disagrees with assumption. Along the way we spend one afternoon with you and the people who run the floor, and every instinct in the room comes out as an assumption with a test attached.

Day 10

You receive the Opportunity ID, signed. The opportunities worth testing, usually three to five, each with a dollar range from your own data, an evidence profile, and its stop condition in writing. If none hold up, the document says so.

Week 3

If one of them deserves a build, the mechanism is architected and the stop condition agreed. The price is fixed and stated in writing.

Week 10

The mechanism is live in your operation. Several builds can run at once, and each one is scaled or killed on what the numbers say.

Guarantees

What is true of every engagement, every time.

Stop condition

Every opportunity carries one, in writing, before anyone builds. It is how we keep the downside capped and the decision to stop never a matter of opinion.

Your staff

Nobody is asked to change how they work unless the OID says so in advance. The mechanism fits the operation, not the other way around.

Price

Fixed, mapped to the outcome, and stated before any money moves. We do not bill by the hour.

Ownership

What we build runs on its own and belongs to you, with no license and no dependency on us.

Your data

Client data is used for the engagement it was sent for and nothing else. We do not keep it beyond the work, sell it, train on it, or share it.

No theater

There is no innovation lab and no workshop series. Every engagement ends with something running in your operation or a written reason it stopped.

The door

Opportunity ID →

The proof

The work →

Start here

Where do you suspect the value is hiding?

Tell us the hunch and what your systems can export. Ten business days later the Opportunity ID comes back signed, with numbers on it.

Run an Opportunity ID →