Portfolio Control Plane Research concept

One founder. Many products. One control plane.

Where should your next hour and dollar go? The Control Plane turns a portfolio of increasingly autonomous products into a short list of decisions, each with its evidence, confidence and a reversible next step.

§ 01Operating model

From portfolio state to outcomes

01Portfolio
02Ventures
03Metrics
04Recommendations
05Founder decisions
06Resource allocation
07Outcomes

The long-term asset is the loop itself: portfolio state, the recommendation made, the founder's decision, the resources moved, and the outcome observed. Every pass through the loop makes the next recommendation better calibrated.

§ 02Decision vocabulary

Seven decisions, used consistently

Every product carries exactly one current decision. Colour is used sparingly: only KILL is red.

VALIDATE: Demand is unproven. Run small, reversible tests until there is a signal, or stop.; Minimal, time-boxed attention; no fixed costs.
BUILD: Demand is validated, but the product is not yet sellable at scale.; Concentrated engineering and agent capacity; little marketing spend.
SCALE: Extra attention or capital earns the highest marginal return in the portfolio.; More founder hours, distribution spend and agent capacity.
MAINTAIN: Stable and healthy. Keep it running with minimal intervention and protect its autonomy.; Monitoring only; act on alerts.
HARVEST: Growth is flat but cash is positive. Reduce attention and extract cash flow.; Cut discretionary spend; automate or drop remaining manual work.
SELL: Worth more to another owner than to your portfolio's scarce attention.; Prepare transferable documentation; reduce your involvement before the sale.
KILL: Consumes attention with no credible path to return. Shut it down cleanly.; Migrate or notify customers, then stop all spend.
Rule: One current decision per product. A product without an explicit decision silently becomes MAINTAIN.

§ 03Interface preview

Decisions first. Charts second.

The primary screen is not a graph. It is what needs your attention, ranked, with the evidence one click away.

Illustrative preview · hypothetical portfolio · not real data

Needs your attention

Product A

SCALE

Growth (30d): +18%
Confidence: 87%

Why

Organic acquisition accelerating
Churn low and stable
Highest marginal attention ROI in the portfolio

Recommended allocation

+6 founder hours / week
+$500 / month content

Review evidence

Evidence against

Growth concentrated in one acquisition channel
Only three months of post-launch history

Reversible experiment

Two-week pricing test on new sign-ups only; revert if trial conversion drops more than 15%.

Product D

SELL

Maintenance tax: 41%
Attention share: 27%

Why

Maintenance tax 41% and rising
Founder burden rising: 12 h / week, mostly support
Growth flat for two quarters

Recommended allocation

−8 founder hours / week after handover
Document operations for a buyer

Review evidence

Evidence against

Small but loyal customer base
Sale multiples for sub-$3k MRR products are uncertain

Reversible experiment

List privately with a broker for 30 days; reversible because no customer-facing change is made.

Product C

SCALE

Growth (30d): +9%
Confidence: 62%

Why

Lowest maintenance tax (11%)
Runs on 3 founder hours / week

Recommended allocation

+3 founder hours / week for customer interviews

Review evidence

Evidence against

Demand signal comes from one segment; needs customer interviews before adding spend

Reversible experiment

Interview eight customers in the top segment before committing marketing spend.

Portfolio summary

44h week

Illustrative portfolio summary
Product	Decision	MRR	Founder h/wk	Attention	Maint. tax	Growth 30d
Product A	SCALE	$15.2k	8h	18%	14%	+18%
Product B	HARVEST	$8.1k	9h	20%	22%	+1%
Product C	SCALE	$6.4k	3h	7%	11%	+9%
Product D	SELL	$2.1k	12h	27%	41%	-2%

This week

01
Product A — Pricing experiment
Highest marginal attention ROI
02
Product C — Customer interviews
Validate the segment before spending
03
Product D — Consumes 27% of founder attention
12 of 44 founder hours per week

§ 04What it allocates

Six scarce resources

Founder attention: Hours per week, by product and by type of work.
Capital: Cash for distribution, tooling and experiments.
AI spend: Model and API budgets per product and workflow.
Engineering capacity: Founder build time and coding-agent throughput.
Marketing & distribution: Channels, content and paid acquisition.
Agent capacity: Which agents run where, at which autonomy level.

§ 05Positioning

Not another multi-product dashboard

Seeing every metric in one place is a commodity. The Control Plane does portfolio resource allocation and decision intelligence.

Analytics dashboard compared with the Portfolio Control Plane
	Analytics dashboard	Portfolio Control Plane
Primary output	Charts of every metric	A ranked list of decisions
Question answered	What happened?	Where should the next hour and dollar go?
Unit of analysis	Metric	Product × resource allocation
Uncertainty	Hidden in averages	Confidence, evidence for and against
Action	Left to the founder	Reversible experiment proposed, decision logged
Memory	Time series	State → recommendation → decision → outcome

§ 06Methodology v0.1

Signature metrics

Proposed definitions that will evolve with use. They are not universal standards.

Autonomous MRR

MRR from products that run below a defined founder-intervention threshold.

v0.1

Autonomous MRR = Σ MRRₚ for every product p where hₚ ≤ T

MRRₚ: Monthly recurring revenue of product p
hₚ: Founder hours per week on p's recurring operations (trailing 4-week average; excludes strategy)
T: Intervention threshold; v0.1 default 2 hours per week

The share of revenue that keeps arriving if the founder looks away. Track it as a share of total MRR.

— Sensitive to the threshold T; always report T alongside the value.
— Hours must be logged, not estimated from memory.

Full definition →

Maintenance Tax

What it costs to keep a product running, as a share of its revenue.

v0.1

Maintenance Tax = (Hₘ × r + infrastructure + software + AI + contractor maintenance) ÷ revenue

Hₘ: Founder maintenance hours in the period (fixes, support, ops — not new features)
r: Opportunity cost of a founder hour, set explicitly by the founder
revenue: Revenue of the product in the same period

Rising maintenance tax with flat growth is the classic HARVEST or SELL signal.

— The founder-hour rate r is a judgement; publish it with the metric.
— Exclude one-off migrations, or report them separately.

Founder Attention ROI

Incremental portfolio value created per founder hour invested.

v0.1

Attention ROI = Δ portfolio value ÷ founder hours invested

Δ portfolio value: Change in value attributable to the work (e.g. Δ MRR × an explicit multiple, or Δ cash flow)
founder hours: Hours invested in the product or initiative over the same window

Allocate the next hour to the product with the highest marginal, not average, attention ROI.

— Attribution lags: value often arrives weeks after the hours.
— Average ROI flatters mature products; use marginal estimates for allocation.

Full definition →

Attention Concentration

Share of founder time consumed by the most demanding venture.

v0.1

Attention Concentration = maxₚ(hₚ) ÷ Σ hₚ

hₚ: Founder hours on product p in the period (all work types)

High concentration is fine on a SCALE product, and a warning sign on a HARVEST or SELL product.

— Read it with the decision label: concentration is a choice, not a defect.

Portfolio Autonomy Rate

Share of recurring portfolio operations that run without direct founder intervention.

v0.1

Portfolio Autonomy Rate = recurring operations completed without founder action ÷ all recurring operations

operations: Recurring workflow runs across all products (billing runs, deploys, ticket resolutions, reports…)

The portfolio-level counterpart of the Zero-Employee Score's workflow automation dimension.

— Needs workflow-level instrumentation; until then, estimate per product and label it an estimate.

§ 07Principles

Read-only before autonomous

Read-only intelligence comes before any execution. The first version connects to data; it never moves money or changes production.
Every recommendation carries confidence, evidence for and against, and a reversible experiment.
Founder decisions are logged with their outcomes, so the system learns from what actually happened.
Founder attention is treated as the scarcest economic resource in the portfolio.

Start with one company: the Zero-Employee Architect How to manage multiple SaaS products today

Research cohort · early access

The Control Plane is a research concept, not a live product. Founders running several products can join the research cohort that shapes it, and reserve a plan at its intended price. No payment is taken.