← Soft MeasuresEthos

How we work.

Soft Measures is a small team that builds and operates a portfolio of software companies. The companies serve different markets, but they are built and run the same way. This page describes how.

01Build to own

01 · Build to own

We build companies and continue to operate them.

Soft Measures is a studio and a holding company. We start businesses, ship them, and continue to operate them. The goal for each company is steady operation and revenue over many years. Operating what we build gives us direct feedback on whether it works.

Long holding period. Each company is expected to operate and cover its costs for years.

Direct feedback. We operate these companies ourselves, so problems in the software surface quickly.

Independent companies. Each company stands alone. Shared infrastructure benefits all of them.

02One engine

02 · One engine

Everything runs on the same engine underneath.

Every company in the portfolio is built on EightMind, our multi-agent framework, and the shared kit of capabilities that comes with it. A new venture starts with RAG grounding, a report engine, memory, human-approval gates, and daily automation already in place. This is what allows a small team to run eleven companies at once.

Inherited infrastructure. New companies start with existing tooling instead of building their own.

Teams of agents. Each agent has a defined responsibility and works within its own domain.

Shared improvements. Fixes and improvements made for one company apply to the rest.

03What we take on

03 · What we take on

Established markets with little modern software.

We work in markets such as surgical supply, home-services leads, grant writing, local SEO, and career verification. Demand in these markets is established, and most incumbents run on manual processes. Within them, we look for problems where a technical or operational advantage is possible and where agents can carry a meaningful share of the work.

Established demand. People in these markets already pay to solve these problems.

A technical advantage. The problem must be one that better software can solve.

Work agents can run. The work has to be something agents can carry out on an ongoing basis.

04Human approval

04 · Human approval

Consequential actions require human approval.

Agents work autonomously up to the point of consequence. Any action that affects a live system, such as a payment, a customer message, a published page, or a trade, waits for human approval. This rule is implemented in the architecture rather than in prompts, and it remains in place as the portfolio grows.

Autonomy with limits. Agents work independently and stop for approval before consequential actions.

Part of the architecture. The approval requirement is built into the system, not into a prompt.

Auditable. Every agent action is recorded and can be reviewed.

05Shipping

05 · Shipping

We ship in short cycles and test before release.

We do not plan in multi-quarter roadmaps. Work is scoped so it can ship in the near term, and quality checks run before release. We use established patterns and existing tools where they fit, and write new code only when nothing suitable exists.

Near-term scope. If a task cannot ship in the near term, we reduce its scope until it can.

Reuse first. Existing, widely used tools are preferred over new code.

Tested before release. Automated checks run before anything reaches a customer.

Contact

Work with us.

We partner with operators who have deep knowledge of a specific market. The partner brings the domain expertise; we build and operate the software.

Get in touch See the portfolio

All conversations are confidential.