TEAF Light for a company just starting out: one process, one loop, one objective

Enterprise architecture is spontaneously associated with large organisations: repositories, boards, maps of several hundred applications. That is a perspective error. A three-person company makes exactly the same structural decisions as a group — where data lives, what is outsourced, what it refuses to automate — only faster, and without writing them down. TEAF Light exists for that situation: the viable minimum for those decisions to leave a usable trace.

The 1-1-1 principle

One process. One loop. One measurable objective. Not two, not seven.

• One process: the one carrying the value and hurting today. For a service business, usually "from inbound request to cash collected". For production, "from order to compliant delivery".
• One loop: Intent → Capability → Knowledge → Decision → Automation → Execution → Observation → Learning. Applied to a single process, it runs in weeks, not a year.
• One measurable objective: a figure you can read without an instrumentation project — average response time, quote conversion rate, days between delivery and invoice.

The principle is developed in TEAF Light and the 1-1-1 principle.

Why a loop rather than a plan

A plan assumes you know. A company just starting out does not yet: its offer moves, its target customer sharpens through rejections, its model self-corrects. A short loop produces knowledge where a plan produces delay — see the TEAF loop, component by component. The gain is sharpest precisely when uncertainty is highest, which is year one.

The four statuses, applied to a small structure

TEAF separates what is declared, observed, inferred and validated. The distinction looks bureaucratic until the day it prevents an expensive decision. In a micro-business it reads like this:

• Declared: "our customers come from word of mouth." That is what the founder says.
• Observed: of the last twelve deals, nine came from a referral, three from online search. That is what the data shows.
• Inferred: "so investing in the website isn't a priority." That is a hypothesis, not a fact.
• Validated: the hypothesis was tested over a quarter against a criterion set in advance.

Most expensive early decisions are inferred items treated as validated ones.

The first Living ADR

A Living ADR is a written decision carrying its own expiry: context, options, choice, consequences, and the signal that will force it open again. For a company starting out, three are enough: where data lives and who holds it, what stays reversible, what you refuse to automate this year. Those three pages prevent the discussion that returns every six months and the choice quietly contradicted — the mechanism is detailed in the Living ADR.

What it actually prevents

The full rebuild of the information system at the first growth step. The cause is almost always the same: tools bought before capabilities, data with no authoritative home, and untraced decisions. The execution side is described here: building an architecture that doesn't have to be rebuilt, then the technical foundation and capacity allocation. TEAF Light templates are published under the MIT licence: any organisation can take them without asking anyone. And for five companies being created, the loop is set up with them in the 5 Founders programme.