Seams

1  Seam, not interface

The difference between the two words is the whole point — it decides whether an organisation needs to be steered from outside or steers itself.

In the Beta world, seam and interface are opposites, not synonyms. An interface separates, and it can only be bridged by a third party doing the steering — a superior, a process, a plan coordinating across the gap. A seam, by contrast, connects, and it makes self-steering possible: through a direct agreement between the parties involved.^[1] The term and the concept come from the Austrian consultant Erich Weichselbaum; Niels Pflaeging and Silke Hermann adopted them into the cell structure design of the BetaCodex.^[1]

That seams are the decisive design level at all is not a Beta quirk but a long-standing insight of systems theory. Melvin Conway put it in 1968: organisations that design systems "are constrained to produce designs which are copies of the communication structures of these organizations".^[3] The structure of a product mirrors the communication paths of its builders — place the seams wrongly and the result is bound to be wrong too. Weichselbaum sums it up: "Interfaces separate departments, seams connect teams."^[12]

2  Two types of seam

The organisation is a network of cells (teams), not a pyramid. Between them there are exactly two kinds of coupling.^[4]

3  The internal market

Between the cells there is no budget and no cost allocation, but a market-like mechanism of yes and no.

Centre cells charge the other cells for their services — "on the basis of actual use. That is the only way an internal market comes about."^[5] The centre explicitly has no decision-making power and no power over resources: no budgets, no authority to coordinate, no control. The periphery may say no to any internal service and source it instead from another cell or from the external market.^[5]

The pricing rule

Internal services always carry a price, and that price should "just cover its costs" — "centre cells must not make a profit!" And: "Five to seven services per centre cell are enough."^[6] The org-shop / info-shop pattern structures the centre: org shops deliver organisational services (admin, HR, legal, marketing), info shops informational ones (accounting, finance, internal IT).^[9]

Every cell gets a full profit-and-loss account, deliberately phrased in a different language: no costs, but three kinds of contribution (anything else would be waste).^[7]

* Example split for a total output of 100, taken from Zellstrukturdesign.^[7]

The binding rule behind it: "Periphery cells should generate margin or 'profit'. Centre cells should not."^[8] If a centre cell made a profit, the distribution of power towards the periphery would be undermined — hence margin in the periphery and exact break-even in the centre.

4  The seam agreement

Weichselbaum formalises the seam as a contract-like document — the most concrete expression of the concept.

"In essence, a seam agreement is a simple, written contract between two teams."^[11] On a single sheet, the supplying team records which service it hands over to the next team — the internal customer — in what quality, by when and at which location. What is agreed is a result, not the route to it: how the team produces the service internally is its own business; all that counts is that the agreement holds.

The fields of a seam agreement

The form in the book (the example "Seam agreement No. 1" between two teams) carries exactly these fields:^[13]

Quality is not a column of its own; it sits in the remarks together with a referenced quality guideline (QRL).

In day-to-day operation, seam agreements are binding: "The agreements at the seams are sacred. There are no exceptions."^[14] One failed link stops the whole chain. Whoever takes delivery at the seam carries full responsibility from that point on — "nothing gets pushed back across the seam"; the span of responsibility and the span of influence must coincide. Change happens only when there is a genuine reason: "When it really matters, the parties must renegotiate and reach new agreements."^[17]

In Weichselbaum's model the O.K. point — the most important seam between winning a contract and fulfilling it — additionally sets the cadence for the daily portion: "In a company it is the O.K. point that sets the cadence, not 'somebody'."^[16] The practical evidence: at the office furniture manufacturer Blaha, after self-steering teams with seam agreements were introduced, lead times fell from 20 to 9 days and profitability tripled.^[18]

5  Handshake economics

The BetaCodex patterns describe how cells strike their internal deals — deliberately against the precision logic of cost accounting.

The guiding principle is handshake quality: "All internal agreements between cells should have handshake quality." The point is not accuracy but fairness and collegiality between the cells.^[10] Several patterns follow from it:

6  Negotiation, roles & conflict

Who negotiates seams — and what happens when the parties cannot agree?

Seams are struck directly between the cells concerned, "directly with one another and among themselves" — never via a centre cell. Even the coordination of large, cross-cell orders runs exclusively between the cells involved; a dedicated coordination role is explicitly abolished.^[9] What is negotiated are not posts but roles: "Posts can be handed out, whereas roles are taken." Roles without market contact belong to the centre, roles with market contact to the periphery — and one person can hold both at once.^[4]

The canonical example: Morning Star

At the Californian tomato processor Morning Star (around 600 permanent staff plus some 4,000 seasonal workers), the CLOU — Colleague Letter of Understanding — replaces the job description. Every year, each person negotiates with the roughly ten colleagues they work with most closely, without any approval from a superior. The CLOU records commitments, metrics ("stepping stones") and responsibilities.^[22]

7  Team APIs & the cybernetic foundation

Modern software organisation theory has developed a precise vocabulary for seams — and cybernetics explains why they have to exist.

In Team Topologies (Skelton & Pais, 2019) every team has a clearly defined team API and interacts with others in exactly one of three modes:^[19]

The aim is for teams to be able to deliver "without requiring high-bandwidth communication between teams".^[20] Anyone who takes Conway's law seriously will shape team boundaries deliberately so that the desired architecture emerges — the reverse Conway manoeuvre.^[21] That is seam design, expressed in the language of software architecture.

8  For BetaOS

What the product represents today — and where the seam entity itself is still missing.

BetaOS already represents the internal market: the Value Creation Report shows the three contribution categories per cell, margin in the periphery and break-even in the centre; transactions carry a supplying team, contracts know their internal buyers, and the service catalogues of centre cells can be priced team by team. The couplings are therefore economically visible.

The roadmap follows directly from this article:

• A seam / commitment entity between two cells — with a lifecycle (proposed → agreed → in force → renegotiated), fed by the existing value creation report.
• The seam-agreement fields as a data model, including the quality reference and the handover cadence (analogous to the O.K. point).
• A flow visualisation in the value creation structure: seams as directed edges (value from the inside out, money inwards), not merely as hover lines.
• A lightweight interaction-mode marker per seam (collaboration / as-a-service / facilitating) to make team APIs explicit.

That would turn today's implicit coupling into a named, negotiable and observable seam — precisely the element that distinguishes the BetaCodex from an org chart.