Value Stream Mapping

From Flow to Outcomes

In the previous post we examined the physics of flow — how queuing theory reveals the hidden dynamics governing work as it moves through networked organisations.

We saw that every organisational boundary is a queue. That high utilisation causes exponential degradation. That protocols function as queue disciplines, managing arrival rates, service rates, and priority.

But understanding flow dynamics alone does not tell you what to coordinate around.

A firm can have beautifully designed APIs, event systems, and workflow engines — and still produce fragmented outcomes, because the protocols were organised around internal structures rather than the flow of value to the customer.

Protocols define how participants interact. Value streams define why.

Value Stream Mapping — originally from Lean manufacturing — offers a way to align protocols, architecture, and organisation around a single question:

What is the sequence of activities required to deliver value to the customer?

The Three Forces, Revisited

Before exploring value streams, it is worth restating the three forces from Posts 10 and 11 — because value stream mapping addresses all three simultaneously.

These forces create a tension. Without careful design, scaling a networked firm leads to complexity collapse — where coordination cost overwhelms the value of additional connections.

Value stream mapping offers a way through.

What Value Stream Mapping Does

Instead of mapping teams, systems, or org charts, value stream mapping traces the path from a customer request to a delivered outcome.

The emphasis is on flow — the sequence of transformations that produce value.

Every step that does not contribute to the outcome is waste. Every hand-off is a potential friction point. Every waiting state is latency in the network.

Why This Matters for Platforms

Platforms typically evolve around features or teams.

A recruitment platform, for example, might organise itself around:

• ATS Team
• Marketplace Team
• Payments Team
• Messaging Team
• Compliance Team

This is Conway’s Law in action — the system mirrors the organisation.

But customers experience value streams, not features.

A hiring manager does not think in terms of “ATS” and “Payments.” They experience a flow:

Value stream mapping forces you to design around this flow instead of internal structures.

Countering the Three Laws

Conway’s Law: Align Architecture to Outcomes

If you design systems based on value streams, teams naturally organise around them.

Instead of:

You get:

Architecture follows customer outcomes, not internal silos. Conway’s Law works for you instead of against you.

Brooks’s Law: Create Stream Boundaries

Brooks’s Law emerges from unconstrained coordination complexity — everyone needing to coordinate with everyone.

Value streams create clear boundaries. Each stream becomes a semi-autonomous system. Communication between streams flows through defined interfaces rather than ad-hoc meetings.

Communication becomes:

Stream A → Protocol → Stream B

instead of:

Everyone → Everyone

This dramatically reduces coordination overhead.

Metcalfe’s Law: Define Protocol Primitives

Metcalfe’s Law says networks become valuable when connections increase. But connections must be low friction.

Value streams define standard interaction points — the places where participants connect:

submit_candidate(vacancy)
schedule_interview(candidate)
accept_offer(candidate)
submit_timesheet(contract)

These become protocol primitives. Participants can connect through them without negotiating each interaction. Network effects emerge because the cost of joining the network falls.

Value Streams Reveal the Real Platform

Many companies believe they run one platform.

Value stream mapping often reveals multiple overlapping platforms, each with different actors, workflows, and scaling dynamics.

Each stream has different participants, different protocols, and different network geometries. Treating them as one system is a Conway’s Law trap.

Exposing Friction

A key goal of value stream mapping is identifying waste — the steps that add latency without adding value.

Common friction points:

• Waiting: manual review queues, email-based scheduling
• Hand-offs: data re-entry between systems, context lost between teams
• Rework: rejected candidates resubmitted, contracts revised after signing
• Manual steps: processes that could be automated but require human intervention

Each waiting state is friction that limits network scaling. These are the points where protocols can replace manual coordination — and where AI agents can participate most effectively.

From Value Streams to Event Architecture

Once mapped, value streams naturally become workflow graphs. Each state transition becomes an event, an API endpoint, and a protocol interaction.

Each event can trigger downstream services:

Services remain loosely coupled. The value stream provides coherence. The protocol layer provides the interaction rules. The events provide the execution mechanism.

Four Levels of Design

Value stream mapping operates at multiple levels of abstraction, each deriving from the one above.

Architecture that derives from value streams is architecture that serves outcomes.

This is the inverse Conway manoeuvre — instead of letting organisation shape architecture, you let outcomes shape organisation.

Why This Matters for AI

AI agents operate best in structured workflows.

Value streams provide exactly what agents need:

• Tasks with clear inputs and outputs
• States with defined transitions
• Data that flows through the stream
• Validation rules at each step

Without structured flows, agents hallucinate, diverge, and produce unreliable outputs. With value streams, agents can participate as nodes in the workflow graph — executing steps, triggering transitions, and routing work through the protocol layer.

Recent empirical work supports this directly. Kim and Liu’s research on scaling agent systems found that task decomposability is the strongest predictor of whether multi-agent coordination improves or degrades performance. Parallelisable tasks benefit enormously from distributed agents; sequential tasks degrade when agents fragment the reasoning process. Value stream mapping reveals exactly this structure — which steps are parallelisable, which are sequential, where hand-offs occur — making it the natural design tool for deciding where agents should participate and which coordination topology they should use.

It is no coincidence that Google’s A2A protocol launch used a hiring workflow as its canonical example — candidate sourcing, background checks, consolidated recommendations — with specialist agents coordinating along the stream. The value stream is the natural coordination unit for agent networks.

This connects directly to the Cyborg Cell from Post 9. The cell operates within a value stream. The human anchor defines the outcome. The agents execute the flow. The protocol layer governs the interactions.

The Minimum Viable Network

Value stream mapping answers a question that network geometry alone cannot:

What is the minimum network required to reliably produce this outcome?

Once you know that, you can design:

• Protocols around the interaction points
• Architecture around the event flows
• Teams around the stream boundaries
• Automation around the friction points

The value stream becomes the organising principle that aligns all four.

This is the deeper insight: value streams are not just a mapping technique. They are the coordination grammar of the protocol firm — the structure that tells protocols what to coordinate, tells architecture what to build, and tells teams what to own.

References & Intellectual Lineage

• Womack, J. & Jones, D. (1996). Lean Thinking — value stream mapping methodology.
• Rother, M. & Shook, J. (1999). Learning to See — canonical VSM guide.
• Conway, M. (1968). “How Do Committees Invent?”
• Brooks, F. (1975). The Mythical Man-Month.
• Metcalfe, R. (1980). Network value and telecommunications economics.
• Skelton, M. & Pais, M. (2019). Team Topologies — stream-aligned teams.
• Kim, J. & Liu, M. (2026). “Towards a Science of Scaling Agent Systems.” Google Research.
• Post 9 in this series: The Hybrid Topology — the Cyborg Cell.
• Post 10 in this series: Coordination density and the three laws.
• Post 12 in this series: The Protocol Layer.