Skip to the content.

Speedboat: Full Guide

Home Docs Guides Ceremonies Templates Setup

Speedboat helps AI-accelerated teams finish meaningful work, not just start more work. Each week, the team limits starts, makes Run visible, and proves what Landed.

Why now

Scrum was designed for a world where writing the code was the expensive part. That world has changed.

AI now materially accelerates how quickly we can understand unfamiliar systems, generate implementation options, build prototypes, refactor, write tests, explore edge cases, and produce demo-ready versions of future functionality. The bottleneck is no longer “how quickly can we write the code?”

AI makes you faster at building the wrong thing, too.

It’s increasingly:

Faster code generation alone does not guarantee faster customer impact. Without a better operating model, AI can create the opposite: more half-finished work, more prototypes, more branches, more demos, more context switching, and more operational drag.

The cost of starting has collapsed. The cost of finishing (judgement, integration, review, rollout, support, learning) has not. That asymmetry is exactly what Speedboat is built to manage.


The shift

Acceleration is easy. Direction is hard.

Speedboat is not about going faster. It is about finishing more meaningful work by starting less, learning faster, and building fewer things at once.

The core question changes:

FROM TO
What tickets are we committing to this sprint? What meaningful thing are we landing this week?

And the cultural shifts behind it:

FROM TO
How much work can we start? What meaningful thing can we land?
Is the ticket done? Who benefited, what changed, what did we learn?
Product writes ideas, Engineering builds them. Product and Engineering use fast Previews to learn together before committing deeply.
AI helps us produce more code. AI helps us learn faster, focus better, and land value sooner.

Team size

A speedboat has few seats by design. The model works best with small teams of 4–5 engineers, where everyone has context, WIP limits feel natural rather than imposed, and the Friday Landing is a genuine conversation rather than a status parade.

Larger groups should split into multiple speedboats rather than scaling ceremonies up. Each boat has its own board, its own rhythm, and its own Landings. Coordination between boats should happen through a lightweight Fleet Sync rather than by scaling the boat-level ceremonies.


Roles

Speedboat uses lightweight roles, not a heavyweight process structure.

The Captain is not a Scrum Master and does not own the backlog. Product Lead is separate from the crew. Facilitation rotates, but accountability for the health of the boat does not.


How AI changes the mechanics

AI doesn’t just make Speedboat possible. It changes how work moves through the lanes:

The operating model creates the structure. AI provides the acceleration within it.

If you are using autonomous or semi-autonomous AI agents, the same rule still applies: they can accelerate exploration, implementation, and triage, but they do not reduce the finishing cost. Review, integration, rollout, and judgement remain human work. See AI Agents in Speedboat.


Core concepts

These are the parts a team needs first.

The model

Preview, Build, and Run are the work lanes. Land is the outcome layer.

Land is not a planning lane where work sits. It is a meaningful outcome: something reached a real beneficiary, or new evidence changed what happens next.

Preview / Build / Run -> Landing

Preview

Quick prototypes, PoCs, or early demos that turn ideas into something tangible.

Build

Production-grade work: turning validated or committed ideas into capability that ships.

What a WIP item means

A WIP item is not a subtask. It is a meaningful unit of work the team is steering: a prototype, a feature slice, a production change, or another outcome-sized piece of work.

The limit applies to active Preview and Build items at that level. A single WIP item may contain multiple tickets, subtasks, or implementation steps underneath it.

Big bets and multi-week Build

Some work will take 6–8 weeks. That is normal.

Speedboat does not require big bets to finish in a week. It does require them to stay visible, steerable, and sliceable.

Treat a big bet as a multi-week initiative with one umbrella outcome, but shape the active work into the next meaningful partial Landings.

Examples of partial Landings:

These are real Landings if they reach a real beneficiary or materially reduce risk, enable downstream work, or make the next slice possible.

If a Build effort runs for more than 2 weeks without a meaningful partial Landing, Route Planning should re-slice it or name explicitly why no such Landing is currently possible.

Landings

Landings are the heart of Speedboat: meaningful outcomes reaching a real beneficiary, or new evidence changing what happens next.

How to interpret Landings

Not all Landings mean the same thing.

A healthy team may produce all four. The important distinction is not to confuse learning or enablement with customer impact.

Run

The honest, visible stream of operational work that keeps the business healthy.


The weekly rhythm

Three weekly steering points keep the boat moving. Zero status updates.

AI accelerates mistakes as efficiently as it accelerates progress. That is why the steering loop has to get shorter.

The default rhythm is Monday / Wednesday / Friday, but teams should shift the cadence to match their working week. The important pattern is start-of-week steering, mid-week course correction, and end-of-week Landing.

When Monday priority is unclear, choose the highest-value Landing you can credibly get to a real beneficiary this week, after making urgent Run explicit. See How to Choose This Week’s Landing.

Ceremony Duration Purpose
Monday: Set Course 30 min Where are we going this week? Choose the intended Landing, confirm WIP, set ownership.
Wednesday: Course Check 15 min Are we still on track? Surface blockers and create permission to course-correct before the week is lost.
Friday: The Landing 30 min What landed, who benefited, what did we learn, and what decisions does each active Preview need?

Route Planning keeps the next few weeks shapeable. It is part of running Speedboat properly over time, but teams can defer the first session if they already have enough shaped work to begin.

Ceremony Duration Purpose
Fortnightly: Route Planning 60 min Keep the next 2–4 weeks shaped enough that weekly steering does not become backlog grooming.

Learning Review and Fleet Sync are periodic or adaptive loops.

Ceremony Duration Purpose
Every 4 weeks: Learning Review 30 min Step back from the week-to-week flow. What is the work teaching us about WIP, Landing mix, Run load, and how the model is working?

For teams running multiple boats, add one extra coordination point:

Ceremony Duration Purpose
Fortnightly: Fleet Sync 30 min Surface cross-boat dependencies, align on shared platform priorities, and flag risks affecting multiple boats.

Fleet Sync is not another planning layer and not a status meeting. It exists only for work that crosses boat boundaries.

The weekly learning slot in Friday’s Landing is tactical: what changed this week and what should we adjust next week? The Learning Review is the slower feedback loop for the operating model itself.

Before the Learning Review, prepare a lightweight monthly stakeholder summary from the Landing Log and the last 4 Weekly Snapshots. It should answer three questions: what landed in the last 4 weeks, what the landing mix is telling you, and what is likely coming next.


Controls

These are the rules that keep the weekly rhythm honest once the team is moving.

Intake flow

New work enters Speedboat through an unshaped backlog.

This is where customer feedback, Sales requests, engineering observations, roadmap items, leadership asks, and other new ideas go before the team decides whether they should become Preview, Build, or Run.

Keep the flow simple:

Between Route Planning sessions, the Product Lead and Captain triage incoming work against current WIP and current commitments.

Only Route Planning should promote non-urgent work from the unshaped backlog into the shaped queue.

Do not quietly inject unshaped work into Build mid-week. New work becomes visible when it enters the system, not when it surprises the team.

Decision rights

Decision Default owner
Intended Landing for the week Product Lead and Captain together
Preview decision: kill / park / continue / promote Product Lead and Captain together
Production readiness for Build Captain / Engineering
Priority of what enters Preview or Build Product Lead
Run prioritisation during incidents Captain

When Product Lead and Captain disagree, they should resolve it explicitly and quickly. If they cannot resolve it within the same working day, the Captain initiates escalation to the next product and engineering leaders rather than leaving work in limbo.

Two rules

Landing mix

Over any 4-week window we expect at least some Customer or Business Landings, unless the team has explicitly chosen a Platform/Run-heavy period. Platform-heavy by accident is a signal to course-correct.

Exception protocol

WIP limits are a steering rule, not a wall. Urgent Run, customer escalations, and leadership overrides are allowed, but the team names what is being paused to make room during the next Set Course or Course Check. The limit applies to meaningful work items, not individual subtasks. The Captain owns enforcing this exception protocol. Urgent Run does not violate Speedboat; hiding it does.


Scaling patterns

These matter when a single-boat rhythm is no longer enough.

Multi-boat coordination

Fleet Sync is optional for a single boat and useful for environments running multiple boats.

Ownership stays lightweight:

Use Fleet Sync to expose cross-boat issues early. Do not turn it into a second Route Planning session.


What Speedboat is not

This matters as much as what it is.

Not “move fast and break things.” Preview can be rough because it’s for learning. Build and Land must meet production standards.

Not one-week Scrum. It’s a steering rhythm for continuous flow, not a sprint commitment under a different name.

Not “ship every Friday or fail.” A Landing can be customer, business, platform, or decision-oriented. The ask is meaningful, not a forced release.

Not a replacement for product strategy. Speedboat is the delivery operating system. Roadmap, customer judgement, and commercial context still matter.


Principles

  1. Start less than you could.
  2. Make Run visible.
  3. Land outcomes, not activity.