Cockpit
    October 13, 2026 · 17 min

    Shape Up vs. Scrum — an improvement from my perspective

    Shape UpScrumAgileLeanMethods

    If you have hit the limits of Scrum, it is worth taking a serious look at Shape Up.

    Shape Up was developed by Ryan Singer and published in 2019 as the book "Shape Up: Stop Running in Circles and Ship Work that Matters." The method grew out of how work is actually done at Basecamp (today 37signals), where Singer shaped it over years of day-to-day product development.

    The further a company moves towards agentic AI in software development, the more the Scrum process changes as well.

    It is therefore worth looking at other ways of thinking and other methods too.

    Updated October 2026: All 7 dimensions have been extended with the impact of agentic software development with AI agents.

    I want to start with a short assessment. After that you will see the differences laid out side by side.

    At a glance: the key differences

    • Work queue: Scrum runs on a permanent product backlog; Shape Up uses a few clearly framed bets per cycle — no backlog.
    • Time model: Scrum uses continuous sprints (one month or less); Shape Up uses 6-week cycles plus a 2-week cool-down.
    • Scope: In Scrum, work is estimated and planned into a sprint; Shape Up reverses this: the appetite fixes the time, scope gets cut ("fixed time, variable scope").
    • Stopping: Shape Up kills projects that do not deliver by default (circuit breaker); the Scrum Guide has no such mechanism.
    • Portfolio level: Shape Up includes the decision level above teams (pitches, Betting Table); the Scrum Guide is silent on it.
    • Roles and events: Scrum defines three accountabilities and five events; Shape Up deliberately defines neither roles nor mandatory ceremonies.

    Part 1: My assessment

    Shape Up offers real advantages. You can take inspiration from the following elements and put them into practice with simple measures — without turning your existing organisation upside down.

    Bets instead of a backlog

    Shape Up works with bets instead of a backlog. I consider this a very significant improvement over the Scrum Guide.

    In the agile transformations I have helped to shape, we also worked from hypotheses or had some kind of business case. Both are close to the idea of a bet.

    A backlog quickly fills up with things that are not relevant. You easily lose focus, you may end up building things that were not necessary, and you drift away from the Lean philosophy, which takes care to reduce waste across the eight types of muda.

    Few people know this, but Scrum is founded on Lean thinking — it says so right on page 4 of the Scrum Guide.

    Shape Up helps you return to the roots of Lean and concentrates on bets. Their tight frame contains the work sharply and so helps keep waste out from the start.

    Relevance for software development with AI agents

    In some organisations, AI agents already create backlog items on their own. That in turn gives people the opportunity to think upfront about why something should be built at all — currently discussed under the term "judgement".

    So you could try to think in bets already. In most companies it will not work one-to-one as in Shape Up, because many work on large, complex initiatives with many systems and legacy, which makes it hard to think in an end-to-end outcome. But at least you can start thinking about how to establish this way of working.

    Cycles instead of sprints

    Another step forward is the cycle. We all know "waterfall in sprints" and the jokes about it — and the jokes come from real experience: teams work forever, in waterfall-like phases, on things that may never deliver the intended value.

    In Agile, working software and satisfied customers were the yardstick. In very large and complex initiatives, the solution may work at the end of every sprint, but the customer is often miles away. That in turn leads teams to fall back into waterfall-like behaviour. Shape Up prevents this.

    Fixing work to set cycles in Shape Up prevents exactly that and minimises the risk of waterfall-style development.

    Relevance for software development with AI agents

    The moment AI agents take over large parts of software development, much more effort automatically goes into defining the architecture and the governance guardrails upfront.

    This means different phases of software development emerge, and they may not all be worked through tickets at the same speed as in the steady Scrum rhythm so far.

    If I suspend the sprint construct and think in cycles instead, I gain more freedom to let these phases play out.

    The challenge is to keep a sensible cadence across the company. Especially when several teams have to integrate with one another, you need a shared rhythm.

    If everyone has been working in sprints and periods so far, I cannot simply announce that everyone now works in cycles. But you can certainly start thinking about which rhythms add up to a sensible cadence and which teams may already work in cycles.

    Small, dedicated teams with focus

    In the Shape Up model, teams are smaller and dedicated to one bet.

    In Scrum teams, reality often looks different: teams are too big, not truly cross-functional, not fully empowered to make all decisions on their own, and pulled across many projects at once.

    In most companies, a complete Scrum implementation has rarely happened — except in those that took the agile transformation seriously and combined Scrum with other methods from the start.

    Relevance for software development with AI agents

    In some organisations, agentic software development has already made teams smaller. The most positive changes are those that bring the user/customer/citizen closer to the developer, so that teams truly become cross-functional.

    When it comes to focus, almost every organisation still has a lot of room for improvement: too many things are often done in parallel, which leads to cognitive overload on the one hand and delayed ROI on the other (keyword: Little's Law).

    Shape Up's idea of small, dedicated teams working with focus on one bet should be adopted for several reasons: you learn faster, you overload people less, and you realise the return on investment faster and more often.

    Outcome instead of tickets

    There are organisations that hold every event mandated by the Scrum Guide, yet the intended effect never materialises, because following the process does not guarantee that the philosophy behind Agile is actually lived.

    Shape Up focuses more on projects, on outcome.

    Scrum, when it is not reflected on carefully, can quickly lead people to focus more on the process than on the outcome.

    There are organisations where engineers just work off a heap of tickets and lose sight of the bigger picture.

    Shape Up prevents this from the start by putting the focus on bets, so that a team even has the chance to work cross-functionally on a piece of work with outcome and value.

    Many Scrum teams never get that chance.

    Relevance for software development with AI agents

    The freedom to shape the process and the meetings yourself can become a real game changer in agentic software development. Architecture and governance questions may need different meetings and a different process than the actual creation of the software by AI agents.

    The focus on outcome is always the right one.

    Kill criteria from day one

    There are organisations that introduced Scrum but never trained their leaders on what Agile would change for them.

    The consequence: no one defines kill criteria or stop conditions. Outcome is not measured with hard facts and data, so there is no basis for deciding why something should be stopped.

    There are organisations that did understand this and ran their agile transformation with exactly these guardrails. But that is not the case everywhere.

    Shape Up is a good alternative for a second attempt. Shape Up has several levers that ensure not too much time and energy is spent on initiatives that deliver too little business value. There is no runaway train — the situation where something gets extended simply because no one defined how to stop it.

    Der Zug hat keine Bremse ("The train has no brakes") is a German party hit. And I sometimes think of it when companies have no method for stopping initiatives that are not delivering the desired value — because the desired value was never defined clearly enough upfront, and no one defined the grounds on which decision-makers may stop.

    The moment I understand kill criteria as part of the way we work, I can stop without anyone losing face.

    Shape Up is an ideal alternative for learning exactly that approach.

    Relevance for software development with AI agents

    If I set it up correctly, I am already defining kill criteria indirectly: everything AI agents build that does not meet the architecture and governance requirements inevitably has to be reworked or discarded.

    All that is missing is the additional step of also defining kill criteria from a business perspective.

    So Shape Up helps you think through, even more consistently, what should already be happening with agentic AI anyway.

    Cutting scope, not quality

    Scrum, taken seriously, already states that quality is binding — so scope inevitably has to become variable.

    And that logically means scope actually has to be reduced regularly in order to deliver an increment of quality every sprint.

    In practice this rarely happens, because it is not exactly intuitive.

    Shape Up is lighter-weight here and forces the team to deliver something within the cycle that fulfils the bet.

    As a result, scope is automatically cut smaller while quality stays constant. In other words, the tight boundaries help teams learn the counter-intuitive work of scope reduction.

    Relevance for software development with AI agents

    The more rigorously quality criteria are defined upfront, the less waste I produce through AI agents. If AI agents develop extremely fast, I would no longer need to limit scope at all.

    But who can review all of that? And which results are actually good? So I have to reduce scope again, so that I do not create new bottlenecks in quality assurance afterwards. Shape Up's focus on quality is therefore very helpful when working with AI agents.

    Factoring in the flight level above

    The Scrum Guide does not prescribe a scaling framework. That is why things like the Spotify model or SAFe emerged.

    Shape Up includes the next level from the start and calls it the Betting Table.

    In the agile transformations I worked in that really brought change, the level above the teams was also designed in from the very beginning.

    Because that is the lever for not doing too many things at once, for course-correcting, for stopping.

    It is always necessary to think about this level — and, if possible, the one above it.

    Shape Up helps enormously here. And it is lighter-weight than SAFe.

    Relevance for software development with AI agents

    Organisations that have not yet developed a way of discussing what teams should actually be working on will notice very quickly, when introducing AI agents into software development, that this is missing. The bottleneck moves one level up. If I can develop at breakneck speed, what should be developed at all? Decisions could pile up if I do not yet have the decision-making body.

    For these organisations it is worth taking a look at Shape Up and drawing inspiration from the Betting Table. Everyone else will already be living it in one form or another.

    Limits of Shape Up and prerequisites

    Mixed types of work in teams

    Shape Up will be difficult to implement in teams that do not only develop products or software, but also handle lots of requests, daily business, maintenance, operations and service requests.

    Shape Up alone would not be enough for that; you would need additional methods.

    Several teams integrating their deliverables

    I also see a major challenge when several teams are needed whose outcomes build on each other or have to be integrated with one another.

    These questions have to be thought through upfront, so that the bets the different teams deliver do not just hang loosely in the air but can be integrated with one another.

    For this, you would have to look very closely at each company to see which tailored frameworks need to be developed in addition so that everything interlocks.

    Rigid hierarchies outside the teams

    Shape Up will also not work where hierarchical structures or reporting lines make implementation impossible, and where leaders have not yet developed their own vision of agile working.

    Agentic software development will make these problems visible as well.

    The ground has to be prepared through active engagement with the philosophy behind it — which, in turn, is rooted in Lean and has evolved over decades.

    And for the projects that are to be run with Shape Up, hierarchies and reporting lines have to be adjusted so that they do not counteract it.

    Top management is needed

    Lean was never a grassroots movement; it was always introduced by top management. That is why it is important for organisations exploring Shape Up to discuss, starting with top management, how they understand it and how they want to apply it — because thinking in bets changes the work of leaders.

    The same applies to software development with AI agents. The bottlenecks move into the phase before development and the phase after development. That brings new tasks and questions for leaders — questions teams cannot answer on their own.

    In Lean practice in the automotive industry, war rooms are popular — Obeya in Japanese. It is worth considering whether something like this, literally or figuratively, is needed during the transformation phase to give a new method like Shape Up the ground it needs to thrive.

    Anyone facing these questions needs an AI consultant with a strong background in software development, agile working and the Lean philosophy, as well as in the latest trends in working with AI agents. Ideally, the consultant also has leadership experience and can think through the leadership and portfolio level.

    Personally, I find that Shape Up has many good approaches that can also be applied in large companies.

    In this article I have only touched on a few points; you can read more about Shape Up here: basecamp.com/shapeup

    Part 2: Overview — Shape Up vs. Scrum Guide (2020)

    Sources: Shape Up (Ryan Singer) · Scrum Guide 2020 (Schwaber/Sutherland). Only statements that can be supported directly by these two sources. Where one source says nothing on a topic, this is marked as such.

    1. Character of the method

    Scrum Guide

    • defines itself as a lightweight and deliberately incomplete framework
    • declares itself immutable: implementing only parts of it, according to the Guide, is not Scrum
    • describes no concrete techniques; states that tactics are context-dependent and described elsewhere

    Shape Up

    • a book that describes how Basecamp works
    • contains concrete techniques (e.g. breadboarding, fat marker sketches, hill charts)
    • the "Adjust to Your Size" appendix explicitly separates core truths from specific practices, i.e. adaptation is expected

    2. Work inventory & selection of work

    Backlog

    • Scrum: the Product Backlog is the single source of work for the Scrum Team; emergent, ordered, continuously refined
    • Shape Up: the chapter "Bets, Not Backlogs" — there is intentionally no central backlog; instead there are decentralised lists and a small number of potential bets per cycle. The book's reasoning: important ideas come back on their own

    Deciding what gets built

    • Scrum: the Product Owner orders the Product Backlog; the Developers pick items from it during Sprint Planning
    • Shape Up: the Betting Table — a meeting during cool-down in which stakeholders decide which pitches to bet on for the next cycle

    Preparing the work

    • Scrum: Refinement as an ongoing activity (breaking items down, adding detail, order and size)
    • Shape Up: Shaping as its own track running in parallel with building; the output is a pitch with five components: problem, appetite, solution, rabbit holes, no-gos

    3. Time model

    Cycle length

    • Scrum: Sprints of one month or less; a new Sprint starts immediately after the previous one
    • Shape Up: six-week cycles plus two weeks of cool-down between cycles

    Cool-down (Shape Up only)

    • time for ad-hoc work, bug fixes and the Betting Table
    • the Scrum Guide contains no comparable element

    Handling scope and time

    • Shape Up: "Fixed time, variable scope" — the appetite defines how much time you want to spend, instead of estimating how long something will take
    • Scrum: the Developers decide how much fits into a Sprint; the Guide names past performance, capacity and Definition of Done as the basis for the confidence of the forecast; the Developers are responsible for sizing items

    4. Focus, interruptions & parallel work

    Shape Up

    • according to the glossary, a bet is the decision to dedicate a team to a project for one cycle without interruption, with the expectation to ship
    • an explicit "Uninterrupted time" section in the chapter on the Betting Table
    • "Keep the slate clean": the next cycle starts without carried-over commitments
    • bugs are not pulled into a cycle; they are handled during cool-down (section "What about bugs?")

    Scrum

    • the Scrum Team is one unit focused on one objective at a time: the Product Goal; it must be fulfilled (or abandoned) before the next one begins
    • no changes are made during the Sprint that would endanger the Sprint Goal
    • working in Sprints at a sustainable pace, according to the Guide, improves focus and consistency
    • the Guide does not regulate whether people or teams may work on multiple products/projects in parallel; the topic is not covered

    Team composition

    • Scrum: one Scrum Team of typically 10 people or fewer; when multiple teams work on the same product, they share Product Goal, Product Backlog and Product Owner
    • Shape Up: project teams of one designer plus one to two programmers; teams are staffed per cycle/project — one of the questions at the Betting Table is "Are the right people available?"

    5. Roles

    • Scrum: three defined accountabilities — Product Owner, Scrum Master, Developers; the Product Owner is one person, not a committee
    • Shape Up: defines no roles in the sense of the Scrum Guide
      • describes activities and groups of people: shapers (senior people, according to the book), participants at the Betting Table, building teams
      • a process role comparable to the Scrum Master does not appear in the book

    6. Events

    • Scrum: five events — Sprint (container), Sprint Planning (max. 8 h), Daily Scrum (15 min), Sprint Review (max. 4 h), Sprint Retrospective (max. 3 h); time boxes apply to a one-month Sprint
    • Shape Up:
      • describes the Betting Table during cool-down as the recurring meeting
      • Daily, Review and Retrospective do not appear in the book
      • status communication runs via hill charts, section "Status without asking"

    7. Planning during execution

    Scrum

    • Sprint Backlog = Sprint Goal + selected items + plan; continuously updated during the Sprint
    • items are often broken down into work units of one day or less; how this is done is decided by the Developers alone

    Shape Up

    • teams are assigned projects, not tasks ("Assign projects, not tasks")
    • the book distinguishes "imagined tasks" (thought of up front) from "discovered tasks" (found during the work)
    • work is organised into scopes — parts of the project that can be built, integrated and finished independently
    • nice-to-haves are marked with "~" and are dropped when time runs out

    8. Non-delivery & termination

    • Shape Up: circuit breaker — projects that do not ship within a cycle are not extended by default; they end. Extension is the exception (section "When to extend a project")
    • Scrum:
      • a Sprint can be cancelled if the Sprint Goal becomes obsolete; only the Product Owner has this authority
      • items that do not meet the Definition of Done go back to the Product Backlog
      • a default mechanism that ends initiatives does not exist in the Guide

    9. Completion & quality

    • Shape Up: "Done means deployed" — done means shipped within the cycle
    • Scrum:
      • Definition of Done as a formal description of the quality state of an Increment
      • increments can be delivered at any time; the Sprint Review is explicitly not a release gate

    Handling growing scope

    • Shape Up: scope hammering — design, implementation or use cases are questioned to finish inside the fixed timebox; the book states: "Cutting scope isn't lowering quality"
    • Scrum: scope can be clarified and renegotiated with the Product Owner during the Sprint without endangering the Sprint Goal; quality does not drop

    10. Theoretical foundation

    • Scrum: grounded explicitly in empiricism and Lean thinking; three pillars (transparency, inspection, adaptation) and five values (commitment, focus, openness, respect, courage)
    • Shape Up: contains no comparable theory or values chapter; as recurring motives the book itself names, among others, targeting risk and handing responsibility to teams

    11. Portfolio level (Flight Level 3)

    • Shape Up describes a recurring decision mechanism above the team level: pitches → Betting Table → one bet per cycle; together with the Betting Table questions (Is the problem relevant? Is the appetite right? Is the solution attractive? Is now the right time? Are the right people available?)
    • the Scrum Guide describes no level above the Scrum Team or above multiple teams working on one product; scaling is only mentioned in that multiple teams on the same product share Product Goal, Backlog and PO
    • neither source uses the term "portfolio"

    Frequently asked questions about Shape Up vs. Scrum

    What are the advantages of Shape Up compared to Scrum?

    The advantage of Shape Up is its focus on the outcome, the tight boundaries that force decisions, the discipline to follow through within the agreed framework, and small teams that can work with focus. The possibility that something may be discarded is built in from the outset, so people have not failed; instead, the method creates a framework for stopping initiatives without anyone losing face.

    Why are bets better than a backlog?

    A backlog quickly fills up with things that are not relevant. You lose focus, you may build things that were not yet needed, and you drift away from the Lean philosophy. Bets frame the work sharply and keep waste out from the start.

    Why cycles instead of sprints?

    The fixed cycle length prevents "waterfall in sprints": teams no longer work forever on things that may never deliver the intended value.

    What changes for the teams?

    Shape Up teams are smaller and dedicated to exactly one bet — instead of being too big, not truly cross-functional, and pulled across many projects at once.

    How does Shape Up prevent process theatre?

    Shape Up prescribes no mandatory events and creates no tickets. The focus is on projects and outcome — not on running a process correctly.

    What happens when a project does not deliver?

    It is clear from the outset when work will be stopped. There is no train without brakes — nothing gets extended simply because no one defined how to stop it.

    Does quality suffer when scope is cut?

    No. Scope is cut while quality stays constant. The tight boundaries are useful precisely for learning the counter-intuitive work of reducing scope.

    Where does Shape Up reach its limits?

    With a heavy stream of daily business, operations and service requests; when multiple teams must integrate outcomes that build on each other; and wherever hierarchies and reporting lines work against it. Like Lean, Shape Up has to come from top management — it does not work as a grassroots movement.

    Keep reading: