The four values of the Agile Manifesto

The Manifesto begins by saying: "We are uncovering better ways of developing software by doing it and helping others do it." It then lists what they have learned to value.

It was written with software in mind, but if you replace that word with marketing campaigns, training materials, financial analyses, or a new service, the text still works. The values do not depend on what is produced, but on how the people who produce it work.

"Individuals and interactions over processes and tools"

Now we have the most capable tool a team has ever had.

The analyses we bring to meetings are well written, with charts and clear conclusions. But when we ask about a figure that catches our attention, the answer is: "AI came up with it."

The tool can propose, draft, and execute, but if we give it responsibility for what it does, we are abdicating: we are handing it the baton.

"Working software over comprehensive documentation"

This is the most surprising value when rereading the Agile Manifesto. In 2001, exhaustive documentation was expensive: it cost weeks and had to be justified. Today it is free. Thirty pages come out in five minutes, producing reports that are flawless, well presented, and possibly even rigorous.

But this abundance has a cost. A report generated in five minutes needs an hour to determine whether it meets expectations, and if we skip that, we lose the most valuable thing: trust in the person who sent it.

"Customer collaboration over contract negotiation"

When building required time, we could see whether we were getting confused or drifting from the vision. When the result arrives immediately, that margin disappears. That is why, the faster development happens, the more important it is to know what needs to be built.

The temptation is to let the machine decide that too. And AI is good where reality repeats itself: it can predict how a fluid will behave, but customers do not work that way. What they value is discovered by talking with them, observing how they use what we give them, and making mistakes together. It is tacit knowledge, and it does not come from data processed in silicon, but from lived reality distilled through neurotransmitters.

AI reasons about what customers say and do: it processes that information and structures it better and faster than we can. But there is a third ingredient that only people provide: the understanding and intuition that come from being present in the real world.

"Responding to change over following a plan"

Responding to change is not changing quickly; it is realizing what needs to change. And to realize it, you have to stop and look.

With AI, a detailed plan takes minutes to produce and looks so solid that it is tiring to question it. But a plan that cost nothing to make should not cost anything to change either.

The useful question is not how much we can produce. It is how often we need to look at what we have done.

"While there is value in the items on the right, we value the items on the left more"

The sentence that closes the values reads differently today. Everything on the right (processes, tools, documentation, plans) is exactly what AI produces in abundance and almost for free. What is on the left (people talking to each other, results that work, customers we collaborate with, the capacity to realize things) remains scarce. The Manifesto asked us to value the left more; now, on top of that, it is the only thing that does not come standard.

The twelve principles

Complementing the four values, the Manifesto sets out twelve principles that are more specific. Reread today, AI has not left them behind; it has given them a relevance they did not have before.

1. "Our highest priority is to satisfy the customer through early and continuous delivery of valuable software"

With AI, early and continuous delivery come standard. The key lies (now more than ever) in the last two words: valuable software. The question is no longer how much we deliver, but whether it adds value.

2. "Welcome changing requirements, even late in development. Agile processes harness change for the customer's competitive advantage"

Redoing work no longer costs much, so accepting changes seems easier than ever. But there is a trap: when change is free, it happens without learning. The principle does not talk about changing a lot, but about harnessing change. And to do that, you have to understand why a requirement changes, something the tool does not figure out for us.

3. "Deliver working software frequently, from a couple of weeks to a couple of months, with a preference to the shorter timescale"

What used to set the frequency was production. Now it is review.

There is an easy sign to recognize: if the "in review" column on the board grows every day, AI is producing faster than people can check. Reviewing is also work, and if it is not planned, it becomes a bottleneck. The agile answer is the same as always: small batches, even when the tool allows producing in large ones, and a pace the team can review with care.

4. "Business people and developers must work together daily throughout the project"

This principle was necessary because neither side could move forward without the other. With AI, they can. The business side builds a prototype or a small tool in an afternoon without going through the team, and the team builds without consulting the business because it no longer needs to wait for anyone to move forward.

AI makes it possible to work separately exactly where the Manifesto asked us to work together. And what gets lost is not coordination, it is understanding: a business side without the team does not see what it costs to maintain what it asks for, and a team without the business does not know what it is building it for.

5. "Build projects around motivated individuals. Give them the environment and support they need, and trust them to get the job done"

If AI takes over the part of the work that gives craft its satisfaction and people are left with reviewing, motivation does not sustain itself: it has to be nurtured. And trusting the team is not the same as trusting what its tool produces. Trust is placed in people's judgment, since they are the ones accountable for the result.

6. "The most efficient and effective method of conveying information to and within a development team is face-to-face conversation"

The principle does not say that conversation is the fastest method. It says it is the most efficient.

With AI it is easy to sacrifice the value of the conversation that enriches (used to enrich) the team. Replacing the daily meeting with an automatic summary may be faster, but it is a way of losing the talk that spots problems, lets ideas take root, and feeds the collective intelligence of the human team.

7. "Working software is the primary measure of progress"

Dashboards are starting to appear showing the team's percentage of AI use. It is a convenient metric, but it measures adoption, not results. A team using AI a lot says nothing about whether what it delivers is useful. Progress is still measured by getting results that work, not by volume.

8. "Agile processes promote sustainable development. The sponsors, developers, and users should be able to maintain a constant pace indefinitely"

We hear a lot about what AI saves and very little about what it wears down. Reviewing all day what another produces, even when that other is a machine, is exhausting: it demands constant attention and leaves little craft satisfaction. A sustainable pace is no longer just not extending the working day; it also means spreading the work so that nobody becomes a full-time reviewer.

9. "Continuous attention to technical excellence and good design enhances agility"

A demo built in an afternoon works in the meeting, convinces everyone and, without anyone deciding it, ends up in production. A prototype that works is not a finished product.

That something looks finished does not mean it is. With AI, the bar for "done" goes up, not down: what AI produces must be reviewed by the person accountable, the one who knows what is meant to be achieved. Technical excellence is precisely the gap between what seems to work and what works when people actually use it.

10. "Simplicity--the art of maximizing the amount of work not done--is essential"

It was written when development cost time and effort. Today it is probably the most demanding sentence in the text. When the tool makes it possible to do everything, the art is no longer in doing, but in deciding what not to do.

11. "The best architectures, requirements, and designs emerge from self-organizing teams"

"Ask Marta, she's the one who knows about AI." When everyone uses AI in their own way and knowledge concentrates in one person, the team does not organize itself: it depends. Self-organizing today includes agreeing together on how AI is used and how what each person learns is shared.

12. "At regular intervals, the team reflects on how to become more effective, then tunes and adjusts its behavior accordingly"

With AI it is easy to think there is no longer time or reason for retrospectives, but it is exactly the opposite: the faster things get produced, the more those moments must be protected. A team that produces three times as much and reflects half as much does not go faster: it accumulates errors faster than it sees them.

What the Manifesto takes for granted: people

Behind the four values is an idea the Manifesto did not need to write, because in 2001 it was obvious: the quality of the work depends on the judgment of those who do it. Now AI puts knowledge within our reach on a scale we never had before, but deciding what is valuable, understanding the context, and being accountable for the result remain our responsibility.

And that judgment cannot be delegated: it is cultivated by doing. Here is a question we cannot ignore. If artificial intelligence does the tasks through which beginners used to learn, who will know how to review what AI does five years from now? Looking after the team's learning is not secondary; it is what guarantees we will still have something to contribute tomorrow.

At this point there are two roles that are sometimes considered obsolete, but I believe they are more necessary than ever.

The product owner, because the cheaper it is to produce, the more valuable it is to decide what to build. A Product Owner who only writes and orders tasks is limited to the part AI does best. The one who holds the vision, who treats every idea as a bet to be validated, and who knows how to say no with strategy in hand, keeps teams from running faster and faster in the wrong directions.

The Scrum Master, because AI can prepare and summarize meetings, but it does not sense how the team is doing, nor does it build the trust needed for someone to dare say "I approved it without looking." And because new work appears: helping the team decide what is delegated to AI and who reviews it, protecting moments of reflection against the pressure of speed, and supporting people through a change that creates uncertainty for many.

AI provides the capacity, but it is values that decide whether it becomes something that matters or just noise.

If the team produces twice as much today as it did a year ago, is it delivering twice the value?