How to lose your head well
When your B2B SaaS product no longer needs an interface

If you have not heard of the headless shift, and your business is SaaS, you may want to start thinking about how to lose your head… well.
There are still some product leaders who describe their interface as their moat. They see it as the thing that keeps customers loyal, makes switching difficult, and justifies the fees. Now, though, enlightened product leaders are asking different questions: “What happens when AI agents are our users?”, and “Do we even need an interface?”
The headless shift is more of a conceptual reframing than a technical challenge, but it’s an important development in the story about what B2B software was actually selling for the last twenty years, and what it now needs to do instead.
Where did ‘headless’ come from?
The term “headless” originated in content management, not enterprise software, but it’s worth understanding the original idea. For most of the time we’ve been online, the place you wrote your content was the same place you published it. WordPress is a good example: you wrote your blog in WordPress, it looked like WordPress; the content and its head (that’s the interface to you and me) were inseparable.
Then, in around 2013, Contentful began suggesting this was the wrong architecture. What if content was just structured data, stored cleanly, delivered via API, and rendered by whatever front-end made sense? A website, a mobile app, a digital display, a voice assistant: the content layer need have no opinion about how it was presented. The head was detached. Sanity, Storyblok, Prismic, Contentstack followed, each bringing their own flavours. A category emerged around the concept that separating content from its presentation layer unlocks new opportunities.
Headless, as a concept, is gaining ground from outside of Content Management - as identified by Andreesen Horowitz and as demonstrated by a recent headless Salesforce launch. Plus, of course, if you understand anything about Agentic AI, you’ll have already pushed ‘human interface’ down your list of priorities and begun moving towards agent-native.
So how is this relevant to B2B SaaS?
B2B SaaS tends, in some quarters, not to move so fast. Maybe it’s the “why fix what ain’t broken?” question that holds Systems of Record where they are? For decades, the interface has been the mechanism through which knowledge was built and maintained. And for decades, the ‘HCI’ or Human<>Computer Interface, was all-important.
Across all types of enterprise software, the UI did something important that nobody fully articulated at the time: it made humans responsible for the data. When a salesperson logged a call in Salesforce, they were making a judgment about value, adding context to a record that ultimately informed forecasting, pipeline management, and strategy. The dashboard was what people bought.
The dashboard is the head.
The new user
I see many SaaS businesses grappling with the AI opportunity; many are still leveraging AI ‘at the edges’ and have not yet understood the fundamental shift of a headless architecture. That’s because they are busy worrying about how AI can improve their existing product; how AI can improve the experience for their existing users doing existing things, in their existing jobs.
They run AI initiatives and AI pilots to flow conversational AI across their interfaces, they use AI to personalise and streamline workflows and better empower existing users to do the jobs they do today a little better.
But what if this strategy was based on an entirely baseless assumption: the assumption that the SaaS user tomorrow will look like a SaaS user of yesterday; the assumption that your user is in any way, human?
Your new user, in, say, 12-24 months, is not so likely to be a human. Sure, there may be some humans using a dashboard, but there will be other users building, adding value to, and consuming the data you are creating. Your new user is, of course, an agent. Or rather, multiple agents. Oh, and instead of the predictable humans whose workflows you could model, these new agent users will be doing things you never dreamt of, and can never predict… with your data.
Agents at work
AI agents don’t need a dashboard to write new data to a CRM. They don’t need an approval workflow to update a financial record. They read and write through the API, faster and more consistently than humans.
In this future, while you may have some form of user interface for certain aspects of your product, much of the value in your product will be created and consumed via agents. While swathes of your interface are rendered unnecessary at best. At worst, the very presence of UI for data entry and retrieval will smack of ‘old-skool thinking’ and mark you out as pre-AI-era.
Maybe you were concerned with building in stickiness for the humans, making your product so invaluable that it was used multiple times a day, and was core to ‘the work’. You measured daily actives, session frequency and duration, actions per session, and action completion rate. All sensible things to be doing, of course, in the SaaS world, no good Product person would do otherwise.
Until that is, you realise the future is headless. In the headless world, you recognise that your user is no longer a human. Or maybe it is a human, but it is very much also an agent. In the headless world, you serve your agent users with different things from your human users. An agent doesn’t need to find your onboarding flow intuitive or to appreciate your clean design. It doesn’t need to build habits. The only thing it cares about is whether it can reliably read and write the data it needs, in a structure it can understand.
For agents, the things that matter are schema clarity, API reliability, data consistency, predictable versioning, clean permissioning, and so on.
Who wins in a headless world?
Most B2B SaaS companies have been competing on dashboards, UX and feature sets, not data architecture, but the ones that will do well are the ones that’ve invested in data infrastructure and building a future-proofed data foundation. For some, it will be because they anticipated agents. For others, it will be because they always understood that a good data infrastructure compounds value.
I don’t think you need to be AI-native to win in this space, but you need to have the ability to shift to data-first, fast. And there’s plenty that survives the headless shift. Some of the core features of systems of record become more defensible, not less, as you become headless. Workflow rules, approval chains, permissions, and business logic that took years to build, years of clean, consistent, well-structured records, and clean audit trails are all vital. If you’re making the shift, you don’t need to start from scratch; in fact, much of what you already have, if built well, can help you see off AI-native competitors if you move quickly enough.
If you are not AI-native or agent-native, here is where to start
Most businesses didn’t start with agents in mind. The challenge they’re facing is how to make the transition without tearing up what seemingly works. It’s worked so far, right, so it’s hard to even imagine it might soon stop working.
Even if you have plenty of human users paying you good MRR to use your interface, start here:
- Data Audit - Start with a data (not a product) audit. This informs your agent strategy.
- API as Product - Treat your API as a product. Most B2B SaaS companies, of course, already have APIs, but few have invested in them the way they would invest in a user-facing product.
- Define the logic and build for orchestration - Rules, workflow, the stuff that was in the UI can’t be inferred by agents, you need to be explicit in the schema and then build for orchestration (deciding which agents run, in what order, with what data, and what happens when something fails).
- Redefine the human role - Give humans something new and better to do when the agents can do the ‘easy’ work.
- Sunset old ways of working - Don’t just maintain the old interface because nobody has decided to change it.
- Act fast - Don’t wait for data quality to be perfect before experimenting.
- Monitor - Create infrastructure that tells you what agents are doing, and lets you intervene when something goes wrong.
Embrace the headless shift
The companies that make this leap well will not necessarily be the ones with the most sophisticated AI strategy, but the ones that understand what their system of record actually is: a store of valuable, structured, trusted, operational knowledge (which up until recently just so happens to have been accessed via a human-centric UI).
Losing your head is inevitable; the question is whether you do it deliberately or by accident.
For help making the headless Shift, Rebel Futures is here to help. If you want help with a data audit, or a landscape analysis of who’s doing what in your vertical, or you want help with the cultural shift to AI-first, we can provide bespoke consulting services to help you create new value with AI.
Katie Lips is CEO and Founder of Rebel Futures. She helps organisations grasp the opportunities at technology inflection points, from the early web and pre-iPhone mobile to cloud-based SaaS and now AI. She has held senior roles at EY and Aimia, and was Head of Digital Policy at Which?. Katie holds a Harvard Kennedy School qualification in Leading in Artificial Intelligence and is a Founder Institute graduate.
Related: explore our work on AI Strategy.