Not long ago, Notion was the platonic ideal of a venture-backed rocket ship. A $10 billion valuation, a cult-like user base, and a mission to replace every other productivity app on your desktop. It was the apex predator of a category it helped define: the 'all-in-one' workspace. A single, elegant container for your company's documents, wikis, databases, and project plans. The promise was seductive: one subscription to rule them all.
That promise is now obsolete. The all-in-one category is a dead end. And the culprit wasn’t a slicker competitor or a market shift, but a technological phase change. Generative AI didn't just add a new feature to the Notion stack; it fundamentally invalidated the core premise that a single container for knowledge is where value resides. For UK founders building the next generation of software, understanding this post-mortem is not an academic exercise. It is the new playbook.
The Fallacy of the Perfect Container
The bull case for Notion, Coda, Slite, and their brethren was built on a simple, powerful idea: centralisation. By corralling an organisation's unstructured data—from meeting notes to marketing plans—into one system, you could create a 'single source of truth'. This would eliminate data silos, reduce the cognitive overhead of app-switching, and create powerful network effects. The more you put into Notion, the more valuable Notion became.
This logic drove astronomical growth. Industry estimates suggest that between 2019 and 2022, the UK market for 'integrated work management' software grew at a compound annual rate of over 30%, with enterprise procurement teams at firms from Revolut to the BBC signing increasingly large cheques. The per-seat licensing model was a cash machine. A mid-sized London tech firm with 250 employees could easily be spending upwards of £40,000 annually on its primary 'knowledge' platform, before even accounting for the litany of other specialised SaaS tools that stubbornly refused to be replaced.
This was the first crack in the façade. The 'all-in-one' was never truly all-in-one. Your customer data lived in Salesforce, your code in GitHub, your financial models in Excel, and your most critical, time-sensitive conversations in Slack or Teams. The 'single source of truth' was, in reality, just another (admittedly very flexible) silo.
Then came the wrecking ball: large language models (LLMs).
Initially, the market saw LLMs as a sustaining innovation. Notion, naturally, integrated 'Notion AI' to help you write, summarise, and brainstorm within its walls. It was a classic feature-add, designed to make the container more attractive. But this was a profound misreading of the technology's impact. The true power of generative AI isn't its ability to operate within a single, pristine garden; it's the ability to act as an intelligent fabric stretched across every application, repository, and data source an organisation uses.
The Contrarian Take: Your Moat Was Never Your UI
The conventional wisdom is that Notion's moat was its flexible, block-based editor and interconnected database functionality. This is incorrect. Its real, unrealised moat was the vast, structured, and interlinked proprietary data it was accumulating on behalf of its customers. The problem? It was trapped.
Generative AI’s core function is reasoning over data. The winning platform in this new era is not the one with the best container, but the one with the best access and reasoning capability. A system that can answer the question, "What are the key action items from last week's sales calls in Slack, cross-referenced with the Q3 product roadmap in Notion and the latest customer churn data from our internal dashboard?" is infinitely more valuable than a system that can only tell you what's in its own database.
By focusing on perfecting the container, the all-in-one category made itself a sitting duck for a new class of tool: the AI-native reasoning engine. These tools don't seek to replace your existing software stack; they seek to unify it at the intelligence layer. They plug into Slack, Google Drive, Notion, and your bespoke internal tools, using AI to provide answers and automate complex workflows across them. The value has shifted from the database to the query.
A recent, as-yet-unpublished survey of 200 UK CTOs, shared with Creative Agent, highlights this pivot. When asked about their 2024-2025 software procurement priorities, 'Consolidating on an all-in-one platform' was ranked near the bottom. The top priority? 'Investing in AI-native tools that integrate with our existing stack'. This is a tectonic shift in how UK businesses buy software, moving away from monolithic, multi-year seat-licence agreements and towards more flexible, consumption-based 'intelligence' services.
The New Founder Playbook for a Post-Notion World
For founders, the collapse of the all-in-one paradigm isn't a threat; it's the single greatest opportunity of the decade. The enterprise software market is being reset. But succeeding requires a new playbook.
Playbook 1: The Moat is the Model, Not the App
Stop building another slick SaaS interface. The new competitive advantage lies in proprietary AI models that deliver unique capabilities. This doesn't necessarily mean training a foundational model from scratch—an eye-wateringly expensive endeavour. It means fine-tuning open-source models (like Llama 3 or Mistral) on domain-specific data sets or creating novel architectures for multi-step reasoning (Agents).
Consider a UK-based FinTech firm. Instead of building another 'all-in-one' client portal, a smarter play is to build an AI agent that can ingest data from the FCA handbook, market data feeds, and a client's own financial documents to provide real-time, auditable compliance advice. The value isn't the chat window; it's the verifiable accuracy and domain-specificity of the underlying model. The moat is defensibly intelligent.
Playbook 2: Shift from SaaS to Intelligence-as-a-Service (IaaS)
The 'per-seat, per-month' model is a relic of the cloud's first wave. It charges for access, not outcome. As AI automates tasks previously done by human 'seats', this model breaks down. Why pay £20 per user for a tool when an AI agent can perform the work of 10 users for a fraction of the cost?
The new model is consumption-based: pricing per API call, per automated workflow, per insight generated. This aligns a founder's revenue directly with the value their AI delivers. It also fundamentally changes the procurement conversation. A CFO at a FTSE 250 company isn't buying 'software'; they're investing in a system that promises a quantifiable reduction in operational expenditure or a measurable increase in revenue. Founders who can't frame their offering in these terms will be routed.
At Creative Agent, we've seen a surge in enquiries from UK enterprises not looking to buy another off-the-shelf SaaS, but to build proprietary 'reasoning engines' that connect their existing data silos. They want to quantify the efficiency gains and build strategic assets, not just rent more digital real estate.
Playbook 3: Build for the Fabric, Not the Throne
The ambition to be the 'one tool to rule them all' is now a liability. The future of enterprise software is a composable, interconnected fabric of best-in-class applications unified by an AI intelligence layer. Founders should obsess over their APIs and integration capabilities. Your product should be designed from day one to have its data read and its actions triggered by an external AI agent.
This means prioritising things that were previously afterthoughts for many SaaS firms: robust APIs, comprehensive documentation, and event-driven webhooks. The most successful 'apps' of the next decade might not even have a user interface. They will be headless services that provide a unique capability—be it financial modelling, legal contract analysis, or supply chain optimisation—to the overarching intelligence layer that orchestrates work across the entire organisation.
The Next £10 Billion Opportunity
The game has changed. The race to build the prettiest, most integrated container for information is over. The AI wave didn't just add a feature; it broke the container and scattered its contents. Value is no longer in possession, but in access and interpretation.
This is disruptive in the truest sense of the word. It levels the playing field, creating openings for agile UK startups to outmanoeuvre lumbering incumbents who are trapped by their own business models and technical architecture. The next £10 billion software company won't look like Notion. It will be leaner, more powerful, and likely won't have an army of users logging in every day. It will be the invisible, intelligent engine humming beneath the surface of the enterprise, making sense of the chaos that the all-in-one platforms failed to tame. For founders, the message is clear: stop building better walls; start building smarter bridges.
