AI can’t succeed in silos. But most businesses are built in them.

AI is no longer a future-state conversation.

It’s being used to detect revenue risk, optimise hiring pipelines, accelerate month-end close, and design personalised customer journeys. It’s drafting legal summaries, forecasting cash flow, spotting procurement anomalies, and automating onboarding. That’s happening now. It’s not a future state.

Across departments from sales to operations, finance to HR, customer service to marketing, AI is starting to shift how decisions are made, the cadence we make those decisions in and, fundamentally, how work gets done.

But as the use cases multiply, the thing that’s becoming most clear… adoption is fragmented.

Different teams are solving similar problems with different tools, built on different data, producing disconnected outputs, and rarely speaking the same operational language. Finance's truth looks different to marketing, sales look different to delivery.

This isn’t a capability problem. It’s a design problem. And if we don’t fix that bit, the impact of AI will remain capped by organisational structure.

Everyone’s building with AI. Few are building around it.

Walk into almost any mid-market or enterprise company right now and you’ll see it. Marketing automates campaign sequencing and performance reporting. Customer success deploys a renewal assistant. Finance experiments with intelligent spend controls. HR adds AI to screening and onboarding flows. Sales plugs in AI for opportunity prioritisation and coaching cues.

On the surface, it looks like progress. Underneath, it’s fragmented. 

Each department is selecting its own tooling, running its own experiments, generating its own logic, often with no shared definitions of value, limited cross-functional visibility, and no central accountability for outcomes (I spoke about this bit last week).

This siloed approach doesn’t just create inefficiencies, it undermines trust in the system itself. One team’s AI-generated insight contradicts another team’s version of reality. Adoption stalls. Metrics clash. The loop breaks.

According to McKinsey, while 90% of companies have launched AI or digital transformation initiatives, fewer than one in three have achieved meaningful financial impact. One of the biggest reasons? A lack of coordination across teams. McKinsey: Rewired to Outcompete

And that’s before the complexity of agentic AI even enters the frame.

As agentic AI matures, fragmentation becomes a business risk

We’re entering a stage where AI isn’t just delivering insights, it’s executing multi-step tasks, often based on the very insights it created.

This is the world of agentic AI: tools that don’t just respond to inputs, but take action autonomously or semi-autonomously across systems. Drafting contracts. Following up on tasks. Recommending pricing changes. Triggering workflows. Acting on behalf of human operators in real time.

This capability unlocks huge value, but it also introduces far greater complexity. If you don’t have alignment, you don’t just get inefficiency. You get exposure. The more autonomous your AI becomes, the more alignment matters.

And that’s precisely why frameworks become non-negotiable.

What AI needs now isn’t more pilots. It’s better architecture.

Too many organisations approach AI as a toolkit, something departments can dip into independently to drive local optimisation. That’s very much a mistake we made.

But the biggest wins don’t come from local optimisation. They come from connected workflows, consistent data, and shared ownership across functions.

McKinsey’s most recent research into AI at scale shows that the companies seeing outsized returns aren’t those with the most pilots. They’re those with the clearest structure for prioritisation, integration, and governance.

You can read more about that here.

These companies think less about “AI use cases” and more about “AI systems.” There’s horizontal design, not just vertical. Teams are grounded around shared metrics. There’s standardisation in data, and how it should be used. The company excels in defining not only what AI does, but also who reviews, refines, and owns it when things go wrong.

In a sentence, they stop experimenting in isolation and start building with intent. And that is only possible with a strong framework.

What a framework unlocks that tooling can’t

When we talk about an AI framework, we don’t mean a list of prompts or a backlog of automation ideas. We mean a structured, repeatable method for designing, implementing, and scaling AI across the business.

A real framework provides:

• Foundational assessment: Where is the business ready for AI? Where is it not? Technically, culturally, operationally?
• Cross-functional visibility: What processes cut across departments? Where does data pass between systems or teams? What are the dependencies?
• Value mapping: Not just time saved, but revenue impact, decision quality, risk reduction, and speed to outcome.
• Tooling guardrails: Not every team needs their own assistant. Not every use case needs a separate model.
• Change management: Training, support, and accountability don’t sit in a Slack thread; they’re built into the design.

It’s what turns AI from a series of experiments into a core capability of the business.

At Six & Flow, that framework is called FLAIR. It’s built to assess readiness, prioritise intelligently, and ensure AI is embedded not just in one department, but across marketing, sales, service, finance, HR, ops, and beyond.

It’s what powers the implementation of systems like SCALE+ and CS+ that form the backbone of our revenue operations, but it’s bigger than either. SCALE+ is how we apply AI within sales strategy and execution, from risk identification to sales coaching. CS+ embeds AI across customer success, expansion and revenue protection. But both rely on the structure FLAIR provides, because that’s what ensures adoption, consistency, and compound value. The framework ensured we got to business-wide impact fast.

Case in point: Klarna didn’t scale AI by accident

Klarna has become a headline example of enterprise AI done well (for the most part), but not because of a clever chatbot.

They redesigned their operating model to support AI. They trained teams. Built consistent architecture. Created connection between data, process, and outcome. Today, over 96% of employees use AI in their day-to-day workflows.

They didn’t start by choosing tools. They started by designing for scale. This is what they delivered:

• Revenue per employee up 152% in 24 months
• 40% reduction in customer service costs
• A more agile, more intelligent operational backbone

Read more about that here and here.

For me, that’s wild. In the “great COVID SaaS growth phase”, EVERYTHING growth-related was about headcount growth. Obviously, that had future repercussions with the tech layoffs, but my point is more that in less than 4 years, we’re now in a space where we have the ability to more than double the revenue impact per employee.

Were there failures along the way for Klarna? Of course. AI isn’t plug-and-play. But their structure meant they could recover, iterate, and adapt. That’s the benefit of framework-led AI: it’s not brittle. It evolves.

The future of AI isn’t prompt engineering. It’s operating model design.

This next phase of adoption isn’t about faster pilots or fancier assistants. It’s about building a business that knows how to work with AI, not just use it.

That means thinking beyond teams. Beyond tools. Beyond quarterly KPIs.

It means designing systems where AI and human capability co-operate, where workflows are shared, context is visible, and outputs drive action automatically or with minimal friction.

It’s where agentic AI thrives, because the business is built to support autonomous execution, not just individual experimentation.

But that future isn’t something you can brute-force through headcount or budget. It takes structure. It takes rhythm. It takes a framework.

Stop scaling tools. Start scaling systems.

The companies that win the AI race won’t be the ones with the most models. They’ll be the ones with the clearest frameworks. They’ll win because they treat AI not as a departmental lever, but as a company-wide operating shift. They’ll have embedded intelligence where decisions happen, not just where data sits. That’s not something a single tool can do, but a framework can.

And right now, that might be your most important AI investment.