Two things happened in the same stretch of weeks.

Salesforce will auto-enable Agentforce in Winter ’27 for every org that already has access. The rollout started in the first week of September. Later in the release, the on/off toggle comes off Setup. Salesforce says there is no extra cost. Admins can still turn Einstein off, or deactivate an agent.

Then Salesforce and Anthropic announced Claudeforce. The first product is Salesforce in Claude: a plugin with 37 prebuilt sales skills, live CRM context, and governed actions from inside Claude. It is in pilot now, with open beta this month. Claude is also the default reasoning model inside Agentforce and Slack.

The names can be confusing, so to be clear:

• Agentforce is Salesforce’s agent platform.
• Claudeforce is the Anthropic partnership.
• Salesforce in Claude is the plugin.

Now we can talk about what actually changed.

• The stack is no longer the bottleneck.
• Turning agents on is a default.
• Talking to the CRM from Claude is simply a plugin.
• Skills come prebuilt.

Availability is not adoption.

A toggle, a plugin and 37 skills do not change how a brand sells, serves or decides. They change how easy it is to look busy. The hard part is still process, change and governance: which work an agent is allowed to finish, who works differently on Monday, and who is accountable when an agent sends, scores or spends.

Salesforce is not ignoring that. The product has permissions, Einstein controls, and “governed action.” The gap is organisational. If Einstein is already on, auto-enable puts Agentforce Builder in front of admins. That is not the same as a sales team using it, a service team trusting it, or a marketer letting it act on live campaigns.

We have watched this pattern on platforms for years. Paid for, switched on, still not earning their keep. The gap is rarely the platform. It is the layers around it. (We set that out here: 10 signs your digital platform is underdelivering.)

Claudeforce makes that sharper, not easier. If the CRM interface is no longer the only place work happens, data quality, permissions and the operating model matter more, not less. An agent that can update pipeline from Claude will do it with whatever records it can see. If nobody owns those records, the agent does not create a data problem. It industrialises one.

So what does the next 90 days look like if the agents are already on?

Days 1–30 – Discover

Find out what auto-enable actually switched on. Which Agentforce, Claude and Slack surfaces people are already using. Name the two or three workflows an agent is allowed to finish. Be explicit about who owns data quality, permissions and the commercial outcome.

Days 31–60 – Define

Write the policy: what can act without a human, what stays in draft until someone publishes. Choose a tool architecture. Agentforce, Salesforce in Claude and Slack are not all required on day one. Write the risk appetite down so the board can stand behind it.

Days 61–90 – Design and enable

Take one priority workflow and map it end to end: data, permissions, checkpoints, human approvals. Train the people who will work differently on Monday against that workflow, not a product demo. Measure completed governed work, not “agents enabled.”

That is not a software project. It is the same digital transformation work brands have needed around CRM, content and experience for decades. The agents just made it urgent.

If the agents are on and the organisation is not, maybe this is the 90 days you need: AI Foundations.

The post Salesforce just turned the agents on. The operating model is not. appeared first on Remarkable.

A digital platform can be stable, supported and doing exactly what it was built to do — and still return far less value than the business expected.

That gap is rarely obvious. It does not show up as an outage or a failed project. It shows up in how long a campaign takes to launch, how many people have to touch a piece of content before it goes live, and how much specialist capacity goes on keeping the current environment running rather than improving it.

In our Digital Experience Platform work across Sitecore, Optimizely, Contentful, Vercel and custom estates, the same ten signals come up again and again. None of them is fatal on its own. Any one of them is enough to mean you are paying for capability you are not getting.

The post Is your digital platform delivering what you paid for? appeared first on Remarkable.

Not long ago I watched a customer of an industrial technology business try to find out whether a piece of equipment he had bought four years earlier was still supported. He knew the product name. He knew roughly which part of the business made it. He spent about ten minutes on the manufacturer’s website, gave up, and phoned someone he had known for years. That contact did not know the answer either, but he knew who to ask. 

Everyone in that chain was competent. The company still got the outcome it wanted. And I would guess nothing about that interaction appeared in any reporting anywhere. 

For most of the time I have worked with clients in this sector, that was survivable. Industrial buying ran on relationships, technical trust and long memories. The website was a brochure. If the brochure was confusing, someone picked up the phone, and the phone worked, so no harm done. 

That is the part that has changed. 

The first place buyers look is no longer your website

Gartner surveyed 646 B2B buyers in late 2025 and found that 67% now prefer a rep-free experience, and 45% used AI during a recent purchase. In a follow-up published in May 2026, 70% said they would prefer a completely digital, self-service buying process. Forrester’s 2026 State of Business Buying puts it more bluntly: generative AI searches are now the starting point for B2B buyers.

Sit with that for a moment, because it is a bigger change than it first appears.

It does not just mean buyers are researching before they contact you – they have been doing that for many years. It means the first thing that reads your content is not a person. It is a model, summarising your category for someone who may never see the page it came from.

Google searches in the US that ended without a click reached 68% in early 2026, up from around 60% in 2024, according to SparkToro’s analysis of Similarweb clickstream data. That figure covers all searches rather than industrial B2B specifically (I’ve not found a credible equivalent measured on industrial buyers alone) which is worth saying plainly. But the direction is not really in dispute.

So, the question stops being whether a buyer can find your page and starts being whether a machine can understand your business well enough to describe it accurately. If your software documentation sits three levels under a hardware taxonomy, or if two divisions describe the same capability differently, or if half your technical content is in undated PDFs, then the model has exactly the problem your buyer has. And it resolves it the same way a buyer does, by going with whoever explained it most clearly. Sometimes that’s a competitor. Sometimes it’s a distributor. Sometimes it is a forum post from 2019.

Findability used to cost you a slow sale. It now costs you a place on the list.

There is a useful counterweight in the same Gartner data. 69% of buyers say they go back to a sales rep to validate what AI has told them, and slightly more buyers think they will be misled by generative AI (51%) than by a salesperson (49%). The relationship still matters. It just no longer gets first look, and it increasingly gets handed a set of assumptions formed somewhere you have no visibility of.

This is not really a website problem

Here is where I think most of these programmes falter, including some I have been part of.

The instinct is to treat this purely as a digital estate issue. Better search, better navigation, a redesign, a content audit. All of which help, and none of which hold.

McKinsey’s 2026 B2B Pulse, based on responses from nearly 4,000 decision-makers across 13 countries, lists the top reasons buyers switch supplier. They are

• Inconsistent information across teams
• Not being able to reach someone who actually knows the answer
• Friction across omnichannel experience

Read those again. Not one of them is a website feature. All three are what an internal operating model feels like from the outside.

Buyers experience your org chart whether you intend them to or not, that’s the reality. Divisions that grew through acquisition and kept their own sites. Hardware and software owned by different P&Ls with different content standards. Service data in one system, order data in another, neither of them talking to the thing the customer logs into. The buyer does not see any of that structure. They just see a company that can’t give them a straight answer.

I used to put the fix solely in the digital team’s plan. That wasn’t strictly accurate, and it is the single most useful thing I have learned doing this work. Digital teams can rebuild the front end, but they cannot make two divisions agree on what a product is called.

The patterns that show up almost everywhere

You can’t tell where one part of the business ends and another begins. Brand architecture in industrial technology usually maps to acquisition history rather than to how customers buy. That is understandable, and can be invisible internally. Externally it means a buyer cannot work out who they are dealing with, which subsidiary holds the contract, or which of three similar-looking sites is the current one. What they do next is predictable. They default to a person, or a partner – or nothing at all.

Nothing is visible after the money has been spent. Order status, lead times, repair progress, warranty position. Sana Commerce’s 2025 survey of 750 B2B buyers found 73% prefer to buy online but 85% experience online-ordering frustrations linked to outdated systems and inaccurate data, and 75% would consider switching supplier as a result. Worth noting that Sana sells B2B commerce software, so read the framing accordingly, but the pattern matches what I see. Meanwhile McKinsey found 73% of buyers are now comfortable placing orders above $50,000 online, up from 59% in 2022. People will spend serious money through a screen. But they will not spend it blind.

The comparison buyers make is not with your competitors. It is with everything else in their life. An engineer who can watch a parcel move across a map cannot see where a repair worth thousands has got to – which is a reasonable expectation.

Content is organised around your products rather than the buyer’s problem. Buyers do not usually search by product family. They search by the thing that has gone wrong, or the outcome they need, or the standard they have to meet. Most industrial content estates are built the other way round, then patched with a search box that cannot bridge the gap. This is the one that has started to matter most now, because the same structure that defeats a buyer defeats a model.

I will note one thing without dwelling on it. Some of this stays unfixed not because it is technically hard but because fixing it forces a conversation about who owns the customer relationship, and that conversation involves partners and channel economics. That is a real constraint, and pretending otherwise is not useful.

What it actually costs

Forrester puts the typical B2B purchase at 13 internal stakeholders and nine external influencers. Every one of those people can now look you up independently, in seconds, and form a view before anyone from your business is in the room. The account relationship you have spent a decade building covers perhaps two of them.

The wider picture is not encouraging. The 2026 Industry 4.0 Barometer, a study of 1,206 respondents from industrial companies run by MHP with LMU Munich, scores UK industrial digitalisation maturity at 62%, down two points year on year, with the DACH region flat at 57% against China at 72%. That index measures operational digitalisation rather than customer experience, so it is not a direct read on any of the above. But the constraints it identifies – legacy systems and fragmented data – are the same ones that stop the customer-facing work.

None of this shows up as a crisis. It shows up as longer sales cycles, more quotes that go quiet, more revenue that depends on individuals rather than on the business, and a slow drift towards whoever got the basics right.

Three things you could test this month

If any of this sounds familiar, there are three checks that cost almost nothing.

• Ask someone outside your business to find a specific, real answer on your own digital estate. Support status for a four-year-old product works well. Time them, and watch where they go when they give up.
• Ask a mainstream AI assistant a question a buyer in your category would genuinely ask. Read what it says about you, and look at where it got it from.
• Then look at the last five deals you lost and check honestly whether the real reason sits in the McKinsey three: inconsistent information across teams, no way to reach someone who knows the answer, or friction across the omnichannel experience.

If you sell into industrial or manufacturing buyers, I would be interested to know which of those three tests you would least like to run.

We are working through the data behind this properly — benchmarking how industrial technology companies actually perform against the three tests above. The report lands soon. Watch this space.

The post Industrial technology’s digital reckoning: a practitioner’s view appeared first on Remarkable.

We’re proud to share that Remarkable is now one of the first agencies globally to become Vercel Agent Foundations Certified, with eight of our developers achieving the certification. It means our team is fully equipped to build and deploy production-ready AI agents across Vercel’s entire agentic infrastructure: the AI SDK, Workflow SDK, AI Gateway, Sandbox and Fluid Compute.

The certification followed a partner workshop earlier this summer, where our developers were set a single brief: build a working AI agent using the Vercel AI SDK. Eight hours on the day, with Vercel engineers on hand throughout, on top of eight hours of pre-work per developer working through Vercel Academy’s Builders Guide to the AI SDK. It was hands-on by design, and a test of how our team can put their skills into practice, not just a test of memory.

Why Vercel is betting everything on agents

Vercel has spent the last year rebuilding itself around a single idea: that agents, not humans, are becoming the primary way software gets built and shipped. At its Ship 2026 event in London, Vercel said that six months earlier, fewer than 3% of deployments to its infrastructure were triggered by coding agents. Now, it’s more than half. Token volume through its AI Gateway has grown roughly tenfold in the same window, from around 2 trillion to 20 trillion a month.

That’s the backdrop our certification sits against. Vercel’s pitch, in founder Guillermo Rauch’s words, is that “each new generation of software needs a new generation of infrastructure. For the agent era, that’s Vercel”, and the five tools covered in our certification are the building blocks of that infrastructure:

1. AI SDK – one consistent way to build against any AI model, switching providers with a single line of code.
2. Workflow SDK – makes long-running, multi-step agents durable, so they survive crashes, deployments and hours-long pauses without custom queues or retry logic.
3. AI Gateway – a single API key that reaches hundreds of models across providers, with automatic failover and one unified bill.
4. Sandbox – isolated, disposable environments where an agent’s code can run safely, without ever touching production.
5. Fluid Compute – a compute model built for the bursty, wait-heavy nature of AI workloads, so you’re not paying for idle time.

What it means for our clients

Becoming certified isn’t the point — being able to build with real competence is. As agents take on more of how the web is designed, built and run, our clients need partners who have not just read the documentation, but actually done the work. That’s what this certification demonstrates: our team built a production-ready agent under real time pressure, with Vercel’s own engineers watching.

That capability is what powers AI Drive — our service for building, deploying and running AI agents in production — and the wider AI consultancy journey that takes clients from strategy to live agents.

If you’re thinking about what agentic infrastructure means for your digital experience, we’d love to talk.

The post Remarkable is one of the first Vercel Agent Foundations Certified Agencies in the world appeared first on Remarkable.

On 1 June, Salesforce announced a definitive agreement to acquire Contentful — the Berlin-founded, API-first headless content platform used by more than 4,800 enterprises, including close to a third of the Fortune 500. The deal is still subject to regulatory approval and isn’t expected to close until Q3 of Salesforce’s FY2027, so nothing changes for customers today. But the strategic signal it sends is worth paying attention to right now, especially if content and customer data sit in different corners of your technology estate.

What Salesforce is actually admitting

For years, Salesforce has positioned itself as the system of record for the customer — the single source of truth for who your customers are, what they’ve bought, and how they’ve engaged. Data 360 handles the unification. Agentforce handles the reasoning and action. What’s been missing is the layer that decides what a customer actually sees: the words, images, and assembled experiences that get delivered once an AI agent has figured out what to say.

That’s the gap Contentful fills. Salesforce has been explicit that the acquisition is about giving Agentforce “a native, headless, composable content layer” so it can assemble and deliver personalised experiences across every channel, rather than handing off to a separate CMS that doesn’t understand customer context. In plain terms: Salesforce has decided that owning the customer record isn’t enough if it doesn’t also own — or at least tightly integrate — the content that gets served against that record.

This isn’t Salesforce’s first move in this direction. Contentful joins a run of acquisitions — Informatica for data integration, Momentum for conversation intelligence, Qualified for AI-driven sales engagement, and Cimulate for experience simulation — that together sketch a picture of a company assembling a full AI agent stack rather than building each capability in-house. Content was the visible gap. Now it isn’t.

Why ‘missing piece of the puzzle’ is the right framing

Composable, API-first content management has spent the last several years as a build-your-own-stack decision: pick your CMS, pick your CRM, integrate them yourself. Salesforce buying rather than building here is itself telling. It suggests that stitching content and customer data together well enough to power dynamic, AI-assembled experiences is hard enough that even a company with Salesforce’s resources chose to acquire proven, in-market capability instead of building it from scratch.

For any organisation that has treated its content platform and its CRM as separate procurement decisions, owned by separate teams, on separate renewal cycles — this is the moment that assumption gets tested.

What this means for digital leaders

If you’re a CIO or Chief Digital Officer, this acquisition raises a governance and architecture question before it raises anything else: does your current content and customer data architecture assume these systems stay separate, or does it already anticipate them converging? Vendor consolidation at this level changes negotiating leverage, changes integration roadmaps, and — as several analysts have already flagged — raises data sovereignty questions for organisations operating under European or UK regulatory regimes, since a US acquirer brings different jurisdictional exposure than a Berlin-founded independent vendor did. None of that is a reason to panic. It is a reason to get ahead of the conversation with your architecture and compliance teams before a vendor’s roadmap decisions get made for you.

If you’re leading a digital transformation programme, the more immediate question is practical: what does this do to the frameworks you’re building your business case around? “Composable” has been sold as a hedge against vendor lock-in — best-of-breed content management, decoupled from the CRM, integrated on your terms. A major CRM vendor absorbing one of the category’s most credible independent players doesn’t kill that logic, but it does mean the independent-versus-suite decision needs revisiting with fresh eyes. It’s also a useful case study to bring upward: boards understand acquisitions, and this one is a clean illustration of why “content” and “customer experience” are converging into a single strategic line item rather than two separate budget conversations.

The questions worth asking now

Three things are worth putting on the table with your teams before the deal closes, not after:

• Where does content actually live in our stack today, and how dependent is that decision on assumptions about vendor independence that may no longer hold?
• What does “AI-assembled experience” require from our data, structurally, that our current content model doesn’t yet support?
• If our CMS vendor were acquired by our CRM vendor tomorrow, would that be a consolidation we’d want, or a lock-in we’d be trying to avoid?

The deal itself changes nothing operationally for another few months. The thinking it should prompt starts now.

Remarkable helps digital transformation and technology leaders make sense of moments like this — where a single vendor decision reshapes the assumptions an entire content and data strategy was built on. If this raises questions about your own architecture, we’re happy to talk it through.

The post Salesforce just bought the piece it was missing. Here’s why that should change your roadmap. appeared first on Remarkable.

Composable stopped being the decision some time ago. Gartner’s Magic Quadrant for Digital Experience Platforms, published in January 2025, carries a planning assumption worth reading twice: by 2026, at least 70% of organisations will be mandated to acquire composable DXP technology rather than monolithic suites, compared with 50% in 2023. Modular, API-first, independently deployable capability is no longer a differentiator in this category — it is increasingly what the category means.

So if you run a complex multi-site estate, the interesting question is no longer whether to go composable. It is who you assemble it with.

The word “partner” does double duty in that sentence — the platforms you licence, and the team who wires them together. In a composable world those two decisions have stopped being separable, which is why the seven factors below apply to both.

1. Composability is a given. The seams are the product.

Every serious vendor now claims API-first modularity, so the claim itself tells you nothing. What separates one composable stack from another is the orchestration layer: how content, commerce, data and identity actually meet, who owns that code, and what happens when one component gets replaced.

Ask a platform where its boundaries are and what it expects to be handed. Ask a partner to sketch the integration layer of an estate they run, from memory. If they cannot, they have never had to maintain one.

2. Multi-site is an architecture, not a feature

The gap between “supports multiple sites” and “supports forty sites across fourteen markets with shared components and genuine local autonomy” is where most programmes come unstuck. What you want is a content model that lets a market inherit, override and diverge without forking the build. What estates often end up with is site one, cloned thirteen times, and thirteen quietly diverging codebases within two years.

Ask to see a live estate at your scale and complexity. Not a demo tenant.

3. The front end is now its own decision

Composable decouples delivery from content, which promotes the presentation layer to a platform choice in its own right — hosting, edge behaviour, build performance, preview workflow, and how quickly a marketer sees their own change. That is a real gain in speed and control, and a real addition to your supplier map.

The partner test here is simple: does the front end come from the same team that owns the experience platform, or have you just acquired two suppliers who will point at each other across an API boundary?

4. Governance fails before the technology does

Gartner’s second planning assumption in that same Magic Quadrant is the one I would put on the wall: by 2027, 40% of organisations will fail to deliver impactful digital CX due to a lack of AI-driven intelligent content coordination and content operations strategy.

That is a content operations problem wearing a technology costume. Across a multi-site estate the failure is specific and recognisable — components used inconsistently, five versions of the same product description, brand drift by market, nobody quite accountable. Ask who owns the design system and brand governance the day after go-live, and what authority that person has.

5. Personalisation you cannot staff is personalisation you will not run

Every DXP demo personalises beautifully. Eighteen months later, a lot of estates are running two rules, both written by an agency that has since left. Composable can make this harder before it makes it better, because the data layer is now yours to assemble rather than the vendor’s to bundle.

Ask what the smallest useful piece of website personalisation looks like on the proposed stack, who writes the next one, and what evidence exists that the last one worked.

6. Agents are moving to the centre, and they read your estate differently

Forrester’s Wave for Digital Experience Platforms in Q4 2025 was blunt about the direction: “Agents are now the center of the DXP. Not a feature. Not an add-on. The center.”

There are two implications and they pull in different directions. Internally, agentic workflows change how content operations work — which is only useful if your content is structured well enough for an agent to act on. Externally, the first thing reading your content is increasingly a model summarising your category for a buyer who may never see the page it came from. A fragmented estate that contradicts itself across markets hands an answer engine exactly the problem it hands a human buyer, and it gets resolved the same way: whoever explained it most clearly wins. That is why AEO and GEO have moved out of specialist SEO territory and into architecture conversations.

7. Cost lives in the integrations, and so does the exit

Licence cost is the number in the business case. It is rarely the number that hurts. The recurring spend in a composable estate is maintaining the connective tissue — version upgrades landing on five different vendor timetables, renewals falling in five different quarters, and the specialist knowledge each seam quietly requires.

Ask a platform what breaks when you remove it. Ask a partner what they have decommissioned, not only what they have built.

What this actually means

Seven factors, one underlying test. Composable architecture moves risk out of the platform and into the assembly. A monolith could be judged largely on its own merits because the vendor owned nearly everything, including the failure. A composable estate is judged on the joins — and joins are somebody’s ongoing responsibility, named or otherwise.

Which changes the shape of a shortlisting exercise. Rather than scoring vendors against a feature matrix everyone passes, work out which of these seven you cannot get a straight answer to, and treat that list as your real risk register.

We work across Sitecore, Optimizely, Contentful and Vercel, usually on estates carrying more sites, markets and stakeholders than anyone originally planned for. The pattern is consistent: the platform choice matters slightly less than most selection processes assume, and the operating model matters considerably more than almost any of them measure.

The post 7 Things to Know About Composable DXP Partners appeared first on Remarkable.

For years, Jacksons Fencing had a quiet problem. It made some of the best fencing, gates and security barriers in the country (the kind specifiers trust and homeowners pay a premium for), and it led with the product itself, the craftsmanship, the security, the quality engineered into every panel, gate and post.

But that was only ever one side of the coin. The people actually buying the fence weren’t just weighing up the product; they were picturing the life it protected and the moments it kept quietly safe: the school playground behind it, the dog that couldn’t be trusted near the road, the garden they’d finally have privacy in.

The post A fence is never just a fence. appeared first on Remarkable.

Sitecore has just acquired Scrunch, an AI search optimisation platform, and the move is worth paying attention to – not because of the deal itself, but because of what it reveals about where digital experience is heading.

Before a prospect reaches your website, fills in a form, or speaks to a salesperson, they may already have an opinion of you — formed not from your carefully crafted homepage or your latest campaign, but from an AI-generated answer inside ChatGPT, Perplexity or Microsoft Copilot.

If your brand appeared in that answer, clearly and accurately, you’re already in the conversation. If a competitor appeared instead — or if you were misrepresented, or simply absent — the evaluation may already be over.

That’s the problem Sitecore is moving to solve. And it matters well beyond the Sitecore ecosystem.

The content estate problem

Most enterprise brands running a DXP like Sitecore have spent years building substantial content estates. Product pages, FAQs, thought leadership, case studies, regional variants, campaign content. The investment is real and the content is good.

The problem is that most of it was built for human visitors — rich, interactive, layered with JavaScript, personalisation engines and media components. AI agents don’t experience any of that. They retrieve, parse and compress. If your content isn’t structured clearly enough to be retrieved and understood, it won’t be cited — regardless of its quality.

Scrunch’s Agent Experience Platform addresses this directly. It creates a lightweight, machine-readable version of your site that AI agents can actually use, running in parallel with your existing experience rather than replacing it. Your human visitors get the full digital experience. AI agents get a clean, structured layer they can parse, trust and cite.

Sitecore has confirmed that Scrunch’s recommendations will be built into its AI Content Management, Content Marketing and DAM workflows — which means the insight and the action sit in the same place, rather than in a separate dashboard nobody has time to action.

Where AEO and GEO come in

At Remarkable, this connects to work we’re already doing with clients on Answer Engine Optimisation and Generative Experience Optimisation. The principles are consistent regardless of platform: content needs to be structured for retrieval, entities need to be clearly defined, answers need to be findable, and the content estate as a whole needs to be coherent enough for AI systems to trust it.

The questions worth asking are practical ones:

• When a buyer in your sector asks an AI to recommend a shortlist, do you appear?
• Which prompts do your competitors win that you don’t?
• Is your most valuable content structured in a way that makes it citable — or is it buried in a PDF, behind a form, or wrapped in a component an AI agent can’t parse?
• Where is your content estate fragmented, outdated or contradictory in ways that undermine your credibility with AI systems?

The Sitecore acquisition signals that these questions are moving from specialist SEO territory into the mainstream of digital experience strategy. For digital leaders already managing complex content estates, the audit of AI readiness is becoming as important as the audit of human experience quality.

What to do now

The opportunity isn’t to rebuild everything. It’s to understand where the gaps are — which content is working hard in AI search, which is invisible, and which is actively undermining your position — and to close those gaps systematically.

That’s a content and DXP conversation as much as it is an SEO one, and it’s one Remarkable is well-placed to help with.

Talk to us about AEO, GEO and AI search readiness

The post Sitecore acquires Scrunch – and your content estate is the point appeared first on Remarkable.

There’s a version of this conversation that the asset management industry has been having with itself for about a decade. It usually starts with something like “digital transformation is coming” and ends with a refreshed website and a new app that nobody uses. The consultants go home, the PowerPoint slides are filed, and the real business (mandate management, fee negotiation, relationship dinners) continues almost entirely as before. 

That era is ending. Not gradually. Fast. 

The challengers that have spent the last five years building digital infrastructure are now scaling it. The fintech disruptors that were once dismissed as “interesting but not really our market” are now sitting in front of the same clients. And the arrival of genuinely autonomous AI agents, not AI as a feature but as a real layer of investment intelligence, is about to change the way firms compete in this industry in ways that legacy players are structurally unprepared for. 

This isn’t a piece about technology adoption. It’s a piece about strategy. Specifically, about why so many traditional asset managers keep making the same category of mistake, and what it reveals about how they think.

The Real Problem Isn’t the Technology. It’s the Strategic Frame. 

Ask the digital or transformation lead at any established asset manager what they’re focused on and you’ll hear a familiar set of themes: operational efficiency, client portal experience, data and analytics, maybe something about AI in the back office. These are real programmes, with real budgets, often led by genuinely capable people. 

But they share a defining characteristic: they are defensive. They are designed to protect existing relationships and reduce cost-to-serve, not to fundamentally rethink what the firm offers or how it competes. They treat digital as infrastructure rather than as a strategy. 

Challengers don’t make this mistake because they can’t afford to. When you’re building from scratch, with no legacy custody arrangements, no inherited client reporting stack, no 30-year-old model that just needs to be left alone, you must build something that’s better. That necessity produces genuine innovation. At scale, it produces a structural advantage. 

The traditional asset manager’s digital budget is almost always a fraction of the technology spend already committed to keeping existing systems running. The challenger’s entire capital base is pointed at building the new thing. 

That asymmetry doesn’t resolve itself. It compounds.

What Challengers Are Actually Getting Right 

The temptation when writing about “fintech disruption” is to focus on the product layer: the slick interface, the lower fees, the frictionless onboarding. These things matter, but they’re downstream of something more fundamental. 

What the best challengers have built is a channel model that fits how modern clients behave. They don’t think about digital as a service delivery mechanism, the kind of portal you log into for information that used to arrive in a PDF quarterly report. They consider it the primary relationship. The data model, the communication logic, the investment narrative: all of it is designed around continuous, personalised, multi-channel engagement. 

That’s not a UX decision. It’s a strategic one. It requires a clear point of view on your client segments, a distribution model built to serve them at scale, and an operating model that can execute without grinding to a halt every time three people in the relationship team need to sign off on a piece of communication. 

Legacy firms tend to have excellent individual client relationships managed by excellent people who are entirely unreplicable at scale. Challengers tend to have scaled relationships that are less warm but far more efficient, and they are closing the quality gap quickly. 

The firms genuinely winning in the middle ground, the Schroders, the Nuveens, the Man Groups of the world, are the ones that have figured out how to scale without losing the human element. That’s a hard organisational problem, not a technology one.

AI Is Not Arriving. It’s Already Here. Just Not Evenly Distributed. 

The industry conversation about AI in asset management has, until recently, been dominated by two narratives that are both somewhat misleading. 

The first is the productivity narrative: AI will help analysts process more data faster, help compliance teams review documents more efficiently, and help client service teams draft communications more quickly. This is true and largely underway. It’s also table stakes. Every firm, legacy or challenger, is doing some version of this. 

The second is the risk narrative: AI in investment processes creates model risk, regulatory uncertainty, and fiduciary exposure. Also, true. Also, not the real story. 

The real story is AI agents operating as a client-facing intelligence layer, autonomous systems capable of monitoring portfolios, identifying opportunities or risks, generating contextualised commentary, and initiating communication without a human trigger. Several challenger platforms are already deploying early versions of this. The client experience isn’t “you can log in and ask our AI a question.” It’s “our system noticed something relevant to your portfolio this morning and has already surfaced it with context.” 

For a traditional asset manager, this creates a genuinely uncomfortable competitive position. Their value proposition, in its simplest form, is that they have smart people who make good decisions about your money. AI agents don’t replace that. But they do make it possible to deliver a version of that proposition at a fraction of the cost, at a speed humans can’t match, and at a level of personalisation that a human-led relationship model can only approximate. 

The firms that will lead in five years are those asking right now: what does our value proposition look like when AI agents are a commodity? What are we, specifically, still better at? And how do we build a digital model around the genuine answer to that question, rather than the answer that makes us feel most comfortable? 

Those are strategy questions, not technology questions. Very few traditional asset managers are treating them as such.

Channel Optimisation Is Where the Battle Is Actually Being Fought 

One of the persistent blind spots in legacy asset management is treating distribution as a separate function from client experience. The distribution team does its thing: managing intermediary relationships, running roadshows, handling institutional sales. The digital team does its thing. Occasionally, they coordinate. Often, they don’t. 

The challengers are laughing at this. Not unkindly, but genuinely. 

Because what the best digital firms have understood is that distribution and experience are the same thing now. Every touchpoint is a channel moment. The way your investment updates are structured is a distribution strategy. The way your client portal presents performance attribution is the distribution strategy. The way your thought leadership reaches a CIO at a pension fund before anyone from your firm has called them is a distribution strategy. 

SEO, content depth, structured data, AI-optimised copy: these aren’t marketing department concerns. They are how your firm’s ideas reach the right people at the right moment in the right context. A well-researched, authoritative piece of content that ranks for the right professional search terms will reach more relevant prospects than a hundred relationship dinners. That’s not speculation; it’s visible in the data for firms that have chosen to look. 

Channel optimisation in 2025 means understanding how your ideas travel: through search, through generative AI surfaces, through social distribution among professional communities, through recommendation. Legacy firms typically have one strong channel (the institutional relationship) and a scattering of underinvested ones around it. Challengers are systematically building the architecture to compete across all of them. 

The point isn’t that relationships don’t matter. They absolutely do, especially at the institutional end of the market. It’s that they increasingly exist within a competitive context where your digital presence either reinforces or undermines the relationship long before the meeting happens.

The Consulting Paradox 

There’s a final dynamic worth naming directly. 

Many of the large traditional asset managers have spent considerable sums on strategy consulting over the past decade. They have roadmaps, transformation programmes, centres of excellence, and innovation labs. They have done the work on paper. 

What they often haven’t done is change. 

This isn’t a criticism of the individuals involved. It’s a structural observation. Legacy asset managers are, in many cases, partnership cultures built around investment excellence. The culture is calibrated to protect the alpha generation engine, and rightly so. But it is also often calibrated to kill off transformation at the point where it starts to feel genuinely disruptive. 

A challenger doesn’t have that cultural antibody. It has survival pressure. Those are very different operating environments. 

The firms that have successfully navigated this (and there are some) have done it by treating digital strategy as a CEO-level priority, not a technology initiative. They’ve accepted that the discomfort is the point. They’ve built external partnerships not to outsource the thinking but to maintain external pressure on a culture that will, left to itself, default to “let’s do this incrementally.” 

Incrementalism is a perfectly rational strategy if you have time. The evidence increasingly suggests that traditional asset managers have less of it than they think. 

What Comes Next 

The next 18 months in asset management will be revealing. AI agent capabilities will move from experimental to structural. Challenger platforms with genuine digital maturity will start winning mandates that would previously have been unthinkable for them. Regulatory frameworks around AI in financial services will start to crystallise, reducing uncertainty for firms bold enough to have already built for it. 

The firms that enter this period with a genuine digital strategy (not a digital team, not a digital programme, but an actual view on how technology changes their competitive position and a plan to respond to it) will be very well placed. The firms still treating it as a back-office efficiency story will find themselves having a much harder conversation with their boards. 

Remarkable is currently conducting a comprehensive analysis of digital maturity across UK and international asset managers, closely examining what the leading firms are doing, where the real gaps are, and what good looks like in this context. It’ll be out shortly, and the findings are starker than the consensus would suggest. 

Watch this space. 

The post The Quiet Crisis in Asset Management: Why Legacy Firms Are Losing the Digital Race appeared first on Remarkable.

Content teams have always faced the same pressure: publish faster, stay on-brand, and don’t let quality slip. For many organisations, that tension results in copy that goes out the door slightly too long, slightly off-tone, or littered with inconsistencies that only a second pair of eyes would catch – except, nobody had time to look.

Sitecore’s AI features are designed to close that gap – not by replacing your writers, but by giving them a smarter editing environment.

What the AI actually does

Working inside the Sitecore Page Builder editor, editors can select any content component and trigger the AI optimisation panel from the right-hand sidebar. From there, a range of actions are available immediately: improve writing, fix spelling and grammar, make shorter, make longer, summarise, change tone, and translate. There’s also a free-text prompt field, so editors can give the AI a specific instruction (for example, “update Remarkable to Remarkable Group”) and have it applied across the entire component in seconds.

The AI generates an optimised version of the content while keeping the original intact. Editors can toggle between the two side by side, accept the change, or discard it, so nothing is pushed to the live site without a deliberate action.

For teams that want to go further, there’s an A/B test option built directly into the same workflow. Editors can set the AI-suggested version running against the original without leaving the page builder, feeding directly into Sitecore’s testing and personalisation engine.

The AI also extends to metadata. When editing page content settings, it can analyse the page and suggest relevant tags — in the video, it recommended both “Case Study” and “Events” based on the content — reducing the manual effort of correctly categorising pages at scale.

Why this matters for your team

The value here isn’t novelty; it’s time. For marketing and content teams managing dozens or hundreds of pages across complex digital estates, the overhead of reviewing, editing, and tagging content is substantial. When those tasks live inside the same tool editors are already working in, rather than requiring a copy to be pasted into a separate AI tool, revised, and pasted back, the friction disappears.

There are also brand governance implications worth taking seriously. Custom AI instructions mean you can encode your brand preferences — terminology, tone, naming conventions — and apply them consistently across the team, without relying on each editor to remember every rule.

For organisations already on SitecoreAI, these features are available now. For those evaluating their DXP options, they represent a meaningful shift in what “ease of use” looks like for content editors in 2026 and beyond.

Want to see this in action for your own platform?

Remarkable is a Sitecore partner with deep experience implementing and optimising Sitecore for enterprise clients. If you’re curious how AI-native content tools could work within your existing tech stack, we’d love to talk.

The post Your CMS Can Now Write Your Copy. Here’s What That Means for Your Website. appeared first on Remarkable.