The question I am probably asked most often is about the love-hate relationship between CRM and CCaaS vendors.
Who remains and who gradually disappears? In my time as a supplier, I always brushed that off with integration and APIs: “you need both, an API here and an API there,” and other such generalities.

 

Contact center software vendors and CRM vendors have been circling the same piece of ground for well over two decades now: the customer record. Both camps have tried to convince the market that they ought to be the owner of the customer relationship. Millions spent on acquisitions, integrations, and marketing decks later, neither has won – if that was ever the goal at all.

It is the eat-or-be-eaten story of enterprise CX. Most reflections on this boil down to either vendor victory laps or vague warnings about 'the future of work'. But I believe something has changed in the meantime.

Two empires, one border, no winner

CRM vendors have built their empire on data: accounts, tickets, a neat 360-degree view of the customer. Contact center vendors built their platforms on real-time interactions: routing, queues, recordings, literally the voice of the customer.

Both parties have, like in a kind of trench warfare, continuously tried to annex pieces of the other's territory. CRM platforms have incorporated contact center and agent functionality into their systems. And contact center platforms have added applications for case management and back-office integrations. The pitch was always the same: one system for the complete management of the customer relationship.

Neither strategy has been entirely successful. Ask any agent, or better yet: any client organization. They are still juggling two or three screens, still copying a case number from one system to another, still putting a customer on hold to 'just grab the data' in another tab, and still claiming they cannot fully warm transfer a call. That single integrated customer view has been sold for twenty years and has, at best, been delivered in small pieces in recent years.

SaaS was supposed to solve this, but well: it didn't.

SaaS was presented as the great equalizer: lighter deployments, an API-first world in which CRM and CCaaS would finally talk to each other and even strengthen each other.

Some of that came true. Implementations proceeded faster. Pricing models became more flexible, and on paper, roadmaps looked more integrated.

But the core problem was barely touched upon. Most enterprises still run multiple CRM solutions side by side, along with both a contact center platform and a stack of point solutions. Each with its own login and its own definition of 'the customer'. With SaaS, the trench warfare is simply packaged in a subscription model.

What changes with agentic AI

Traditional software wins the match by being the 'system of record'. Switching costs turn you into a loyal customer. But for agentic AI – the autonomous AI agent capable of performing actions – it makes no difference in which system the data is stored. An AI agent capable of looking up an order, checking a shipping status, and answering an invoice query doesn't care that the necessary information resides in the CRM, the contact center platform, or a spreadsheet someone once forgot to clean up. Providing access is sufficient.

As soon as an AI agent can complete an end-to-end task entirely using multiple systems side-by-side, the underlying applications matter less. The CRM system is no longer 'the place where the relationship lives'; it simply becomes one of a data source in a series. CRM and contact center software will then start to look like what they actually always were: databases and routing engines. That is an uncomfortable truth if your business model is based on the idea that you are irreplaceable. But it is good news for everyone who paid for five systems that could have been one.

Fair is fair: both camps saw this coming. Consequently, they have since quickly maneuvered themselves into the orchestration layer – and not just by slapping a chatbot widget on top, although that is part of it too. The question is whether that orchestration layer is a genuine architectural change, or a successful lick of paint on the same monolith. For AI agents can now access the data without entering the application's front door: it is still the same building. I call that agent-accessible. Whether it is also agent-native is a question you may ask any vendor, including those with the most beautiful keynote slides.

The good news: commodity solutions are easier to use

When a CRM application or contact center solution breaks down into separate services behind APIs, those services become easier rather than harder to purchase. A 'get order history' call, neatly exposed at the network level, can be picked up by any agent or team without anyone needing to master a new admin console. The functionality becomes a building block that you can plug in organization-wide, in the same way that electricity was no longer a brand when it came out of literally every socket. No one wonders anymore exactly which power generator lights which room. Although consumers in the Netherlands might actually do ask that, now that I think about it.

Ten years later: will CRM survive this trench war?

It is my opinion, and I am wrong more often than I am right, so here is my shot across the bow. If customer data is further extracted from vendor-specific data structures and ends up in a layer accessible to any AI or human agent, then 'CRM' is no longer something a company *must* buy. Not this year, and probably not in three years either.

Do not be surprised if, within a decade, 'the customer record' is a shared, movable layer instead of a licensed product with a login screen. And reasoning further: the customer record may well be the property of the end customer or consumer themselves, who chooses, on an ad hoc or structural basis, to be a participant in that layer.

When it comes to CCaaS, my final verdict is different. The decision logic, skills-based routing, IVR tree structures, and queue rules are just as exposed as the object model of CRM. You can house those rules in different systems. But things like the telephony infrastructure, number portability, rules regarding wiretapping and emergency calls (such as E911) in every jurisdiction, 'five nines' availability when it comes to volumes, and more: that is 'plumbing,' and you cannot automate that away.

The irony is glaringly obvious. For a decade, both CRM and CCaaS vendors have marketed telephony as legacy, as something to bury under a friendlier UI. With the advent of agentic AI, that is being reversed. It is the user interface layer that is being automated. The very plumbing work that no one wanted to talk about in a keynote is precisely what an AI agent cannot figure out on its own. Telephony is therefore not legacy, but the moat around the fortress, the moment everyone stopped bragging about it.

Software disappearing?

My prediction is therefore not that 'software will disappear'. As far as I am concerned, the situation is more nuanced: that part of the software stack that always contained all the logic will be eaten up first. This is also the domain of CRM applications, which were largely built on logic and object structure; for these vendors, it was therefore logical to aggressively enter the CCaaS domain. The value of CCaaS lies only partially in exposed logic and partly in actual network infrastructure. It is precisely the latter that is more relevant than ever with the growing role of agentic AI. As an enterprise, ensure you know which of the two you are actually buying.

CRM and CCaaS have fought for the same piece of ground for twenty years without either managing to fully conquer it. SaaS led to new packaging, but not new content. Agentic AI is the first real stress test for that old paradigm. Agentic AI may also be the phenomenon that finally settles this issue, in which neither vendor is crowned the winner.

Who has the best chances? My bet: the suppliers who survive the coming decade are those willing to let their own application layer, in which agents are currently trapped, become boring and replaceable. Now, on their own terms, instead of waiting for the market to make that choice for them.

(Ziptone/Merijn te Booij)

Featured, Opinions, Technology