Salesforce turned twenty-five in 2024 and remains the most widely used CRM in large companies. HubSpot has spent more than fifteen years selling the same promise: a CRM that works for any business. And yet, if you ask the sales teams who use them every day how they’re getting on, the most common answer is not an enthusiastic one.
The reason is not that Salesforce or HubSpot are bad products. The reason is that no software designed for any type of company can be optimal for a particular company. And over the last three years, the availability of open-source language models and rapid development tools has fundamentally changed the equation.
For the first time in the history of enterprise software, building a bespoke CRM for the specific sales process of a mid-sized company comes at a reasonable cost. And the difference in effectiveness is enormous.
Why generic CRM has an adoption problem
The number one problem with CRM systems in businesses is not technical. It is one of adoption. Sales teams do not fill them in, do not update them, do not check them before calling a client. The CRM ends up as a repository of outdated data that the sales director reviews with frustration every Monday.
This happens for a simple reason: generic CRMs force the sales team to adapt their sales process to the software’s structure, rather than the other way round. The pipeline stages are the ones the software defines. The fields to fill in are the ones the product designer considered relevant for a hypothetical universal customer. The alerts and reminders work on the default business logic.
For a company that sells industrial machinery with six-month sales cycles and three decision-making levels on the client side, that generic CRM is as useful as a map of another city.
What changes with sector-specific customisation
A CRM built for a sector’s specific sales process has its own vocabulary, pipeline stages that reflect exactly how the customer buys in that sector, and an alert logic that responds to the behavioural patterns that genuinely indicate whether an opportunity is progressing or going cold.
For an estate agency, the pipeline is not the generic "contact → qualified → proposal → close". It is "visit requested → visit completed → interest confirmed → offer presented → counter-offer → reservation → completion". With specific fields for property type, maximum budget, preferred area, decision timeline and the client’s financing status.
For a mechanic’s workshop, CRM is not a sales pipeline. It is a vehicle and owner history where what matters is when the last visit was, what work was carried out, when the customer’s car is due for its MOT, and whether there are any pending jobs that were quoted but not approved.
For a professional services agency, the CRM has to integrate the history of completed projects, the volume billed per client, the activity trend and the referrals each client has generated.
These differences are not cosmetic. They determine whether the team uses the system or not.
The artificial intelligence layer that changes everything
Sector-specific CRM existed before AI. What AI adds is a layer of automation that resolves the adoption problem at its root.
Instead of the sales rep having to remember to update the status of each opportunity, the system detects the signals that indicate progress or stagnation: whether the client opened the quote sent, whether more than ten days have passed with no activity on an opportunity that should be moving forward, whether the behaviour pattern of this lead resembles that of leads that have historically closed or that of those that have been lost.
AI CRM does not simply store information. It processes it, detects patterns, generates relevant alerts and, in the case of the most advanced systems, drafts personalised follow-up messages for each client in the specific context of each conversation.
The practical result is that the sales rep arrives at Monday’s call with a four-line briefing on each client that covers the key points from the last conversation, the quote pending response and a suggestion of what to cover in this call. Without anyone having had to prepare that briefing manually.
The cost argument that no longer works
For years, the argument for using a generic CRM was an economic one: Salesforce costs X euros per user per month but it is already built and works. Building something bespoke would cost ten times more.
In 2025, that calculation has changed. Open-source language models, combined with modern development tools, have reduced the cost of building a bespoke CRM by an order of magnitude compared to five years ago. A custom CRM for the sales process of a mid-sized company can be built in four to six weeks and maintained at a fixed monthly cost that does not depend on the number of users.
Compared to the forty, fifty or one hundred euros per user per month that generic CRMs charge with all features enabled, the economics of bespoke CRM are increasingly favourable for businesses of ten or more users.
What’s coming: vertical CRM as a service
The emerging trend is not that every company builds its own CRM from scratch. It is that vertical CRM platforms appear — specialised in a particular sector — offering sector-level customisation with the subscription model of generic CRMs.
A CRM built specifically for mechanical workshops. One for estate agencies. One for clinics. One for marketing agencies. Each with the right vocabulary, pipeline, fields and alert logic for that sector, built on a platform that can be adapted to the specific nuances of each business without starting from scratch.
At BAI, we are building exactly that. Not a universal CRM that promises to serve everyone. But commercial management solutions designed specifically for the sectors we work in.