Alexandre Pericão The world of marketing and sales is undergoing its deepest transformation since the emergence of search engines. For decades, CRM was seen merely as a “system of record”, that is, a place to store contacts and meeting notes reactively. However, we are entering an era where CRM becomes the “operational brain” of companies, powered by Artificial Intelligence (AI) to predict behaviours, personalise experiences at scale and automate processes that previously required hours of manual work.
1. The Problem of “Blind” AI and the Contextual Solution
Currently, more than 70% of CRM users are already experimenting with AI tools, but most face a critical obstacle: AI is “blind” to business context. If you ask a generic chatbot to write a follow-up email, it will not know whether the client has already received three previous emails, whether they opened the last proposal, or whether they have a pending support ticket.
The great revolution of AI Connectors (present in platforms such as HubSpot, Salesforce and Zoho) is precisely to create a native bridge between AI’s processing power and the truth of your data. By giving AI “eyes” through your CRM, it stops providing generic advice and starts offering insights based on the historical reality of your contacts, deals and interactions.
2. Market Specialists: Salesforce, HubSpot and Zoho
For those who are not technical specialists, choosing the right tool may seem challenging. However, the three main platforms in the market offer distinct approaches to integrating AI into daily operations:
Salesforce and Agentforce
Salesforce introduced the concept of Agentforce, which goes beyond a simple chatbot. These are autonomous agents that help sell more intelligently by eliminating what they call the “Toggle Tax”, meaning the friction of switching between ChatGPT for research and Salesforce for data updates. With direct integration into ChatGPT, salespeople can now update opportunities or create strategic account plans directly within a natural conversation, without ever leaving the interface they already use for research.
HubSpot and the Trio of Connectors (GPT, Claude, Gemini)
HubSpot opted for a native connector approach for the three major models in the market:
ChatGPT: Ideal for deep research (Doctor-level analysis). It excels at analysing patterns in large volumes of data and identifying market trends.
Claude (Anthropic): Currently the most capable for execution. It is the only one that allows “writing back” to the CRM, creating notes, tasks or updating deals directly via chat, including through mobile devices.
Gemini (Google): The logical choice for teams that live within Google Workspace. It allows pulling CRM data directly into the Gmail or Google Docs sidebar, making it easier to draft emails with real-time context.
Zoho and the Zia Assistant
Within the Zoho ecosystem, AI is called Zia. It acts as a decision engine that spans the entire company. Zia allows any employee to ask questions in natural language (such as “What was the most profitable product in the last quarter?”) and obtain instant reports, eliminating the need to know how to build complex dashboards.
Zia Predictions: Anticipating Before Reacting
Zia allows predictive models to be built based on the company’s own historical data.
One of the most relevant features is churn prediction. The system analyses interaction patterns, purchase history, support behaviour and activity frequency to assign a statistical probability of churn to each customer.
This makes it possible to activate retention workflows before the customer leaves, transforming a reactive CRM into a preventive system.
Beyond churn, the Prediction Builder allows the creation of customised forecasts. It is possible to predict:
• Probability of closing a deal
• Probability of upgrade
• Probability of delay or default
• Any strategic field defined in the CRM
The user selects the field to be predicted and the historical data that feeds the model. No advanced technical knowledge or data science team is required.
This capability is particularly powerful in Revenue Operations contexts, where correct pipeline prioritisation directly impacts revenue.
Zia Recommendations: From Insight to Recommended Action
Beyond predicting, Zia recommends.
The Next Best Experience functionality analyses historical patterns and suggests the ideal next action for each customer or opportunity.
Example:
Customers with similar profiles converted after a specific approach.
The system suggests repeating that action.
This is not just about alerts. It is prescriptive guidance.
Additionally, Recommendation Analytics allows measuring the impact of these recommendations:
• Acceptance rate by teams
• Revenue impact
• Effectiveness of suggestions
This creates a continuous learning cycle: AI recommends, measures, learns and adjusts.
In the SME and mid-market context, this prescriptive layer is often underused. However, this is where the difference between basic automation and true operational intelligence becomes evident.
3. The Shift in Information Consumption: From SEO to GEO
The impact of AI is not limited to internal productivity. AI has changed the way customers find you. Traditional organic traffic is experiencing erosion due to AI Overviews (AI summaries at the top of search results) and Zero-Click Searches (where the user obtains the answer without clicking on the website).
For companies, this means that traditional SEO is no longer sufficient. The need for Generative Engine Optimisation (GEO) emerges. The objective now is to ensure that your content is structured in a way that AI models (such as Perplexity or ChatGPT) cite it as an authoritative source. Your CRM plays a role here by providing real data and testimonials that prove your expertise — something AI values when selecting sources.
4. Customer Experience (CX): From Reactive to Predictive
AI makes it possible to elevate Customer Experience (CX) to a level that would be impossible with human teams alone. There are six main ways to apply this intelligence:
- Sentiment Diagnosis: Analysing thousands of emails and tickets to identify dissatisfaction patterns before they become crises.
- Friction Identification: Understanding at which stage of the journey customers drop off or encounter more delays.
- Predictive Segmentation: Identifying who is most likely to buy or who is at risk of churn.
- Real-Time Personalisation: Adapting the content of an email or website based on the user’s immediate intent.
- Conversational Assistants: Bots that not only answer FAQs but also resolve technical issues and schedule meetings by checking team availability in the CRM.
- Dynamic Automation: Flows that adjust to the customer’s pace — if interest is shown, the cycle accelerates; if there is no interaction, the communication channel automatically changes.
5. The Danger of Automation Without Strategy
An important warning for SMEs: technology should serve the relationship, not replace it. Automating everything simply because the tool allows it is a strategic mistake that can drive customers away. When automated messages ignore context (such as sending a generic sales email to someone who has just opened a critical support ticket), the brand appears insensitive and robotic.
Additionally, many SMEs face a crisis of trust in their data. “The numbers don’t match” is a common phrase when information is spread across multiple Excel sheets and isolated tools. Before implementing advanced AI, it is essential to centralise information into a “Single Source of Truth” within the CRM, ensuring that the data feeding AI is clean and reliable.
6. Revenue Operations (RevOps): Teams Aligned for Success
To maximise return on investment in AI and CRM, companies are adopting the Revenue Operations (RevOps) model. This strategy breaks down silos between Marketing, Sales and Support, ensuring that everyone works with the same data and revenue goals. RevOps is based on four pillars:
• Processes: Standardising how a lead moves from marketing to sales.
• People: Evaluating teams not by isolated tasks, but by shared satisfaction and retention goals.
• Data: Centralising everything into a system accessible to all.
• Technology: Using integrated platforms (such as Zoho or HubSpot) to automate what is repetitive.
7. The Near Future: 2026 and the Agentic Era
Looking ahead, marketing will cease to be a static planning exercise and will become a living system. We are entering the Agentic Era, where we delegate complex tasks to intelligent agents that not only suggest, but execute. These agents will pre-qualify leads, prepare meetings and perform follow-ups based on the buyer’s real intent.
Personalisation will move beyond cosmetic (using the name in an email) to hyper-personalisation at scale. Imagine a system that automatically adjusts the offer, the channel and the timing of each interaction based on real-time conversion probability.
Conclusion: The Human Touch in the Digital Age
Artificial Intelligence in CRM does not exist to replace marketers or salespeople, but to augment them. By freeing people from bureaucratic tasks and massive data analysis, technology creates space for what truly matters: empathy, creativity and trust-building.
At Liminal, we believe technology should serve the relationship. We combine strategy, marketing and technology to ensure that these innovations are not just new tools, but real engines of sustainable growth. The future belongs to companies that know how to balance AI’s predictive power with the authenticity of human relationships.
If your company is looking to modernise, there are even incentives such as the “AI for SMEs” Line (integrated into Portugal’s PRR) that financially support the adoption of these solutions.