In a very short time, we have gone from using bots to solve basic queries—for example, when chatting with a company via WhatsApp—to seeing how artificial intelligence steps in as a protagonist in the promotion and sale of products and services. What was once an assistant that answered questions has now become an AI agent, an active AI agent, that searches, profiles, argues, and closes opportunities. This AI agent no longer just responds but anticipates customer needs, functioning as a true commercial AI agent within the company.
For example, companies like the one you mentioned—Mentius, partnered with InConcert technologies—no longer just automate customer service processes but train AI agents with generative AI to “aggressively” make calls, send messages, identify profiles, and offer complete service portfolios in telecommunications. Each AI agent is designed to assume responsibilities that were previously human, becoming an autonomous AI agent within the commercial cycle. This marks a paradigm shift: from the human who sells to the human who closes, and from the AI agent that handles the bulk of the commercial cycle, where each interaction can be managed by a different AI agent.
This type of transformation has profound implications: more sophisticated automation, displacement of routine tasks, new interaction norms, and a redefinition of the human salesperson’s role in relation to the AI agent. In the following sections, led by the ITD Consulting team, we will explore how each AI agent is trained, what their concrete participation is, what limits they have, what benefits they bring, and what precautions should be taken when integrating an AI agent into sales and customer service processes.
How is an AI agent trained to “sell”?
Training an AI agent for sales is not simply allowing an AI agent to answer questions using fixed scripts. In the past, customer service systems basically responded to “if you ask X → answer Y”; but in sales, the environment is more dynamic, and an AI agent must handle objections, improvisation, tone, customer motivations, multiple offers, and competition. For this reason, the technical challenge of each AI agent is much greater.
In traditional systems, AI operated with rigid scripts: “choose option A or B, respond with phrase X,” but today, with generative AI and machine learning, each AI agent can be trained to propose products, argue in real time, and adapt its speech to the conversation. For example, as explained by an executive at Mentius, each sales AI agent must have a certain capacity for improvisation, simulate spontaneity, and manage variations, rather than just following strict rules as a traditional AI agent would.
To achieve this, several techniques are employed with each AI agent:
- Supervised learning: The AI agent is fed examples of real conversations (humans selling), sales scripts, common objections, and successful responses so that each AI agent learns patterns of effective interaction.
- Reinforcement or adaptive learning: AI agents are exposed to varied (simulated) scenarios where they must decide what action to take, and they are rewarded when results are positive, training each AI agent to improve through experience.
- Generative language modeling: Large language models (LLMs) are used so that each AI agent generates proposals, responses, and personalized arguments in real time, making the AI agent behave convincingly and naturally.
- Integration with customer and portfolio data: Each AI agent accesses information about products, conditions, discounts, and customer profiles with the highest likelihood of conversion, ensuring the AI agent can make informed decisions during interactions.
Training also includes variability in the discourse of each AI agent: informal or formal tone, adapted to the type of customer, the channel (chat, WhatsApp, call), and the timing of the interaction. It requires constant testing, fine adjustments, and human supervision to ensure that each AI agent does not act inappropriately or outside the sales script.
The role of human supervision
Although the AI agent performs complex tasks, it is not alone. Humans continue to intervene in critical phases, while the AI agent handles offering, profiling, and initiating the sales funnel. In the Mentius example, the AI agent manages these initial stages, while the final closure, sales legalization, contract signing, and handling of sensitive or regulated issues remain in human hands. This mixed approach (AI agent + human) combines efficiency with compliance assurance.
In practice, the AI agent delivers a pre-qualified lead to the human agent, with filtered and profiled information, ready to close. This allows the AI agent to reduce repetitive workload so the human salesperson can focus on strategic or empathy-driven tasks. Automating the “dirty work” through the AI agent—generating leads, segmenting, and arguing—paves the way for higher productivity for the human team.
When does the AI agent intervene, and when does the human?
To understand the interaction between AI agent and human, it is useful to view the sales cycle divided into phases:
1. Lead acquisition: The AI agent can identify prospects and contact them via automated calls, text messages, or messaging apps.
2. Pre-qualification: The AI agent asks the customer questions (by voice, chat, or app), determines profile, interest, budget, competition, and urgency.
3. Argumentation / offer: The AI agent presents portfolio options, adapts the offer to the customer profile, responds to simple objections, and shows benefits.
4. Transfer to human: When the opportunity is “hot” (appropriate profile, clear intention), the AI agent transfers the lead to the human agent with all context ready: customer data, previous responses, and proposed products.
5. Closure, legalization, and customer service: The human handles signing, regulatory compliance, transaction documentation, and resolution of complex or sensitive cases.
In this framework, humans are freed from most of the “mechanical” or repetitive part of the sales process, which is performed by the AI agent: standard offers, initial calls, frequently asked questions, and interest filtering. This is an advantage because humans are not as consistent throughout the day as the AI agent: they may get distracted, lose the script, become tired, or vary in tone. In contrast, the AI agent maintains constant focus.
Main benefits of using generative AI for sales
The adoption of this type of AI agent brings several benefits, both operational and strategic. The main ones are:
Operational efficiency
By assigning repetitive tasks to the AI agent—contacting, qualifying, arguing—human resources are freed to focus on tasks requiring judgment, empathy, or complexity. This reduces time per lead, improves response rate, and optimizes cost per conversion thanks to the AI agent’s constant work.
Scalability
The AI agent can operate 24 hours a day, without fatigue, shift changes, or loss of quality. This allows a company to manage a much higher volume of leads without proportionally expanding the human team, all coordinated by the AI agent.
Consistency and Quality of Speech
The AI agent does not “forget” the script, nor does it deviate due to fatigue or distraction. This maintains a consistent level of quality in every contact, reduces errors, and improves message coherence. This is especially important in regulated or highly competitive sectors, where the accuracy of the AI agent makes a difference.
Better Qualification of Human Leads
Because the AI agent performs an initial filter, when the case reaches the human it is already more advanced (“pre-qualified”). This improves human agent productivity, as they can focus on closing opportunities with real potential, while the AI agent performs the initial exploration and filtering work.
Regulatory Compliance and Traceability
The AI agent can record every interaction, generate conversation logs, decision data, and offer menus. This facilitates compliance with sales regulations, audits, and transparency. In sectors such as telecommunications or finance, the traceability provided by the AI agent is key.
Customer Adaptation
The generative AI agent allows for more dynamic personalization of speech and offers. Based on customer data, behavior, and context, the AI agent adapts its argumentation strategy so that the customer perceives a more human conversation, even though it is an automated system.
Competitive Advantage
Companies that adopt AI agents for sales can gain an edge over competitors who maintain more traditional processes. They can respond faster, handle higher volumes, maintain consistency, and reduce costs. In a market where the speed and quality of contact matter, the AI agent can make a difference.
Case: How to Structure Implementation in a Service Company?
For a service sector company (e.g., telecommunications), deploying an AI agent for sales can follow a structured plan:
Step 1: Diagnosis of the Current Process
Map the sales flow with a focus on the AI agent: lead capture, qualification, argumentation, closure, legalization, follow-up. Identify bottlenecks: contact time, conversion rate, drop-off, human error, consistency of AI agent speech. Also define lead volume, customer profile, and products/services offered so that the AI agent can operate optimally.
Step 2: Objective Definition
What do you want to achieve with the AI agent? Examples: increase the percentage of leads qualified by the AI agent, reduce time from contact to offer managed by the AI agent, increase conversion rate of leads filtered by the AI agent, reduce cost per sale, improve regulatory compliance, provide 24/7 contact via the AI agent. Having clear objectives will allow measurement of the AI agent’s results.
Step 3: Technology and Vendor Selection
Choose the generative AI platform, virtual AI agents, integration with CRM/ERP, call or messaging capacity, data capture, and reporting of AI agent indicators. Verify that it has human supervision functions, traceability, and regulatory compliance so that the AI agent operates safely. Establish budget, schedule, and resources for deploying the AI agent.
Step 4: Agent Training
Design sales scripts and analyze successful conversations to train the AI agent. Extract common objections, define customer profiles, design the offer speech, and determine the qualification triggers the AI agent will use. Train the AI agent with varied scenarios and run pilot tests to adjust tone, behavior, and variations. Establish metrics for the quality of the AI agent’s speech.
Step 5: Pilot Launch
Start with a controlled volume of leads in a specific segment, allowing the AI agent to operate under controlled conditions. Measure key metrics: AI agent contact time, response rate, percentage of leads qualified by the AI agent, percentage of sales closed after AI agent intervention, error rate, customer satisfaction. Adjust technical, script, or integration issues of the AI agent.
Step 6: Scaling and Optimization
When the pilot shows positive results, expand the AI agent implementation. Adjust patterns according to data: which hours work best, which channels (call, WhatsApp, SMS) generate the best response, which type of message generates higher conversion, which customer segments respond more. Integrate continuous feedback to the AI agent to optimize performance.
Step 7: Human Role Review and Training
Redefine human agent roles considering the tasks now assumed by the AI agent: initial qualification, product offering, standard argumentation. Train the human team to collaborate with the AI agent, clarifying that the AI agent does not fully replace humans but enhances human efficiency, allowing humans to focus on higher-value tasks.
Step 8: Governance, Ethics, and Supervision
Define policies for AI agent use, supervision of interactions, quality control, data auditing, and regulatory compliance. Ensure that AI agent interactions are traceable, that human intervention occurs when necessary, and that the customer is aware they are interacting with an AI agent if regulations require it. Monitor customer perception and acceptance of the AI agent.
Step 9: Continuous Improvement Culture
Promote a culture in the company where teams work with data generated by the AI agent to learn, adjust strategies, improve scripts, and adapt offers. There should be feedback from humans to the AI agent and from the AI agent to humans. This allows the AI agent to evolve, improve performance, and adapt to new market conditions.
The emergence of generative AI agents in the sales process represents one of the most relevant transformations in commerce and services in recent years. It is no longer just about “serving the customer” or “resolving doubts,” but the AI agent actively participates in sales: contacting, profiling, offering, arguing, and preparing for closure.
By integrating the AI agent in this way, organizations can increase efficiency, improve speech quality, scale contact volumes, and offer more personalized experiences. At the same time, human operators are reoriented toward strategic, complex, sensitive tasks that require human judgment, while the AI agent handles most routine interactions.
Of course, not everything is simple or automatic: careful training, technological infrastructure, human supervision, ethical attention, and cultural adaptation are required. Companies that know how to harmoniously combine AI agent + human will be better positioned to compete, especially in high-volume industries with repetitive tasks, digital customers, and intense competition.
Today, some AI agents may already be conducting part of the sales cycle, but the real change lies in thinking of the AI agent as another salesperson, available 24/7, with a fatigue-free script, instant access to customer data, and referral to humans only when closure, signing, or ratification is required. This vision is redefining what it means to “sell” in the 21st century.
If you want to implement AI agents strategically and safely, ITD Consulting can help you design and integrate these solutions into your sales process. Write to us at [email protected] and discover how to enhance your results with generative AI.
EN 
ES