The New Operating Model for AI-Powered Sales

Mario Pufahl · 2026-07-31

Why Copilot, CRM and individual AI use cases do not yet add up to transformation

Many companies are investing heavily in artificial intelligence in sales. They are introducing copilots, testing automated conversation summaries, using generative AI to write emails or building their first sales agents. The number of use cases is growing. Yet the measurable impact often falls short of expectations.

The reason is simple: individual AI applications do not yet change a sales system. Embedding AI sustainably in sales takes more than technology. It requires a new operating model: a clear approach to how strategy, processes, people, data, AI agents, leadership and governance will work together.

The key question is therefore not: which AI solution should we introduce? It is: how should we organise sales so that people and AI achieve better results together?

Traditional sales was built for a different era

Many sales organisations still use an operating model based primarily on human activity. Salespeople research accounts, prepare conversations, document meetings, write follow-ups, maintain opportunities, create forecasts and coordinate internal experts. CRM systems support this work but do not perform it.

The consequence is familiar: much of a salesperson’s time goes into administration, research, coordination and documentation rather than customer interaction.

AI fundamentally changes this starting point. It can combine information, analyse customer histories, identify opportunity risks, prepare meetings, recommend next actions and automate parts of communication. This changes not only the division of work, but the entire system of sales management.

The operating model has seven building blocks

A future-ready approach to AI in sales cannot be designed through tools alone. It needs seven connected elements:

• Strategy and target vision
• Customer-centred sales processes
• Data and knowledge
• AI agents and automation
• Roles and skills
• Leadership and performance management
• Governance and continuous improvement

Only when these elements work together do individual AI initiatives become an effective sales model.

1. Strategy: which problem should AI solve?

Many AI programmes start with technology. That is the wrong starting point. Begin with a clear business objective. Should AI increase productivity, improve win rates, shorten sales cycles, make forecasts more accurate, identify growth opportunities in existing accounts or improve customer interactions?

Without priorities, a collection of disconnected pilots quickly emerges. An effective target vision answers three questions:

• Where exactly should AI create value?
• Which decisions should become better or faster?
• Which tasks will deliberately remain with people?

AI in sales is not an end in itself. It must contribute to revenue, margin, retention, speed or quality.

2. Processes: do not automate poor workflows

Many sales processes have evolved over time. They contain disconnected systems, redundant approvals, inconsistent practices and extensive manual work. Simply adding AI may only automate existing inefficiency. Decide how the process should work before automating it.

Review the core sales processes:

• Lead to opportunity
• Opportunity to close
• Account management
• Proposal management
• Forecasting
• Renewal and expansion
• Sales planning
• Customer handover to delivery or service

For each process, identify which steps should be standardised, supported, automated or deliberately kept human. In a complex opportunity, AI may analyse the account, identify missing stakeholders, flag risks and suggest next steps. Assessing buying-group politics, maintaining the decision-maker relationship and negotiating remain human tasks.

The objective is not maximum automation. It is the best division of work.

3. Data and knowledge: AI’s real infrastructure

AI-powered sales is only as good as its data foundations. Missing CRM entries, outdated contacts, contradictory product information and unstructured documents do not create more intelligent results. They simply accelerate poor decisions.

The new operating model therefore needs a reliable knowledge architecture, including:

• CRM data
• Customer interactions
• Proposal and contract information
• Product and service knowledge
• References and use cases
• Market and competitor information
• Pricing and margin rules
• Internal expert profiles
• Governance and compliance rules

The point is not to collect as much data as possible. It is to make relevant data available in the right context. Data quality becomes a strategic sales capability rather than an administrative task.

4. AI agents: from assistant to digital team member

The next stage goes beyond individual copilot features. AI agents can analyse and prepare tasks independently, and execute them within defined boundaries. Specialised sales agents might include:

• Account research agent
• Meeting preparation agent
• Opportunity risk agent
• Proposal agent
• CRM documentation agent
• Forecast agent
• Renewal agent
• Sales coaching agent
• Pricing agent
• Next-best-action agent

One super-agent should not do everything. A more useful architecture consists of specialised agents with clear responsibilities, data access and escalation paths. The salesperson no longer just works with software; they orchestrate a digital sales team.

This also changes the technology architecture. CRM, collaboration, knowledge management, the data platform and the AI layer need to connect. CRM remains important, but it is no longer the sole centre of sales.

5. Roles and skills: the salesperson becomes an orchestrator

When AI takes on research, documentation and analysis, salespeople do not automatically become less important. Their contribution shifts. More value comes from capabilities that are difficult to automate:

• Building trust
• Explaining economic value
• Facilitating complex decisions
• Recognising political dynamics
• Bringing stakeholders together
• Supporting change
• Setting priorities
• Taking responsibility

New skills also emerge. Salespeople must learn to assess AI outputs, define good assignments, question data critically and use digital agents purposefully. AI literacy becomes as natural as CRM competence is today, but it is not enough on its own. Effective salespeople combine AI skills with business acumen, industry knowledge and relationship skills.

6. Leadership: from managing activity to managing the system

Sales leadership changes too. Traditional models often focus on activity counts: conversations, meetings, proposals or CRM entries. In AI-powered sales, activity alone becomes less meaningful. If agents create emails, research accounts or update data, manual activity volumes no longer provide a useful measure of performance.

Leaders need to manage the whole system, including AI agents. This includes:

• Quality of opportunity decisions
• AI use and acceptance
• Speed through the sales process
• Data quality
• Forecast accuracy
• Customer value
• Win rate
• Margin
• Growth of strategic accounts

The sales leader increasingly becomes the architect of an effective system. They decide where to standardise, where flexibility is needed and how people and machines work together.

7. Governance: autonomy needs clear boundaries

The more AI participates in decisions and processes, the more important governance becomes. Not every task may be performed autonomously:

• A pricing agent may recommend prices, but should not approve individual discounts without defined limits.
• A proposal agent may create drafts, but legally or commercially significant statements must remain reviewable.
• A customer communication agent may prepare follow-ups. Sensitive customers, escalations and contractual questions require a human in the loop.

Governance answers questions such as:

• Which data may an agent access?
• Which decisions may it prepare?
• Which tasks may it execute independently?
• When is human approval required?
• How are results documented?
• Who is accountable?
• How are errors detected and corrected?

Governance should not be seen as a brake. Good governance builds trust and makes controlled scaling possible.

The target: a hybrid sales model

The future operating model is neither entirely human nor fully autonomous. It is hybrid. AI primarily handles data-intensive, repetitive, rule-based or time-critical tasks. People handle tasks requiring judgement, empathy, responsibility, creativity and relationship work.

The strength lies in the combination: AI provides speed, scalability and analytical depth. People provide context, trust and judgement. Companies should systematically strengthen the salesperson rather than try to replace them.

How companies should begin

A new operating model does not emerge through a big-bang launch. A step-by-step approach makes more sense.

Stage 1: Assistance

Start with bounded tasks: meeting summaries, conversation preparation, CRM documentation, email drafts and lead qualification. The aim is quick productivity gains at low risk.

Stage 2: Intelligence

Next, use AI to analyse more complex situations: opportunity risks, missing stakeholders, cross-selling potential, forecast deviations and next best actions. The aim is better decision quality.

Stage 3: Automation

Then automate defined steps: data updates, standard follow-ups, internal information gathering, proposal components and renewal reminders. The aim is scalability.

Stage 4: Orchestration

At the highest stage, several agents work together across the sales process. People lead, assess and decide. The aim is intelligent collaboration, not autonomy at any price.

The new KPIs of AI-powered sales

A new operating model needs new measures. Alongside revenue, margin and win rate, further KPIs become important:

• Administrative sales time saved
• Reduction in sales-cycle length
• Quality and freshness of CRM data
• Forecast accuracy
• AI adoption rate
• Use of recommended next best actions
• Conversion rate at each process stage
• Share of automated process steps
• Quality of opportunity assessment
• Revenue per productive sales hour

Do not confuse technology use with success. The number of agents is not a business KPI. What matters is whether they enable better sales outcomes.

German original on LinkedIn