A Harvard Business Review article argues that agentic AI is weakening the traditional boundary between marketing and sales. Agents can carry customer context across research, personalization, outreach, qualification and follow-up, while companies still tend to deploy the technology inside separate departments. The proposed direction is a more integrated commercial system built around shared data, coordinated workflows and common outcomes.
The article is right about the customer experience. Buyers do not care which department owns a message, and they should not have to repeat information whenever an internal handoff occurs. Separate systems can produce contradictory offers, duplicated outreach and conversations that ignore what the customer has already said. Adding autonomous agents to those systems could reproduce the same fragmentation at much greater speed.
But a continuous customer journey does not make sales and marketing the same function. It makes coordination between them more important. Software can move across an organizational boundary without making the work on either side interchangeable. A CRM already contains campaign activity, account history and sales conversations; nobody concludes from that shared database that positioning a product and negotiating a contract are identical jobs.
Marketing generally works across markets and longer time horizons. It studies audiences, shapes positioning, builds memory, creates demand and tests messages across groups of potential buyers. Sales works with particular opportunities under immediate commercial constraints. It discovers account-specific needs, manages relationships, handles objections, negotiates terms and decides how to advance or abandon a deal. AI may perform tasks in both categories, but the underlying decisions still involve different evidence, skills and risks.
Common revenue measures can also conceal more than they clarify. Revenue is an essential company outcome, but it is a late and incomplete measure of marketing. If every activity is judged by visible pipeline contribution, work with delayed or uncertain effects—brand building, market research, category creation and experimentation—will predictably lose resources to actions nearest the transaction. Better attribution does not solve this problem. AI can make an attribution model more elaborate while leaving its assumptions just as contestable.
The same distinction matters for accountability. When an agent turns a campaign signal into personalized outreach, someone must decide which claims are permitted, when a lead is ready for contact and when automation should yield to human judgment. Pricing, contractual promises and account strategy cannot become ownerless simply because one workflow touches several departments. A unified commercial layer without explicit decision rights may replace a visible handoff problem with a less visible governance problem.
The appropriate boundary will also vary by business. A small company may already have one person handling both functions. A product-led software company, a consumer brand and an enterprise supplier with an eighteen-month buying process do not need the same structure. Treating technological overlap as a general organizational destination risks turning one useful operating model into another fashionable reorganization.
The better addendum is that AI can connect sales and marketing without collapsing their differences. Companies should share customer context, coordinate agents, establish escalation rules and measure outcomes across the journey. They should also preserve specialist judgment and make ownership clearer, not blurrier. Integration is valuable because the functions contribute different things to the same commercial system—not because technology has made those differences disappear.