The Agentic Shift: Why Your RevOps Team is Already Obsolete

The move from human-led to agent-orchestrated operations changes how we define scale. Nate Follen’s work at Perplexity shows that the standard 10 to 15 person RevOps team is being replaced by one human orchestrator managing a fleet of autonomous agents. Competitive advantage no longer comes from headcount. It comes from the ability to build, test, and harden skills. These are reusable, model-agnostic workflows that run on their own across your tech stack. This creates a gap between teams that use AI as a basic chatbot and those that use it as a system for end-to-end business operations. For leaders, the advantage is speed. Work that once required weeks of collaboration now happens in days or minutes.

The Architecture of Autonomy

The most important takeaway from Follen’s approach is the shift from reactive AI--where a human prompts a model for a single task--to proactive agentic workflows. By using a multi-model orchestrator like Perplexity Computer, Follen avoids the risks of relying on one AI.

I usually reach out to people in my network like Rich over at HubSpot and a few folks there to get their opinions but I also like to not trust one model... Perplexity does that automatically, and it's just through something we call Model Council.

-- Nate Follen

This Model Council approach lowers the risk of hallucinations or bias by forcing models to cross-reference their findings against each other and your CRM. The system does not just provide an answer; it provides a map of where models agree, disagree, and offer unique insights. This creates a trust layer that lets the human operator move from manual work to high-level oversight.

From Messy Middle to Hardened Skills

Many organizations currently suffer from AI sprawl, which consists of fragmented, undocumented prompts. Follen argues that true scale comes from a shared skills repository. Instead of letting every employee build their own siloed agents, successful organizations treat skills as a product.

• Evaluation: Skills must be measured against business results, not just output quality.
• Hardening: A thread that works once is a novelty. A thread turned into a recurring skill is an asset.
• Observability: By hooking these agents into logging systems, operators can identify repeatable patterns and automate future workflows.

Once I notice that something's getting long, I will just create it as a skill typically and then run that skill regularly. And so that's where I get the most value out of skills.

-- Nate Follen

This creates a loop: the agent does the work, the human reviews it, the agent learns from the correction, and the workflow is hardened into a recurring job. This is the difference between using AI and building an agentic team.

The Hidden Cost of Manual Reconciliation

Conventional wisdom says CRM hygiene is a manual chore. Follen’s workflow changes this by using agents to reconcile data sources, such as Google Drive contracts and Ironclad CLM data, against the CRM. The result is not just cleaner data; it is the ability to perform audits in hours rather than weeks.

This creates a competitive moat. While competitors struggle with the friction of manual CRM updates, your agents continuously enrich the database and identify opportunities for growth. The effort of setting up these connectors pays off in agility, allowing the team to pivot strategy based on real-time data instead of stale, manually entered records.

Key Action Items

• Map Your Model Council (Immediate): Stop relying on a single LLM for high-stakes decisions. Create a workflow that prompts at least two different models to compare findings against your internal data.
• Audit for Repeatable Threads (Next 30 Days): Identify any task you perform more than twice per week. Stop doing these manually. Convert the prompt chain into a reusable skill or agentic workflow.
• Implement a Skill Registry (Next Quarter): Stop the AI sprawl. Create a central repository for your team’s best prompts and skills. Ensure these are documented, evaluated for accuracy, and shared across the organization.
• Automate the Voice of Customer Loop (Next 30-60 Days): Build an agentic workflow that pulls from transcripts and emails to generate a daily dashboard of key themes, customer feedback, and actionable items for product and enablement teams.
• Transition to Approval-Based Execution (12-18 Months): Shift your role from doer to orchestrator. Move your workflows toward an approval-only model where the agent handles data gathering and campaign creation, requiring only your final sign-off.
• Hardening Infrastructure (Ongoing): If you have technical capability, build a hook that logs your AI session history. Parse these logs to identify patterns you should turn into automated, recurring skills.