AI in RevOps: Scaling Automation Without Losing Human Judgment — Liping Zhang

AI in RevOps: Scaling Automation Without Losing Human Judgment

AI is changing how Revenue Operations teams work, but adopting more AI does not automatically create better revenue execution.

For Liping Zhang, Head of Revenue Operations at Integral, the value of AI comes from using it to reduce operational work, improve decision-making, and give teams better access to revenue insights, while keeping humans involved where judgment matters.

In this episode of Revenue Architects, Liping shares how Integral is applying AI across Lead-to-Cash, forecasting, Customer Success, and sales enablement. She also explains how her team evaluates AI experiments, manages costs, validates AI-generated insights, and decides what is ready to scale.

Here are the key lessons Revenue Leaders can take from her approach.

How should Revenue Leaders think about AI in RevOps?

Liping's starting point is that RevOps should remain a lean function.

Instead of spending most of its time on CRM administration and day-to-day operational work, RevOps should focus on turning data into actionable insights and carrying those insights through to execution.

AI can help remove much of the repetitive work involved in getting there.

At Integral, AI is currently used primarily for automation, data analysis, and insight generation. The company has not given AI complete autonomy. Liping believes the strongest approach is AI working alongside people, with humans retaining responsibility for important decisions.

That distinction matters.

The objective is not to automate RevOps for the sake of automation. It is to give Revenue teams more time to focus on strategy, customer relationships, pipeline quality, and revenue growth.

Where is Integral using AI across Revenue Operations?

Integral has applied AI across several parts of the revenue organization, but Lead-to-Cash and pre-sales have received the largest investment.

Around 90% of the company's AI resources and automation have been focused on pre-sales.

The team uses AI for areas such as:

These workflows reduce the amount of manual work sellers and RevOps teams have to do between customer interactions.

For example, before a meeting, AI can review previous conversations and surface what was discussed, what was promised, who needs to be involved next, and which questions or objections may need to be addressed.

After the meeting, it can analyze the conversation and identify action items or details that might otherwise be missed.

The result is not simply faster administration. It gives sellers more context before they make decisions.

How is AI improving forecasting?

Integral's approach to forecasting had already become more disciplined before AI entered the picture.

The company moved from a usage-based brokerage model to SaaS, which changed the forecasting challenge. The team introduced tighter opportunity criteria, MEDDPICC, multiple forecasting models, pipeline velocity analysis, and customer-health monitoring.

AI now adds another layer to that process.

It can process call recordings quickly and identify information relevant to MEDDPICC and opportunity health. That gives RevOps and sales teams a faster way to understand what is actually happening inside deals.

This is important because a CRM stage alone does not always tell the full story.

A deal may appear healthy in Salesforce while a recent customer conversation reveals uncertainty around budget, decision-making, timing, or business value.

AI can help surface those signals—but Liping does not treat the output as automatically correct.

Why does human judgment still matter when using AI?

Before AI handled many of these tasks, Liping had done the work manually.

She had reviewed customer contracts, built performance metrics, tracked pipeline movement, and calculated conversion ratios herself. That experience gave her an understanding of the underlying process and the mathematics behind the metrics.

So when AI produces a number that does not make sense, she can recognize the problem.

This is particularly important for Revenue Leaders. AI can produce an answer extremely quickly, but speed does not guarantee accuracy.

Human expertise provides the context needed to ask:

AI can accelerate analysis. It cannot remove the need to understand what is being analyzed.

What does “stay in the field one more day” mean?

Her experience has shown her that seemingly simple processes often contain more edge cases than teams realize.

Integral, for example, used spreadsheets to calculate sales commissions for around two years before adopting a commission platform.

Even after automating the process, the company kept manual checks in place for two quarters to make sure the calculations were correct.

That caution was deliberate.

Commission is a particularly sensitive process because errors can quickly damage trust between management and salespeople. The team needed to understand and validate the automated process before fully relying on it.

The same principle applies to AI.

Before automating a workflow, Revenue Leaders should understand how it works manually, where exceptions occur, and what could go wrong.

How is AI changing Customer Success?

Integral has maintained a gross retention rate above 90% and single-digit churn. AI is helping the Customer Success organization build a deeper view of account health.

The team brings together information from multiple sources, including:

Integral also created an internal MCP for its trading-volume tracker and connected it with Claude.

Rather than asking AI to independently determine whether an account is healthy, the team defines the metrics that matter and uses AI to bring those signals together.

This creates a more complete view of each customer.

That context can then inform how account teams work with customers across different regions.

For Integral, trading volume is particularly important because its revenue comes from both a fixed SaaS fee and brokerage commissions generated from customer trading activity with liquidity providers.

So understanding customer behavior is directly connected to revenue.

What can Revenue Leaders learn from Integral's AI experimentation?

Integral did not begin with one centralized AI strategy that every employee had to follow.

Instead, employees were encouraged to experiment.

They created their own dashboards, tasks, integrations, and workflows. Liping compared the result to mushrooms appearing after rain, many ideas emerged quickly.

But the team eventually needed to determine which ones were actually worth scaling.

That led to a collaborative evaluation process.

When someone creates a dashboard, the team asks:

The strongest experiments can then be consolidated and adopted more broadly.

This approach prevents the organization from paying for dozens of disconnected AI experiments that provide little incremental value.

How should companies think about the cost of AI?

AI usage can scale quickly when every employee is running their own workflows, dashboards, and analyses.

Integral addresses this by comparing the cost of AI against the value it creates and against traditional alternatives.

For example, the team found AI-driven data enrichment to be significantly more affordable than traditional approaches such as conferences, trade shows, and paid social campaigns.

The question is therefore not simply:

"How much does this AI workflow cost?"

It is:

“What would this work cost us another way, and what value are we getting from automating it?”

That is a much more useful question for Revenue Leaders evaluating AI investments.

How can RevOps avoid creating an AI free-for-all?

Experimentation is valuable, but experimentation without consolidation can create another problem: fragmented systems and rising costs.

Integral's approach is to let teams experiment, identify the most useful solutions, and then bring those solutions together.

That creates a path from:

Experiment → Validate → Consolidate → Scale

It also gives RevOps a role in governing AI adoption.

Rather than trying to stop experimentation, RevOps can help the organization determine which experiments are actually improving revenue processes.

What does AI mean for the future of RevOps?

AI can analyze thousands of data points, process customer conversations, prepare sellers for meetings, identify missing deal information, and bring together customer signals much faster than a person can.

But people still need to decide:

That is where Revenue Leaders remain essential.

The goal isn't more AI. It's better revenue execution.

Liping's approach offers a practical framework for Revenue Leaders navigating AI adoption.

Understand the process. Experiment with AI. Validate the output. Measure the value. Consolidate what works. Keep humans involved where judgment matters.

The most important lesson may be the simplest one: don't automate a process simply because you can.

First understand it. Then improve it. Then determine where AI can make the work faster, more accurate, or more scalable.

That is how AI becomes part of a stronger Revenue Operations strategy rather than another disconnected tool.

Liping Zhang’s approach to RevOps offers a practical perspective on a rapidly changing revenue landscape: AI can make revenue teams more efficient, but strong processes, human judgment, and disciplined experimentation still matter.

The opportunity for Revenue Leaders is not to automate everything. It is to understand where AI can remove operational work, improve visibility, and help teams make better decisions—then build those capabilities into workflows that can be trusted and scaled.

As AI becomes more deeply embedded in RevOps, the teams that stand out will be the ones that combine technology with business understanding, experimentation with governance, and automation with human judgment.