How Smartsheet runs CX on customer truth instead of opinions

Smartsheet's support team rolled out a redesigned Pro Desk sign-up page. The new design looked clean. Customers were missing their coaching sessions.

Corinne Flanagan, Senior Manager of Enablement and Quality at Smartsheet, runs the team that found the cause. Her group ran the Customer Research Worker. It identified relevant conversations, analyzed the underlying transcripts, and the answer came back: customers were getting tripped up by time zones and booking sessions in the middle of the night. The tech team fixed the display before the issue grew.

This is one example. Smartsheet's Workers deployment spans eight different analysis roles, runs across their full conversation dataset, and feeds the other Level AI products they already use.

Before AI Workers: Data Without Answers on Demand

Smartsheet is the intelligent work management platform that unites people, data, and AI. Corinne's team supports a large customer base across Support, Professional Services, and Finance, and they generate a substantial volume of conversation data every day.

Before Workers, that data was abundant but not interrogable on demand. When support met with product leaders to advocate for changes, the team could describe what customers were experiencing, but pulling the specific evidence to support a claim required manual review.

Corinne described the dynamic this way: "It used to be that we would get in a room and talk about how customers are feeling frustrated about X. And it was all feelings. We had so much data, but we didn't have anything quantifiable."

The data was there. The path from data to a specific answer was the missing piece.

After AI Workers: Eight Specialized Workers, nearly 500 Runs, One Conversation Dataset

Smartsheet runs eight Workers, each scoped to a recurring analysis job:

• Coaching Plan Worker, Generates tailored, structured coaching plans.
• Customer Research Worker, Conducts research with direct quotes as evidence.
• Conversation Analytics Worker, Deep quantitative analysis across all of your conversations..
• Product Feedback Worker, Quantified product feedback from conversations.
• Team Performance Worker, Cross-team analysis and workload trends.
• Executive Research Worker, Compiles deep, leadership-ready, multi-step research.
• Sentiment Insights Worker, Insights from advanced sentiment analysis.
• VoC Insights Worker, Insights from VoC and constituent AI factors.

Each Worker has access to the conversation data Smartsheet already feeds into Level AI for QA scoring and Voice of the Customer dashboards. This is the cross-SKU value Corinne's team gets from the deployment. AutoQA scores every interaction. Voice of the Customer surfaces patterns by topic. Workers make use of the AI enriched data from the fleet of small language AI models used in QA, VoC and other Level AI products and answer specific questions when the team needs to investigate something the dashboards do not already show.

The Pro Desk investigation is what that looks like end to end. The team noticed an uptick in support cases following the launch. They asked the Customer Research Worker “What are the issues customers encountered whilst trying to schedule Pro Desk sessions?” to find what was driving the cases. The output identified the time zone confusion directly from customer language in the transcripts.

Customer Research Worker returned the specific failure pattern in the same language customers used to describe it, which is what the product team needed to act on.

How Workers Connect to the Rest of the Stack

Workers is one part of Smartsheet's Level AI footprint. AutoQA handles quality review across interactions. Voice of the Customer powers role-specific views, including a product dashboard that maps call topics to satisfaction trends and a customer success view that ties sentiment to revenue and account risk.

Workers make the same conversation data queryable on demand. Smartsheet uses workers everyday to ask questions like “Can you investigate convos with the following keywords related to accessibility?”, “What are recurring process or policy issues in the last 30 days?”, and “Can you show convos related to Brandfolder where customers indicated technical issues like the following?”

Corinne summarized the shift this way: "That's been the power of Level for us. It's no longer just debating. These are facts. These are customers' words. It's been incredibly effective and powerful that way."

Why This Matters for the Customer Intelligence Gap

Support organizations hold enough conversation data to answer almost any question about their customer experience. The constraint is the time and skill required to ask each question. By the time the analysis lands, the launch is already in trouble, the account is already at risk, or the product decision has already been made.

AI Workers closes that gap by collapsing the time between question and answer. A support manager who wants to know why cases spiked after a release does not need to file a request and wait.

Workers have evolved to get smarter and more proactive. Workers can react to triggers based on every metric, metadata, and AI enriched data available across Level AI, CRMs, CCaaS, and more to proactively conduct research and email the relevant stakeholders.

Smartsheet's Pro Desk fix is what that looks like in practice. A launch issue identified, located, and resolved before it ever showed up in a quarterly review.

Build your own AI Worker

Level AI is opening up Worker creation to customers. Teams will be able to scope Workers to their own CX functions, connect them to their own data, and put them into production alongside the Workers Level AI ships out of the box. Join the waitlist to get early access.

Join the waitlist →