Client story Scrubbed’s AI Early Warning System Cuts Research Time and Boosts Client Care

Leading through proactive support and constant care

Scrubbed leads the way in the accounting and financial professional services world by taking over back-office work so their clients can focus on growth. Alaina Payawal, the AI project manager at Scrubbed, sees the company as a bridge that connects expert financial support with the daily needs of U.S. businesses.

By handling bookkeeping, tax, and audit support, Scrubbed lets clients stay ambitious and focused on their own missions. As the industry moves toward faster, data-driven decisions, Scrubbed stays ahead by making sure their team has the best tools to care for every client relationship.

The challenge: Facing manual research burdens and reactive client care

Before starting this engagement, Scrubbed relied on the gut feelings of their account managers to track the health of client relationships. The team looked for amber signals, those subtle virtual “yellow indicator lights” like slower email replies or a change in email tone. While these signals often warned of larger risks, the team lacked a way to capture them in a structured or real-time format. “It really depended on who was managing the account, which meant the insights weren’t always consistent,” says Payawal.

This manual approach created a heavy burden for the Scrubbed staff. Skilled professionals spent hours checking LinkedIn, news outlets, and press releases just to stay informed about their clients. This effort added up quickly as the client base grew, making it hard for every team member to stay proactive. When the team missed an amber signal, such as a leadership change or a shift in client strategy, they lost the chance to step in early.

Additionally, their data lived in silos. The internal Customer Relationship Management (CRM) notes showed how often the team talked to a client, but they didn’t show what happened in the external market. Without connecting these two views, Scrubbed only saw part of the picture. Highly skilled staff acted as manual data researchers instead of focusing on strategic work that adds value. Scrubbed needed a way to use AI to handle the heavy lifting of data gathering and processing while keeping their leaders in control.

The solution: Building a modular AI engine with Gemini

Insight worked closely with Scrubbed to realize a vision for a Google Cloud AI-driven early warning system. This system runs in the background, monitoring both internal notes and external signals to give a clear view of client health. To build this, Insight used an agent development kit. This framework let the team build specialized agents, like a search agent and a client hub agent, that work together with an agent manager, or orchestrator, as one cohesive brain to trigger health status updates to the database.

The core of the solution uses Google Cloud’s Gemini Enterprise Agent Platform (GEAP) and the Gemini Large Language Model (LLM). Insight helped Scrubbed translate their business rules into a logic core that defines client health as “green,” “amber,” or “red.” The system analyzes unstructured data, such as free-text meeting notes and news articles, to find the “why” behind a client’s status.

“It was really amazing how the Insight team worked closely with us to define what green, amber, and red actually mean in our business context,” says Payawal. “From there, they translated those definitions into structured logic powered by GEAP and the Gemini LLM. That alignment between business understanding and technical implementation was critical.”

To make the system scalable, Insight used additional Google Cloud technology, including Cloud Run and Cloud Tasks. These tools let Scrubbed process the health status of hundreds of clients at once without slowing down. A Cloud Scheduler integration runs the system twice a day, so no one has to check things manually.

Keeping a human in the loop

The design also includes a human in the loop as a collaboration between AI and human expertise. Account directors review the AI’s findings and can update a client’s status if they have deeper context. When changes are made, the specific reasoning feeds back into the agent to refine its logic, helping it learn and improve over time. To maintain agility, prompts are stored in a central Google Cloud Storage bucket, which lets the business team update health criteria in real time without a developer to rewrite code. “Having a human in the loop actually keeps the system grounded,” says Payawal.

To ensure trust, the Insight team built reasoning guards into the prompts. These guards make sure the AI evaluates the correct company and doesn’t get confused by firms with similar names. The logic also leans toward being conservative; if signals conflict, the system defaults to an amber or red status to make sure the team doesn’t miss a risk.

The outcome: Achieving real-time visibility and proactive client health

The new system transforms how Scrubbed manages their client relationships by moving from reactive fixes to proactive care. By automating the research process, Scrubbed gives their team a single source of truth for every client.

Quantifiable benefits

• Slashed assessment time of client health from hours to under 60 seconds
• Cut manual research tasks of account managers by 80%
• Automated daily health checks with Cloud Scheduler to run twice a day at preset times to gauge client health status
• Scaled client account analysis with Gemini to process hundreds of clients in parallel without performance lags

Qualitative improvements

• Unified framework: The new system replaced individual gut feelings with a shared, objective way to measure client health across the organization.
• Better preparation: Account managers now enter meetings with deep, real-time knowledge about client funding, partnerships, and strategic shifts.
• Stronger relationships: By knowing about client changes before the client mentions them, Scrubbed shows they truly care and are invested in client success.
• Focused expertise: Highly skilled professionals now spend their time on strategic outreach and relationship building rather than hunting for data.

Future outlook

The success of this system gives Scrubbed the confidence to expand AI into other parts of the business. The modular design makes it easy to add new data sources, like email sentiment or payment patterns, in the next phase.

Scrubbed also plans to use this early warning approach to support their own staff. By tracking signals for employee turnover, the company hopes to provide a better environment and retain the great talent they have. “Scrubbed wants to use AI to support people, not replace them,” says Payawal. “You should definitely make AI work for you and not do your work.”

Build an AI ready workforce. Discover a five-step strategy and actionable courses to assess needs, build trust, and upskill teams with Google Cloud.