Quick Take
- Your last month of reviews may already show the service issue you keep missing.
- In CRMBC’s podcast conversation with Harsha Vashisht, AI becomes less abstract and more operational: a way to sort reviews, staffing context, promotions, and customer feedback faster.
- The most useful starting point is not a new AI tool. It is a real restaurant problem.
- Harsha’s advice is practical for operators who want to test AI without turning it into a technology project.
- The full episode gives examples from Uber Eats, customer support, review monitoring, prompting, business data, and local marketing.
Restaurant operators do not need another abstract conversation about AI. They need to know where it can help in the daily work of running a restaurant.
In a recent Self-Insurance Podcast episode, CRMBC CEO Kaya Stanley spoke with Harsha Vashisht, co-founder and CTO of Dyssonance AI, about how operators can use AI to spot patterns, ask better questions, and make better use of information they already have.
Why this conversation matters for restaurant operators
Restaurant operators already have more information than they can easily use. Sales reports, reviews, staffing schedules, customer complaints, menu performance, safety issues, and claims activity all tell part of the story. The challenge is connecting those signals while the business is moving.
Harsha keeps the conversation grounded in operating questions. Are lower reviews tied to a busy night, a shift pattern, a menu item, or a service issue? Are complaints repeating in a way that the manager has not had time to see?
AI can help with that first pass. It can sort comments, group themes, and point to possible connections. The operator still decides whether the pattern is real, but the work starts with evidence rather than a blank page.
“It’s there to help you. It’s there to make your life better. It’s a tool that’s at your disposal.”
— Harsha Vashisht
He also emphasizes that better answers start with better context: “Give the context, tell them who you are, what type of business you’re trying to run, what is the problem that you’re trying to solve.”
Use AI to look for patterns in reviews and service
Harsha’s most operator-friendly example is reviews.
If reviews drop on certain days or during certain shifts, AI may be able to help sort through the feedback and look for patterns. Are complaints tied to long waits? Missing items? A specific day of the week? A rush period? A staffing issue? A menu item?
That does not mean AI knows your restaurant better than you do. It means AI can help organize the evidence so you can see the issue more clearly.
Harsha explains that an operator could ask AI to compare review patterns across Google Maps, Yelp, and other platforms, then flag when something changes. He also describes using AI to watch nearby competitors and summarize what is happening in the local market.
A five-minute review check operators can try
This does not have to start with POS integration or a complicated dashboard.
Start with public reviews. Copy 30 to 50 recent Google or Yelp reviews into a document. Remove names or anything you do not want in the tool. Then paste the reviews into ChatGPT, Claude, or Gemini and ask:
What are the top three recurring service complaints in these reviews? Do any complaints appear to cluster around specific days, times, menu items, or service moments? Summarize the patterns and list what a manager should check next.
From there, compare the output against what you already know: staffing schedules, rush periods, ticket times, manager notes, or recent menu changes.
That small exercise makes the concept concrete. AI does not need to run the restaurant to be useful. It can give a manager a faster first read on information that would otherwise take time to sort manually.
One caution: shift-level patterns should start a conversation, not end one. If complaints appear to cluster around a particular shift, use that as a reason to look more closely, review conditions, and talk with the team before drawing conclusions.
Think of AI as an extra set of eyes
Harsha also talks about AI agents, which are different from a one-time question. Instead of asking the tool once, an operator can set up a recurring review: watch new reviews, check competitor activity, or summarize changes on a regular schedule.
For an operator, that might mean monitoring Google and Yelp reviews, competitor activity, menu changes, and local market trends. It could mean receiving a daily or weekly summary instead of manually checking every source.
That kind of monitoring is also where AI starts to connect back to risk. A pattern in reviews may point to more than customer frustration. It may show where the team is stretched, where service breaks down, or where training and safety issues need attention before they become larger problems.
Be thoughtful with business data
The episode also covers a question many operators will ask: What about security?
If you plan to use AI with POS exports, customer information, employee information, financial data, or other business records, take time to understand the tool’s privacy settings. Avoid uploading sensitive information unless you know how it will be used, stored, and protected.
The safer starting point is information that is already public or low-risk: public reviews, sample menu language, promotion ideas, or general scheduling questions. More sensitive records, including customer, employee, financial, or POS data, should only be used after the operator understands the tool’s privacy settings and internal policies.
Operators should also expect some manual setup. Reviews, POS exports, staffing notes, and manager observations may live in different systems. AI works best when the information is gathered clearly, stripped of unnecessary sensitive details, and organized around a specific question.
What operators should take from the conversation
The most useful entry point is a contained problem: a month of reviews, a recurring complaint, a promotion idea, or a staffing question.
Give the tool enough context to sort the information. Then compare the output against what you know from the floor.
For restaurant operators, that is the practical test. AI is worth using when it helps you see the business more clearly, ask better questions, or spot issues earlier.
FAQs
Do restaurant operators need technical skills to use AI?
No. A practical starting point is to describe your restaurant, explain the problem, and ask the AI how it can help. The better the context, the more useful the answer.
What kind of restaurant problems can AI help with?
AI may help operators look for patterns in reviews, customer complaints, POS exports, promotions, menu performance, staffing notes, and competitor activity.
Should operators upload POS data into AI tools?
Not without understanding the tool’s privacy and data settings. Start with low-risk information first, and avoid uploading sensitive customer, employee, or financial data unless you know how it will be handled.
Watch or listen to the full episode
For operators who want a clearer starting point, the full conversation is worth hearing. Harsha explains how he thinks about AI, what he learned from Uber Eats, and how restaurant operators can begin using these tools without treating AI as a separate technical project.
Watch or listen for the full examples on customer service, review patterns, prompts, business data, and restaurant marketing.
Learn more about CRMBC
CRMBC helps California restaurant operators take more control over workers’ compensation through a member-governed self-insured group model. To learn more, contact CRMBC.
Kaya Stanley is an attorney, published author, business owner, and highly sought-after strategic turnaround expert. Ms. Stanley serves as CEO and Chairman of the Board for CRMBC, the largest restaurant workers’ compensation self-insured group in California, and she is the Licensee for TEDxReno, an independently organized TEDx Event.
Throughout her 22 years of practicing law, Ms. Stanley has served as outside counsel for Wal-Mart and Home Depot. She was voted one of the country’s “Top 25 OZ Attorneys” by Opportunity Zone Magazine and published a best-selling book called “The Employer’s Guide to Obamacare.” Before that, she earned her master’s degree in social work and public policy, after which she worked with at-risk girls in Detroit and lobbied for women and families.
