Can AI help small businesses keep up?

Can AI help small businesses keep up?
Small business owners can connect with customers through AI platforms.—PHOTO FROM CANVA

The locally developed platform promises quicker customer replies without replacing human judgment. Its early deployments show a possible use for AI (artificial intelligence), but costs, accuracy and measurable results remain harder to establish.

A customer messages a business on Facebook about prices, schedules, or enrollment. The question is ordinary. Hundreds of other customers may be asking the same thing while employees have other work to finish. An unanswered message can mean a missed sale. For an employee, it can mean another evening spent repeating information.

Lesmon Saluta, founder of StrataDesk AI, believes this is a problem software can help solve. Developed by Philippine software company Bscale Laboratories Inc., the platform is designed to answer routine inquiries, guide customers through transactions, and forward complicated concerns to human staff.

The proposition is straightforward: Respond faster without asking machines to resolve everything. Whether smaller enterprises can afford it still requires evidence beyond the developer’s account.

From an overflowing inbox

Saluta said the idea emerged from Bscale Labs’ experience building software for local companies. As businesses drew more attention online, their Facebook and Messenger conversations expanded faster than their support teams.

Customers repeatedly asked about services, prices, schedules and registration. Delays could cost prospective buyers. The challenge is familiar to anyone who has messaged a business page, received an automated acknowledgment, and then waited hours for an actual answer. Speed matters, but so does receiving the information requested.

“We built StrataDesk AI so a business could give fast, accurate answers without hiring more and more people just to keep up,” Saluta told CoverStory in an interview. He said the team developed the system with clients, concentrating on what the assistant could say and when people should intervene.

“The hardest part wasn’t getting an AI to talk,” he said. “It was getting it to answer only from what the business has approved, and to know when to stop and pass the conversation to a person.”

Lesmon Saluta, cofounder and CEO of Bscale Labs, the company behind StrataDesk AI—PHOTO FROM LESMON SALUTA

What the customer sees

Consider Saluta’s example of a prospective student messaging a school’s Facebook page about a course. According to the founder, StrataDesk replies through Messenger using details from a business-approved knowledge base: program descriptions, policies, enrollment instructions and relevant links.

Unlike fixed menus, the assistant is intended to interpret ordinary language, including Taglish, explain procedures and guide bookings. Account problems, sensitive issues and discretionary decisions should go to employees.

Knowing when to stop may matter more than always producing an answer. Approved material does not guarantee accuracy: Outdated or misunderstood information can still produce wrong replies. A customer asking for a refund, for example, may need an explanation of the written policy, but also a staff member empowered to examine the circumstances. 

A test in education

One of the platform’s clearest publicly described applications is Funnel Mastery Academy, an education business whose online marketing brings heavy volumes of inquiries. Saluta said its community had more than 160,000 members and received thousands of messages weekly, exceeding 10,000 during advertising campaigns. 

Bscale Labs’ case study describes StrataDesk answering questions on Facebook and Messenger, while staff handle course-access difficulties and unusual requests. Questions may range from certification requirements to whether a beginner can manage the coursework.

The company also reports building three separate payment portals that support GCash and QR Ph transactions, confirmation emails and access to courses after payment.

Bscale reports that response times improved from hours to near-immediate answers. In a testimonial reproduced on the developer’s website, Funnel Mastery Academy COO Charles Andrew said the organization wanted to expand without damaging support quality. Employees may spend less time repeating enrollment answers and more time resolving difficulties. 

A second Bscale Labs case study describes an implementation for Batangas Country Club covering guest inquiries, booking guidance, social channels and a website. The project remains in development.

Accessible to whom?

The potential Philippine market is large, but it is hardly uniform. The Department of Trade and Industry’s 2024 MSME statistics show that micro, small and medium enterprises account for 99.63% of the country’s 1,241,476 registered business establishments. Microenterprises alone make up more than 90%.

These businesses differ greatly in workloads and software budgets. For one shop, a simple FAQ and a reliable employee may be enough. For another, high-volume campaigns might justify an automated first response. Market size alone says nothing about whether a particular product delivers value.

Saluta said Bscale helps clients set up StrataDesk, assemble an approved knowledge base from existing business information, and connect the system to their current pages. The most difficult part, he suggested, may come before automation begins.

“The most common challenge isn’t the technology,” he said. “It’s getting a business’ information written down clearly in one place, since much of it usually lives in people’s heads.”

There is also the question of cost. Saluta said interested businesses should contact the company for a quotation suited to their needs.

Customization may help, but undisclosed prices prevent easy comparisons. Without documented savings, affordability for microenterprises remains unproven. Owners would need to calculate subscription and setup costs against time saved, missed inquiries recovered, and ongoing work required to maintain the knowledge base. The cheapest system is not necessarily the one with the lowest advertised price.

Whose data, whose responsibility?

Customer conversations may include names, contact details, transaction references and complaints. Managing them requires more than polite replies.

The National Privacy Commission’s 2024 advisory on AI and personal data makes clear that existing data-protection duties apply to AI systems processing personal information. These include transparency, lawful processing, appropriate safeguards and accountability, including when processing is outsourced. A customer should be able to understand what is being collected and why, not discover the arrangement only after something goes wrong.

Saluta said business clients control what StrataDesk may do, can review conversations and may intervene at any point. He also said customer information is treated as the business’ data and used to serve its customers.

Those assurances need documented controls. Bscale Labs’ corporate website privacy policy expressly covers that website only. Its company-wide list of service providers identifies OpenAI and OpenRouter for AI processing, alongside Meta and various hosting, storage and monitoring providers, some operating abroad. 

Work that remains human

Automation also raises a less technical concern: What happens to employees once routine customer support can be handled by software?

Saluta argued that StrataDesk removes repetition, not the need for workers. He said teams can focus on problem-solving, customer relationships and sales while the assistant handles standard opening questions. That is a plausible division of labor, especially for businesses struggling with message volume.

For workers, the most meaningful test is whether saved time produces manageable workloads, better skills and decent opportunities, rather than simply higher expectations for fewer people. Human handoff must also mean a real employee can respond promptly, not that a difficult request disappears into another queue.

Beyond the instant reply

Imagine a small business able to answer basic questions after closing time, while its employees return the next morning to conversations that genuinely require attention. AI could make that possible without requiring every shop to operate a round-the-clock call center.

The alternative is automation that misleads customers, mishandles data, or obstructs access to people. The outcome depends on system design, oversight and accountability. For Saluta, success means businesses growing without overwhelming their teams and customers receiving help while they still want it. The measure is reasonable, but it remains to be demonstrated consistently.

StrataDesk is an example of a useful direction for Philippine-made software: solving repetitive local business problems rather than treating AI as a spectacle. Its longer-term credibility, however, will rest on verified outcomes, transparent costs, workable privacy protections and a human handoff that actually works. A fast answer is valuable. A trustworthy one is harder to achieve, and matters more. CS

Hannah L. Tabunda is a journalism graduate of the University of the Philippines Diliman and a contributor to CoverStory. Her interests include feature writing, multimedia journalism, and “stories that explore people, culture, and lived experiences.” She also engages in such combat sports as karate, Muay Thai and boxing, and has taken up bouldering and top-rope climbing.