Fifteen app iterations, patent-pending UI, voice activation. Why Billie Onsite CEO Annie Liang obsesses over one-handed mobile operation.
What’s happening: Annie Liang’s construction tech startup tested at 300+ building sites and iterated 15 times to create a voice-AI system that works onsite. The former UBS banker developed patent-pending features like one-handed operation after discovering why most construction apps fail in real-world conditions.
Why this matters: Construction is a $10 trillion industry where most tech startups flop despite huge market potential. Understanding how to build tools that traditional sectors actually adopt could unlock massive opportunities for companies willing to test rigorously instead of guessing user needs.
The construction site buzzed with activity, but Annie Liang wasn’t watching the cranes or concrete pours. She was observing how workers interacted with their mobile devices, noting when they struggled with two-handed operation while wearing safety gloves, and documenting the precise moments when digital tools became obstacles rather than aids.
This scene played out across more than 300 construction sites as Liang and her team at Billie Onsite conducted what might be the most extensive usability testing programme in construction technology. The result: a voice-activated AI system that workers can operate with one hand while navigating demanding onsite conditions.
“The problem isn’t a lack of tech familiarity,” explains Liang, who co-founded Billie Onsite in 2023 after her finance career and research into prefabrication systems in Auckland. “Many users handle advanced apps on personal devices well. Rather, business software has often been designed without sufficient consideration for practical hurdles in collaborations from field to office.”
Testing reality
The construction industry presents unique challenges for technology adoption that many startups underestimate. Unlike software or services sectors where digital transformation has been embraced, construction involves physical activities that aren’t standardised and fragmented groups of stakeholders with their own methods and schedules.
“A construction project brings together architects, contractors, subcontractors, suppliers, inspectors, regulators, and clients, each with their own methods and schedules,” Liang notes. “Variables like weather and site-specific conditions add further complexity, preventing easy uniformity in processes.”
This fragmentation creates a software integration nightmare. Tools often don’t align with varied workflows, resulting in chronically low adoption rates. Most solutions, according to Liang’s observations, “solve the office need for data organisation well, but don’t provide practicality for onsite use.”
The extensive testing programme revealed something crucial: onsite adoption requires more than intuitive design. It demands interfaces that work within the physical constraints and safety requirements of construction environments. This insight led to Billie Onsite’s patent-pending UI features, including one-handed mobile operation and voice activation that functions even with background noise and safety equipment.
Voice meets construction
Rather than creating another project management tool, Liang’s team focused on what she calls “high-value, critical outcomes that hinder daily operations.” The voice-visual AI system targets specific pain points that consume significant time and cost, such as stakeholder reporting, communication gaps, and undetected problems that lead to delays and budget overruns.
“AI in these sectors can’t simply enhance existing workflows; it must target high-value, critical outcomes,” Liang emphasises. The system instantly categorises and summarises voice and visual data, automates report generation, and verifies inspection findings against codes and standards through a centralised dashboard.
This approach emerged from Liang’s firsthand experience with performance-based certification challenges and quality control issues during her prefabrication research. “To resonate among our target audience, we focused our key deliverable on certain forms of expert reports that focused heavily on verifying onsite issues against building codes and project documents.”
Beyond workflow tweaks
The construction technology graveyard is littered with startups that built elegant solutions for problems that didn’t matter enough to pay for. Liang learned this lesson during her transition from financial services to construction technology, where she discovered that flashy features mean nothing without measurable business impact.
“A key metric for gaining buy-in is showing low effort with high return, which influences decisions on investing in the software,” she explains. “Users and companies need evidence that the tool requires minimal training or disruption while delivering substantial savings or efficiency gains.”
Billie Onsite’s approach focuses on quantifiable outcomes: cost savings from desktop work automation, reduced delays from better communication, and faster compliance verification. These metrics matter more than user experience scores or feature lists when convincing construction companies to invest in new technology.
Different users, different needs
Even within single companies, user groups have distinct needs that must be addressed differently. Liang’s research revealed a fundamental split between bottom-up and top-down users of construction technology.
“Site teams reporting progress bottom-up prioritise quick data capture for logging achievements and requesting information,” she explains, “while in contrast, project engineers going to site have a top-down focus on verifying quality, auditing compliance, and aligning with overall goals.”
This insight shaped Billie Onsite’s development significantly. The same voice-AI system serves both user types but presents information and functionality differently based on their primary objectives. Site teams get streamlined data capture tools, while project engineers receive comprehensive verification and audit features.
The lesson extends beyond construction to any traditional industry where technology adoption has been slow. Understanding not just what users do, but why they do it and what success looks like from their perspective, becomes crucial for building solutions that gain widespread adoption.
Liang’s experience suggests that breaking through in traditional sectors requires more than innovative technology. It demands rigorous field research, iterative design based on real-world constraints, and a willingness to solve unglamorous but high-value problems that existing players have overlooked.
“Building a tech startup in slow-moving sectors requires direct engagement with challenges, rigorous testing, and a focus on targeted value,” Liang concludes. “By addressing varied user needs and proving clear returns, it’s possible to create solutions that gain traction and make a lasting impact in these areas.”
Keep up to date with our stories on LinkedIn, Twitter, Facebook and Instagram.