Reslink

Speech-to-Form AI Integration

"This approach allowed us to validate the value of AI quickly without committing to a long and costly project."

Challenge

Reslink is a Finnish company with a growing international footprint for facilities management. Reslink had already achieved significant success with their SaaS platform used by field workers in security, cleaning and other facilities management operations. While their digital forms had delivered productivity improvements over paper-based processes, the company recognized an opportunity to further enhance user experience through AI-powered voice interfaces.

The primary challenge was reducing administrative overhead for field workers who still faced friction when completing forms on mobile devices. Workers needed a hands-free solution that would allow them to complete forms while continuing their duties. However, Reslink didn’t want to commit to a lengthy six-month AI project with uncertain outcomes and needed to validate the technical feasibility of voice-powered form completion before committing to full development.

Solution

Metosin delivered a collaborative one-day workshop that simultaneously built a working prototype and developed internal AI capabilities for the Reslink team. Rather than creating a solution for Reslink, the objective was to upskill their team to become independent with AI development and implementation.

The technical approach leveraged proven AI services to create an elegant solution. OpenAI's Whisper API provided speech recognition that could accurately handle accented English and technical terminology common in field operations. GPT-4o-mini then processed the transcribed text and extracted relevant information into JSON format matching Reslink's existing form schemas. The mini model was chosen for its cost-effectiveness, making it perfect for structured data extraction tasks.

The solution integrated seamlessly with Reslink's existing infrastructure. Workers could record voice notes on the mobile app, which would upload audio to the backend for processing through the Whisper API. The text transcription fed into GPT-4o-mini with form-specific prompts, and the extracted data populated form fields for worker review before submission to the existing MySQL database.

The workshop structure maximized both learning and development. The morning focused on problem definition and technical assessment, including an introduction to LLM concepts and review of a pre-prepared proof-of-concept. The afternoon involved hands-on collaborative development where the Reslink team actively participated in building the speech-to-JSON conversion pipeline and integrating with their mobile app. The late afternoon was dedicated to testing with real form scenarios and refining prompts for better accuracy.

Results

The workshop delivered immediate and strategic value for Reslink. By 6 PM on the workshop day, they had a working prototype that demonstrated speech-to-form functionality seamlessly integrated with their existing digital forms platform. This created a compelling demonstration capability for potential customers while building on the productivity gains already achieved by their digital forms.

The enhanced user experience allows workers to complete forms by voice while continuing light duties, representing the next logical evolution of Reslink Platform. More importantly, the Reslink team gained hands-on experience with AI implementation, understanding both the technical integration process and how to iterate and improve the solution independently.

The solution established a scalable foundation that works with Reslink's existing XML form definitions, meaning it can potentially extend to any form type across their platform. This approach focused development effort on business logic and integration while leveraging mature APIs for core AI functionality, demonstrating that AI doesn't require massive investments or lengthy development cycles.

Reslink now has both a working prototype and the internal capabilities to move forward independently. They can polish the prototype for production deployment, pilot with willing customers to validate real-world performance, and eventually roll out across all form types based on measured impact.

The key difference from typical AI projects is that Reslink now possesses the internal capabilities to execute this roadmap, iterate based on customer feedback, and apply AI thinking to future product development challenges.

Customer testimonial

"Metosin’s workshop exceeded our expectations. In just one day, we not only had a working prototype of speech-to-form AI, but our own team also gained the knowledge and confidence to continue developing it independently. This approach allowed us to validate the value of AI quickly without committing to a long and costly project. It was truly a pleasure working with the Metosin team."

Elisa Oranen
CEO, Reslink