A Chilean construction safety startup born at Universidad Adolfo Ibáñez is building an AI-powered risk prevention platform that insists on manual oversight, standardized data, and patient execution over rapid automation. The approach offers a window into the operational and talent realities employers face when hiring engineering talent in Chile for AI-driven workplace safety projects.
Neos AI, co-founded by Sebastián Flores, Benjamín Salas, and Santiago Sfeir, now serves construction firms with camera-based hazard detection, document management, incident tracking, and a proprietary risk probability index. The platform reflects a deliberate sequencing: digitize, standardize, then automate.
How Neos AI pivoted from security cameras to construction safety
Neos did not begin as a construction technology company. The three co-founders, who had worked together for eight or nine years through university and a master's program, presented an image recognition solution for security cameras at Prototypes, an event hosted by Universidad Adolfo Ibáñez. Entrepreneur Maria Paz Gillet heard the pitch and suggested adapting the technology for construction, then connected the team with a builder whose primary concern was accidents.
The founders built what is now a multi-module platform. CanarIA Vision monitors live camera feeds to detect missing personal protective equipment, unsafe distances between workers and machinery, and restricted zones. A document manager handles inductions, checklists, and risk analyses. An incident module records preliminary alerts, witness statements, and root-cause analysis. Each incorporates what Flores describes as small layers of AI.
Need help hiring?
See the next step after this guide
If this topic is relevant to your team, these are the most useful pages to check next.
Flores acknowledges the team knew nothing about risk prevention at the outset. Today, clients can customize the platform to match their operational reality, a flexibility Flores says competitors offering generic forms and free-text fields do not provide. The end users are not executives but site supervisors, safety officers, foremen, and workers. The team spent time on-site to understand their workflows.
Why construction's data chaos creates opportunity for AI-driven risk prevention
Chilean construction remains largely paper-based, and digitalization efforts often amount to unstructured text fields that generate no actionable intelligence, according to Flores. The industry sits on valuable data that goes unused because it is neither standardized nor governed.
Neos applies a strict sequence: digitize processes correctly, manage organizational change, order and standardize data, and only then deploy AI. Flores states he would have preferred entering a market already prepared for AI. Even when AI functions well, he insists it cannot operate alone in safety-critical contexts.
The platform enforces human verification. For critical documents, AI may handle 80 percent of the work, but a person must validate the output. Flores frames this as both a safety imperative and a labor-saving measure: automation reduces administrative burden, but final review remains manual.
This design philosophy has direct implications for workforce planning. Employers building or adopting similar tools need AI engineers building safety solutions who understand the limits of automation in high-stakes environments. They also need change management capacity to help field teams adopt new workflows, and data governance expertise to ensure AI models train on clean, structured inputs.
Flores argues that companies serious about safety embed it in their culture, not compliance paperwork. For those motivated only by regulation, he offers an economic case: a fatal accident can halt an entire project, making prevention far cheaper than remediation. Neos positions its platform as a tool to reduce incident probability, maintain operational continuity, and protect workers.
What Chilean startups need to scale safety tech across LATAM
When asked what must happen for safety technologies to become standard across Latin America, Flores points first to data governance. He wants competitors to adopt it as well, because transitions from rival platforms to Neos often involve messy, unstructured data. Only after governance is in place can companies optimize processes, execute digital transformation, and apply AI effectively. Without that foundation, he says, the technology runs like a car on a dirt road full of potholes instead of a highway.
The startup nearly folded. Flores and his co-founders reached a point where they considered quitting, uncertain whether they were delivering value to the industry or just one client. What kept them going was persistence, a value they half-jokingly attribute to anime and Marvel fandom. Flores admits he struggles to disconnect: ideas for clients surface while he watches series at the gym. His one rule is no work on Saturdays; he sleeps late, cycles, and resumes Sunday at five in the afternoon.
For the coming year, Neos aims to add clients and build brand recognition. Anonymity has been costly, Flores concedes, and competitors have begun developing similar technology. He wishes he had known the construction sector better before starting, particularly to build a contact base earlier.
The startup's trajectory illustrates the operational and talent challenges employers confront when deploying AI in traditional industries across Latin America. Success requires technical skill, domain expertise, cultural fit with end users, patience with data infrastructure, and the capacity to persist through uncertainty. Flores has not built the Iron Man suit he dreamed of as a child, but he has constructed a different kind of protection: one that, with order and without rushing AI, aims to get workers home safely.