NxCode Raises Funding as AI Code Outpaces Verification
NxCode has raised a seven-figure U.S.-dollar investment from GSR Ventures, according to information supplied by the company. The round arrives at an unusual moment: generating code has never been easier, and trusting it may be getting harder.
Sonar's 2026 State of Code Developer Survey, based on more than 1,100 professional developers, found that AI now accounts for an estimated 42 percent of committed code. Yet 96 percent said they do not fully trust AI-generated code, only 48 percent always verify it before committing, and 38 percent said reviewing AI output takes more effort than reviewing a colleague's work.
The 2025 Stack Overflow Developer Survey showed the same tension: more respondents distrusted AI accuracy than trusted it, and 66 percent named solutions that were "almost right, but not quite" as their biggest frustration. Obvious errors stop a program. Plausible ones travel much farther.
The gap shows up between demo and production. A generated interface can look finished while authentication fails at account recovery, a database has no safe migration path, a payment integration mishandles an edge case, or an agent exposes information the user did not realize had entered its context. Non-technical users - the people an AI app builder most empowers - are least equipped to spot them.
NxCode sits directly on that line. Its platform uses specialized agents to handle planning, architecture, coding, testing and deployment, so that a founder can describe an application in ordinary language and move toward something deployable. But every layer of automation is also a decision about responsibility: what the platform verifies automatically, which risks it explains, when it should refuse to proceed, and how a user can inspect, export or roll back what the agents produced.
Developers draw the line at exactly that point. In the Stack Overflow data, 76 percent said they did not plan to use AI for deployment and monitoring - the stages where a suggestion becomes an operational fact and someone becomes accountable for the outcome.
Zirong Chi, who founded NxCode and leads its product design, development and operations, has worked on that problem before. Her earlier product, Bibabo, taught AI and coding through debugging exercises rather than presenting model output as something to be accepted. Access, in that view, means enough understanding to question a result.
The funding raises the standard rather than settling it. The measures that matter are unglamorous: how many projects deploy successfully, how many are still running months later, how often users need human help, and whether customers can take their code elsewhere. Speed in AI development is becoming widely available. Confidence is not.
About NxCode
NxCode is an AI development platform that turns plain-English descriptions into working software, with agents handling planning, coding, testing and deployment. Founded by Zirong (Vivian) Chi, it was selected for MiraclePlus's F25 program and is used by 5,000+ non-technical builders. Mountain View, California, USA. https://www.nxcode.io