Gudea Wants to Predict Social Media Crises Before They Erupt
Gudea pitched on the Startup Battlefield stage at TechCrunch Disrupt 2026 with a system it calls the "storm tracker for the internet." The company uses AI to help organizations spot trouble brewing online before it turns into a full social media crisis. TechCrunch
The core is narrative tracking at network scale. That means following how stories move across connected accounts, not only counting mentions. Gudea's models analyze behavior, relationships, and pathways of information travel across more than 500 mainstream and fringe websites and platforms, including X, Reddit, and TikTok. The emphasis is not on keyword volume alone. It is on how a story moves, who amplifies it, and where it gains velocity.
Gudea was founded in 2023 by Keith Presley, Jonathan Sperber, and James Thomin. Presley serves as CEO and previously built infrastructure for Andrew Yang's presidential campaign. Sperber is chief strategy officer and brings a background in disinformation research, the study of false information spread deliberately. Thomin is CTO, holds a Ph.D. in chemical engineering, and has spent roughly two decades working in AI and machine learning.
Three years after founding, the company reports nearly 40 clients. Its public site at gudea.ai frames the offer as rapid detection of emerging narratives paired with expert interpretation of complex signals. The tagline used there is "Foresight for Emerging Narratives." The company also closed an oversubscribed $7 million seed round, meaning investors offered more money than was available, according to an Oct. 10 announcement. PR Newswire
Rather than ranking by follower count, Gudea sorts social media accounts into five behavioral archetypes, or groups defined by posting patterns and network behavior. Gudea The stated goal is to detect and forecast online narratives, not simply to archive mentions after the fact. Traditional social listening tools aggregate posts and score sentiment, which is a measure of tone. They do less early trajectory inference, which is prediction of which threads will grow. The open questions are which fringe thread will stay contained, which will bridge to mainstream feeds, and which actors routinely serve as bridges. Gudea's approach is that graph structure, the map of who links to whom, plus behavioral sequencing, the order of actions over time, provides earlier signal than volume spikes.
For teams that already run a SOC, a security operations center that watches for threats, or a trust and safety function, the team that handles abuse and harmful content, the pitch is to shift detection left, to catch issues earlier in their life cycle. The system would replace manual Boolean queries, hand-built keyword rules using AND and OR, and dashboard watching with a system that surfaces a small number of high-risk narrative paths for human review. Finding many candidates, called recall, is easy. Being correct, called precision, is hard. Any system that monitors 500-plus sources will generate noise unless its ranking layer is disciplined.
In my view, the credibility test will not be coverage breadth. It will be false positive rate under operational load, how often the system raises false alarms during normal work. A crisis prevention tool that pages a communications lead at 2 a.m. needs calibrated confidence and auditable provenance, a clear record of who said what first, where it spread, and what changed in the diffusion pattern. A point worth noting is the staffing mix. A CEO with campaign infrastructure experience, a strategist trained in disinformation research, and a CTO with long ML tenure pairs practical knowledge of adversarial information flows with engineering for production models.
The broader context here stretches across several online eras. I raised two kids through the successive waves of forums, social feeds, and short video. Each wave made rumor transmission faster and harder for outsiders to trace. Tools that map that transmission do not solve misinformation. They do give institutions a better chance to respond with facts while a narrative is still forming. Over time, those gains compound into more resilient operations.
