On Friday, October 30, Ciudad Cultural Konex moved from the main stage to the worktable. The day brought invitation-only sandboxes, two rounds of lightning talks and an afternoon of workshops. Here is what it left behind.
Day 1 of Media Party Buenos Aires 2026 belonged to the international keynotes. In contrast, day 2 belonged to hands-on experience. Digital native outlets, civil society groups, developers and entrepreneurs from Latin America shared what they are building with AI. They also explained what worked and where they hit limits.
The theme of this edition is “Storytelling in the age of agents.” On Friday, it turned into one idea that came up in almost every room: AI can expand journalism, but judgment, data and community ties remain human. That is also the core of the cognitive sovereignty defended in this year’s manifesto.
Sandboxes: four tribes working side by side
The day opened at 10:00 a.m. with a new format: the Media Party sandboxes. These were small working groups, open by invitation or application. Each one gathered a “tribe” of professionals around one topic:
- Media Party Investment Club: Saša Vučinić (North Base Media) and Mariano Blejman led this group on media financing and investment. It was designed for projects looking for venture capital.
- Laia & AI Tinkerers: David Coronel ran a hands-on lab to prototype AI tools for journalism. In 2025, he presented Calíope, a LAIA writing tool.
- Creator Economy: Ernesto Martelli led the discussion on business models, monetization and the creator economy.
- Policy and Regulation: Fundar backed this group, with Agustina Brizio as coordinator. It focused on digital governance, AI regulation and public policy.
Midday lightning talks: audiences, sustainability and trust
At 12:00 p.m., the Gran Sala hosted the first round of lightning talks. The talks were short and direct, and each one started from a real case.
Audiences: journalism tailored to each user
Working from exile, elTOQUE presented Sami. Tayli Sánchez explained that the app personalizes topics, formats and notifications. It draws on the outlet’s verified archive and on more than 1,400 legal answers about life in Cuba. In short, it is a new way to tell the story, with the same rigor as always.
Sustainability: products that also bring in revenue
María Eugenia “Maru” Álvarez, executive director of SembraMedia, looked at outlets that are already making money from AI products. For example, she cited Chequeado’s Desgrabador and Aos Fatos’ Escriba. However, she warned that selling the tool is not enough. First, outlets need to know which business they want to strengthen. SembraMedia has supported digital media in the region for a decade, as Janine Warner explained in her talk on the successes and failures of innovative media.
Next, Sebastián Auyanet, from Factchequeado, showed GranTIA. The tool matches Spanish-language outlets in the United States with funding opportunities. Building the prototype was fast. Getting outlets to use it, however, was the hard part. That is why his advice was simple: measure use, not just delivery.
Meanwhile, Martín Pozzo brought a story from the outskirts of Buenos Aires. Radio Cromo Escobar launched in 2022 as the area’s first digital native station. Since then, it has built its own ecosystem: streaming “with content,” the training platform RCE Educación and the publishing house RCE Ediciones, now on its fourth book.
Trust: open data against disinformation
Juan Pablo Ruiz Nicolini, from Fundar, explained how digital public infrastructure can defuse election fraud narratives. Today, results by polling station go online 48 hours after each election. Similarly, Andrés Vázquez, from the Open Knowledge Foundation, presented pilots with Uruguay and Brazil. Thanks to the MCP protocol, AI can query open data and return exact calculations. The Media Party Advisory Board also discussed this topic.
Adrián Pino and Soledad Arreguez, from Proyecto Desconfío, shared a compass to classify AI-generated disinformation. They built it after monitoring nine Latin American elections in 2025 and 2026. One finding stands out: 92.9% of the pieces they studied have no traceable source. In addition, most of them are video deepfakes shorter than 15 seconds. Armin Senoner, from Google DeepMind, also explored this issue in his talk on tracking synthetic content.
People: get organized before you speed up
Maximiliano Ferreyra challenged a common pattern: teams buy AI and keep working the same way. Instead, he suggested starting from real problems. Finally, Rosario Tapia closed the round with insights from neuroscience. In her view, automating broken processes does not fix them, and the same is true for people. Her idea echoes Leonardo Dalera’s talk on how not to implement AI projects in media.
Read the full story on the first round
Workshops: an afternoon of hands-on work
From 1:30 to 6:30 p.m., the Media Party workshops filled the galleries and rooms of the Konex. Here, the question changed. Instead of asking what AI can do, participants asked what to delegate and where a person needs to step in. Friday’s sessions included:
- The cognitive cost of working with AI: Yanina Pantiga explored when to delegate and how to supervise without burning out.
- Agents in collaboration: the Normsy team, including Juan Marten and Juan Manuel Ortiz de Zárate, showed how to design multi-agent workflows with human approval.
- The day the machines stopped talking: Camila Achuri Rojas, from Datasketch, looked at the skills we lose when we hand work to machines.
- Hack your narrative: Lucila Pelletieri, from Google News Initiative, showed how to build immersive stories with Gemini Notebook and Canvas.
- The techno-political experience of the Police Map: Sebastián Alejandro Sciutti shared how his team maps and reconstructs events with open tools.
- Build your professional portfolio with AI and take it home: from 5:00 to 6:30 p.m., Melisa Avolio, Irina Sternik and Fedra Abagianos covered digital hygiene and SEO. Participants then worked toward a finished portfolio in PDF and web formats.
Closing lightning talks: what AI cannot replace
At 6:00 p.m., the second round of lightning talks returned to the Gran Sala. This time, the stage brought together cross-border investigations, century-old archives and civic games.
Collaboration to investigate power
Natalia Viana, from Agência Pública, shared the story behind Big Tech’s Invisible Hands. Agência Pública and CLIP led the investigation, which brought together 17 outlets from 13 countries. Together, they published an open database of nearly 3,000 lobbying actions by tech giants in Latin America. Her message, that it takes a network to investigate a network, echoes the keynote by Marina Walker Guevara, from the Pulitzer Center.
Florencia Tuchin, from the InContext agency, presented a guide to design symbiotic relationships between journalism and civil society. She developed it with the Pulitzer Center and Universidad de los Andes. The premise is clear: collaborate without losing independence. Moreover, teams should measure impact by resonance, not clicks.
Human judgment: AI proposes, people decide
Camila Velasco, a columnist at Perfil, showed her social listening system. Every morning, a team of AI agents prepares a dashboard of verified topics for her. Still, her rule is clear: the agents listen, but she decides. Marco Túlio Pires, from Google DeepMind, shares this view. For him, no answer should arrive without its source.
Armando Ezequiel Puerta, from Fundar, faced a different problem: AI kept producing inconsistent charts. To fix it, his team built an in-house, open source charting library. Juan Alberto Farías, in turn, presented Perceptor, a self-hosted tool that organizes media libraries with agents and human oversight. Lastly, Agustina González Miguens, creator of Beyond The Source, defended hype-free tech communication. For her, quality comes before quantity.
Archives and audio: rescuing what already exists
Gonzalo Rotonda explained how El Eco de Tandil is digitizing an archive that spans more than 140 years. The project combines OCR with an AI layer that can read the writing of other eras.
Hernán Ordiales, from LAIA, proposed semantic search in audio. His reason is simple: transcripts lose the tone of voice, the applause and the booing. This is not his first time at Media Party, either. In 2025, he showed how to go from a prompt to a sound piece with visuals in 17 minutes.
Community: useful information and everyday democracy
From the outskirts of São Paulo, Anderson Meneses, from Agência Mural, presented Mural Local. The project turns public data into service journalism, and local ambassadors review every piece. It already reaches 11% of the outlet’s audience.
Finally, from Costa Rica, Daniel López Barboza presented Poli-Tico, a video game by Diario Mirador. Players govern a canton and, along the way, learn how local government works.
Read the full story on the second round
What day 2 leaves behind
Above all, Friday showed that media innovation in the region is not waiting for big platforms. Instead, teams are building it with open data, newsroom collaboration and careful use of AI. To keep exploring these topics, read more stories on AI in newsrooms or check the full Media Party Buenos Aires 2026 schedule.
On Saturday, October 31, the event closes with the Media Party hackathon. There, interdisciplinary teams will turn these ideas into working prototypes. Find out what the day will look like.
Friday’s ideas get put to the test this Saturday, October 31, at the hackathon at Konex (Sarmiento 3131), free of charge: register here and help build the future of media.
