Here are the four main objectives:

• Enhance efficiency:
  The goal was to reduce the time spent on operational and repetitive tasks, such as transcribing audio recordings and reviewing lengthy documents. This would allow journalists to free up time to focus on creating content with greater depth and editorial impact.
• Design tailored tools:
  The project aimed to develop custom intelligent assistants that addressed the newsroom’s specific needs. The idea was to design tools that could optimize key processes — such as interpreting government bulletins or police reports — while remaining aligned with the outlet’s editorial style and priorities.
• Expand coverage and personalize content:
  Another objective was to explore how AI could help diversify the range of topics covered and personalize the news offering for local audiences. This involved creating more accessible and relevant content, improving the user experience, and strengthening the media outlet’s connection with its community.
• Strengthen data analysis and decision-making capabilities:
  The project sought to leverage AI tools to improve data analysis and the interpretation of key metrics. This would provide the newsroom with the ability to identify audience trends, evaluate content impact in real time, and adjust editorial strategies quickly. With information efficiently processed and visualized, the team would be able to make more informed and strategic decisions.

With these goals in mind, we began working on the initial stages of the project, transforming early ideas into concrete actions that would lay the foundation for the full implementation process.

Implementation Phases

With the objectives clearly defined, the integration of artificial intelligence at 0221 was designed as a progressive and structured process, combining analysis, training, and customized development. This approach not only allowed for an orderly implementation but also ensured the adoption of useful tools tailored to the newsroom’s workflows.

The process began with an initial assessment, carried out through two parallel streams. On one hand, we worked strategically with senior management and chief editors to identify opportunities and establish priorities. This process helped define a comprehensive vision for how AI could contribute to improving operational efficiency, optimizing content production, and strengthening the outlet’s editorial offering. Discussions at this level focused on prioritizing the areas where AI would have the greatest impact and on establishing criteria for responsible implementation.

At the same time, and as a fundamental part of the process, we began jointly developing AI usage policies for the organization. These policies set clear guidelines to ensure the ethical and transparent application of the technology, always prioritizing human oversight. The regulatory framework included principles for validating AI-generated content, protecting data, and maintaining transparency in communicating with audiences about the use of these tools.

In parallel, we conducted an operational analysis within the newsroom. In this phase, we worked directly with journalists and editors, reviewing their daily workflows and identifying tasks that could potentially be automated. Based on this assessment, we launched a training phase to prepare the team for the practical use of AI tools. This phase included a series of workshops covering topics such as effective prompt creation and the design of custom GPTs, fostering both technological adoption among journalists and a critical, collaborative approach to AI integration.

To support this process, we created a safe environment where teams had the freedom to experiment, make mistakes, and test ideas without fear of failure. This controlled space was key in helping them identify not only the potential of artificial intelligence but, more importantly, its limitations. Beyond exploring its capabilities, the goal was to understand the technology’s boundaries and learn to recognize which tasks were not suitable for AI or did not add value.

The development of customized tools was a central part of this stage. Building on the knowledge gained during the training sessions and the analysis of their own workflows, newsroom teams created their own tailored tools. This participatory methodology ensured that the solutions were aligned with the specific needs of each area while also fostering a critical perspective within the team, leading to a more conscious and responsible adoption of AI in the newsroom.

Among the main tools developed were:

• News Summarizer: An application that generates automated summaries from long texts, providing quick access to key information.
• Writing Assistant: A custom GPT trained to suggest headlines, leads, and editorial angles, offering creative support to journalists.
• SEO Reviewer: A tool designed to optimize texts based on search criteria and digital trends, while respecting 0221’s editorial style.
• LectorIA: A tool for synthesizing information from the Official Gazette of the Province of Buenos Aires, streamlining access to key government data for news coverage.
• PolicIA: A model that interprets police reports, facilitating the quick identification of relevant information and the drafting of news stories.
• B-Note Generator: A solution for the automated creation of complementary content, especially useful for extended or ongoing coverage.
• Las Tres Cabritas: An application designed to organize and present sports statistics on players, coaches, and referees, making key data easily accessible for sports reporting.
• Hugo CM: A tool specialized in drafting text and graphics for social media, adapted to 0221’s style and tone.
• InfografIA: A system that extracts key data from journalistic texts to create visual content and simplify complex information into infographics for both social media and the website.
• BancarIA: A model designed to interpret and synthesize information about bank promotions and discounts, facilitating the creation of updated and relevant content for 0221 users.

As part of the process, we decided to introduce a playful component to encourage creativity and team engagement. We organized an internal competition to reward the best AI-based development created by the newsroom. To evaluate the projects, we developed a custom GPT called JudgeGPT, trained with various parameters to assign scores and justify its decisions. This dynamic not only strengthened the team’s innovative spirit but also provided a practical experience in applying AI to real-world problem-solving.

Throughout an entire day, teams fine-tuned the details of their projects, documented them, and presented them openly to the entire newsroom. The winning project was LectorIA, a tool developed by two of the newsroom’s most experienced members. Their deep knowledge of internal processes and long-standing experience were key to designing an effective solution tailored to their area of work — providing a clear example that generational factors are not a barrier when it comes to adopting new technologies.

The next phase with the newsroom involved testing and iterating on the newly developed tools. We piloted the tools in real-world scenarios before fully implementing them. These pilot tests allowed us to fine-tune functionalities, incorporate improvements, and ensure that the tools integrated seamlessly into the editorial workflows. Throughout this phase, we worked side by side with the journalists, which not only strengthened the relationship between the team and the technology but also built trust in AI’s potential to support their daily routines.

Throughout the entire process, human oversight remained a fundamental pillar. To ensure accuracy, truthfulness, and transparency, validation checkpoints were established, requiring that any information generated or interpreted by AI systems be reviewed by journalists before publication. This “humans in the loop” approach helped to consolidate an ethical and effective integration of technology into the newsroom.

Challenges and Approaches

Throughout the integration process, several challenges emerged that required agile and collaborative responses. Four, in particular, stood out as the most significant: resistance to change, the need for training and learning, technical limitations, and ethical and editorial control concerns. In each case, the team implemented concrete strategies that not only helped overcome these obstacles but also strengthened the adoption of the technology with the aim of fostering a culture of innovation within the organization.

Below is a description of these challenges, along with the strategies and actions that enabled them to be addressed effectively:

1. Resistance to Change
The introduction of new technologies often triggers uncertainty and fears about task automation. At 0221, some journalists expressed concerns about the potential loss of editorial control or even job reductions. There was also skepticism about AI’s ability to produce high-quality content that matched the outlet’s style.

Approach:
To address this resistance, we prioritized clear and continuous communication, emphasizing AI’s complementary role and the non-negotiable commitment to maintaining human oversight at all times. Practical workshops allowed journalists to experiment with the tools and understand their benefits firsthand. Additionally, a culture of experimentation was promoted, where mistakes were framed as learning opportunities, strengthening the team’s trust in the technology.

2. Training and Learning
The primary challenge was equipping the team with the skills necessary to use AI tools effectively. A lack of prior experience with the technology posed a barrier to adoption and limited the full potential of the solutions developed. Furthermore, it was essential not only to teach the technical use of AI but also to encourage a critical understanding of its possibilities and limitations.

Approach:
An intensive training phase was carried out, including personalized workshops where journalists learned to design effective prompts and create their own custom GPTs. The methodology was practical and participatory, allowing the team to apply the knowledge acquired in real-world scenarios. Additionally, a continuous mentoring space was established to answer questions and support the learning process. This strategy not only accelerated AI adoption but also fostered greater ownership of the tools among the team.

3. Technical Limitations
During the implementation, challenges emerged related to integrating AI tools with the newsroom’s existing systems. Differences in data formats and the need to adjust solutions to real-world workflows added further complexity.

Approach:
To address this challenge, a scalable development approach was adopted, involving pilot testing and continuous adjustments. The creation of customized tools such as LectorIA, PolicIA, and the B-Note Generators allowed the technology to be adapted to the outlet’s specific needs. Moreover, close collaboration between technical and editorial teams was essential for quickly identifying problems and applying effective solutions.

4. Ethics and Editorial Control
One of the main challenges was ensuring that AI implementation did not compromise the truthfulness or quality of information. The possibility of the technology producing inaccurate or biased content remained a constant concern.

Approach:
To mitigate these risks, responsible usage policies were established, alongside a “humans in the loop” approach that required human oversight at every stage of the process. All content generated or assisted by AI was reviewed and validated by journalists before publication. Additionally, training sessions on algorithmic biases and best verification practices were conducted, ensuring the ethical and transparent use of the technology.

Results and Impact

To assess the impact of artificial intelligence integration in the 0221 newsroom, a quantitative and qualitative survey was conducted, with 17 editorial team members participating. The survey measured perceptions regarding the use of the implemented tools, the benefits and challenges of the process, and suggestions for future improvements.

The data analysis revealed clear strengths, areas requiring adjustments, and new opportunities for development. Overall, the results reflected a positive evaluation of the experience, while also highlighting critical feedback that offered deeper insight into the current limitations of AI technologies in journalistic environments.

Acceptance and Productivity

Ninety percent of the team rated their experience with AI as either satisfactory or very satisfactory, citing ease of use and the reduction of repetitive tasks as the main benefits. Additionally, 80% reported a positive impact on their daily productivity, particularly due to the time saved from mechanical tasks like transcribing audio recordings or interpreting lengthy documents.

Automation of tasks and time optimization were seen as the most tangible improvements, mentioned by 70% and 65% of respondents, respectively. These results indicate that the tools successfully met one of the project’s primary objectives: enhancing the team’s operational efficiency.

Impact on Quality and Creativity

Regarding editorial quality, 70% of journalists noted that AI significantly or moderately improved the content produced, especially in areas such as SEO optimization and data verification. However, 10% reported a decline in quality, pointing to issues related to so-called “hallucinations” by AI models.

In terms of the creative process, 70% of respondents stated that AI enhanced their creativity or inspired new ideas. This suggests that the tools not only eased operational burdens but also served as a catalyst for more innovative content development.

Usability and Experience

Ninety percent of the team rated the tools as “easy” or “very easy” to use, emphasizing the intuitive design of the interfaces and the support received during the training phase. This positive perception greatly facilitated the integration of the tools into daily workflows.

Main Challenges

Despite the favorable outcomes, the process was not without obstacles. The most cited challenge was a lack of trust in AI outputs (mentioned by 50% of respondents), mainly due to transcription errors, automated analysis inaccuracies, and occasional precision issues. While this distrust diminished over time with practice, it remains a significant barrier for some team members.

Another concern was insufficient training: 30% of respondents felt they lacked adequate preparation to fully leverage the tools. This highlights the ongoing need for continuous training and adaptation of technological solutions to the specific workflows of each newsroom section.

Technical limitations were also identified, mainly related to tool customization and integration with more complex workflows. Although most solutions were functional, some teams noted that certain processes could not yet be fully automated or effectively assisted.

Recommendations and Future Outlook

Among the most frequently mentioned recommendations were:

• Improve staff training (50%)
• Increase the precision of AI tools (40%)
• Explore new areas of application (30%)

Additionally, 95% of the team would recommend AI implementation to other newsrooms, although 20% would do so with some reservations, emphasizing the need for an integration strategy that combines technology, training, and constant editorial review.

Post-Implementation Indicators: Results from the First Quarter of 2025

In addition to the perceptions gathered through the internal survey, the impact of artificial intelligence at 0221 was also reflected in concrete editorial production metrics. One of the most significant data points is the growth in so-called “production stories” — articles developed by the newsroom within a single working day. This increase was driven by the time freed up through the use of AI tools, allowing the newsroom to produce more high-value content over the course of a day.

Statistically, during the first quarter of 2025, the newsroom published 39 production stories, compared to 17 in the same period of 2024. This represents a 129% year-over-year increase in this type of content, which typically demands greater time and dedication from the team of journalists.

This growth in production stories is particularly noteworthy considering that the total number of published pieces on the site increased by only 2.2% during the same period (from 4,048 to 4,138 pieces of content). In other words, the increase was not due to a higher volume of articles overall but rather reflected a strategic focus on improving the quality of editorial work.

This result reinforces one of the core principles of 0221’s AI integration process: not using technology to mass-produce “commodity content,” but rather to automate operational tasks, free up creative time, and boost the generation of original, relevant, and strategically valuable pieces.

It’s important to note that these results were achieved within a particular context: it was a pilot and adaptation period, coinciding with the summer holiday season in Argentina, and the newsroom’s use of AI tools was still developing, requiring ongoing training and daily practice. Even so, the change has set a promising path forward and encourages further progress along this line of work in the coming months.

Conclusion

On a personal note, I believe that the AI integration process at 0221 left valuable lessons that go far beyond the technical aspects and can be useful for other local media outlets facing similar challenges.

Above all, it was not just about implementing new tools — it was about rethinking processes, roles, and priorities within the newsroom. The experience showed that AI can become a powerful ally when it is incorporated critically, gradually, and with a clear editorial orientation.

For local media, which daily face the tension between urgency and depth, and between limited resources and the need for innovation, initiatives like this open a concrete path toward new models of work. The key is not simply to adopt technology, but to appropriate it with editorial purpose and strategic focus, looking toward the future.

Artificial intelligence is far from solving all the challenges of journalism. But when implemented thoughtfully, it can become a powerful ally in what has always defined our craft: telling relevant stories with context, sensitivity, and purpose.

The goal is not to replace who we are, but to strengthen it. To use artificial intelligence to reinforce the journalistic mission: to produce better, with greater discernment, and with a more genuine connection to our audiences.