Nerdearla Argentina 2026 CFP: AI, agents, and a community that wants to build

We closed the Nerdearla Argentina 2026 Call for Papers and the first thing we want to say is: thank you!

We received 718 proposals to share knowledge, experiences, lessons learned, demos, mistakes, hypotheses, and questions with the community. More than 700 ideas submitted by people who want to get on a stage —or on a screen— to talk about what they are building, what worked, what did not, and what they think comes next.

Now the Evaluation Committee begins its work: reading, discussing, and scoring each proposal anonymously, track by track. Meanwhile, we took a first look at the full set to understand which topics the community is pushing forward this year.

This is not the Nerdearla Argentina 2026 agenda yet. It comes before that and, in a way, is just as interesting: a snapshot of the moment. What worries teams, what excites them, and which questions are showing up across the region’s technical communities.

AI stopped being “a topic” and started showing up almost everywhere

The track with the most submissions was Data Science / AI, with 175 proposals. But if we look at titles, abstracts, and tags, artificial intelligence appears far beyond that track: in development, infrastructure, product, security, UX, management, and even conversations about team culture.

The difference from previous years is not just quantity. It is tone. We are no longer seeing only talks like “what is generative AI” or “how to use a prompt”. This year, more concrete questions keep appearing:

• How do I take a proof of concept to production?
• How do I evaluate whether a model actually helps?
• What happens with quality, cost, observability, and governance?
• How do I integrate AI into existing workflows without breaking everything?
• Which decisions should be automated, and which ones should not?

Agents, MCP, and the end of the “magic demo”

One of the most visible topic clusters was agents, MCP, and autonomous workflows. In a keyword-based reading, around one in five proposals mentions agents, agentic AI, MCP, or related patterns. And 32 proposals explicitly mentioned MCP or Model Context Protocol.

That says a lot. The community is not only asking “what can a model answer”, but what it can do when it has context, tools, memory, permissions, and clear limits.

We saw proposals about coding assistants, agents for analytics, AI-powered incident response, declarative infrastructure, skills, memory for copilots, MCP servers, spec-driven workflows, and tools that connect models with real systems.

There is also a healthy concern: how to move from the viral video or the perfect demo to systems that survive real-world use. How to design agents that are not black boxes, that can be audited, observed, limited, and turned off when needed.

Data: the foundation that keeps coming back

AI took many of the headlines, but the dataset reminds us of something obvious: without data, there is no magic.

Data, analytics, data engineering, pipelines, lakehouse, BI, observability, and data quality appear in more than 200 proposals. Often connected with AI: RAG, embeddings, semantic search, model evaluation, source integration, and data preparation for agents.

The message is clear: the data layer is still where many projects are won or lost. The promise of “adding AI” becomes fragile if data is scattered, hard to understand, ownerless, or unreliable.

Infrastructure, Kubernetes, and cloud are still at the center

Infrastructure received 87 proposals and, combined with mentions of DevOps, SRE, Kubernetes, cloud, platform engineering, CI/CD, observability, and FinOps, the topic remains very present.

What is interesting is that infrastructure is also starting to mix with AI. Not just as “where do I run the model”, but as a broader question:

• How do we operate infrastructure in a world with copilots and agents?
• What changes in observability when systems are making decisions?
• How do we control the cost of increasingly dynamic workloads?
• How do we design internal platforms that do not just deploy software, but help build it?

Kubernetes, cloud, and platform engineering no longer appear in isolation: they are part of a conversation about productivity, reliability, and scale.

Security: more surface area, more questions

Security had 51 proposals, and mentions of security, privacy, supply chain, authentication, compliance, vulnerabilities, and cybersecurity also appear across other tracks.

The expansion of AI and agents adds a new layer of questions. If a system can read context, use tools, execute actions, or make decisions, then security is no longer only about “protecting the perimeter”. It is also about designing permissions, traceability, limits, human review, and safe-by-default behavior.

We saw classic topics —OWASP, cloud security, containers, secrets, privacy— and newer topics related to the surface area opened by intelligent automation.

The human side did not disappear: it grew

Softskills, Management, and Product added up to 155 proposals across the three tracks. And if we also look at mentions of culture, teams, leadership, career, communication, community, and mentoring, the topic appears in roughly a third of all submissions.

That is also a signal. In a year where so much technical conversation is shaped by AI, the community keeps bringing human questions:

• How do we lead teams in changing contexts?
• How do we learn faster without burning out?
• How do we communicate technical decisions?
• How do we build meaningful products?
• How do technical, product, design, and business profiles work better together?

Nerdearla has always had that mix: code, infrastructure, data, and security, yes; but also community, careers, real stories, and conversations.

Development, UX, Open Source, Maker, and the rest of the map

Development was the second-largest track, with 109 proposals. Architecture, backend, APIs, performance, testing, programming languages, legacy systems, patterns, and new AI-assisted ways of building software all appeared.

UX had 39 proposals, focused on design, accessibility, digital experiences, and product. Open Source received 25. Maker/Retro, 20. Nerd, 47: that hard-to-classify space where history, culture, science, creativity, technical oddities, and the kinds of talks that make Nerdearla feel like Nerdearla all fit.

There were also proposals for Testing and Kids. Few in number, but important to sustain two conversations we care about: how we build software with higher quality, and how we open doors for people who are just getting started.

What comes next

Over the next few weeks, the jury will read and score the proposals. After that comes the difficult work of building a diverse, interesting, and representative agenda: not only by topics, but by levels, formats, approaches, stories, and voices.

If you submitted a talk: thank you for taking the time. We know that writing a good proposal takes work, and that behind every abstract there is experience, curiosity, and a desire to share.

If you did not get to submit for Argentina, there is still a nearby opportunity: the Nerdearla México CFP remains open until June 30.

You can submit your talk at nerdearla.com/cfp.

See you at Nerdearla.

Submissions by track

Distribution of the 718 proposals received for the Argentina 2026 CFP, by track.

Data Science / AI | ██████████████████████████████████ 175 (24.4%)
Development       | █████████████████████ 109 (15.2%)
Infrastructure    | █████████████████ 87 (12.1%)
Softskills        | ██████████████ 74 (10.3%)
Security          | ██████████ 51 (7.1%)
Management        | ██████████ 49 (6.8%)
Nerd              | █████████ 47 (6.5%)
UX                | ████████ 39 (5.4%)
Product           | ██████ 32 (4.5%)
Open Source       | █████ 25 (3.5%)
Maker/Retro       | ████ 20 (2.8%)
Testing           | █ 7 (1.0%)
Kids              | █ 3 (0.4%)