Tecnologías VM
Work
AI sub-brand · since 2018

Nairu AI

Nairu is the AI sub-brand of Tecnologías VM: a boutique intelligence practice that has served political parties, presidents, ministers and mayors since 2018, reading social media at scale to surface how public conversation really moves.

Selected country
Costa Rica
  • 3 presidential campaigns (2018, 2022, 2026)
  • Mayoral campaign
Where we work: five presidential campaigns, three mayoral campaigns and government work, across four countries. Hover or tap a country.
Imagery

The platform, in production.

A look at Nairu: the nairu.ai site, the live social-listening dashboard, the interactive weekly report, the mobile PDF summary and the public AI insight briefings at insights.nairu.ai.

The nairu.ai homepage: the public face of our political and brand intelligence practice
The Nairu Social Analytics platform: secure client sign-in
Social listening: post volume, reach and the day's top posts
The interactive weekly report: cover and section navigation
Weekly report — Topics: leading themes, sentiment and top mentions
Weekly report — Topics benchmark: share of conversation by theme
Weekly report — Public figures: reach, sentiment and top mentions
Weekly report — Figures benchmark: share of voice by public figure
The same intelligence as a mobile-friendly weekly PDF summary
Nairu Insights: AI-written candidate briefings, Costa Rica — insights.nairu.ai
A candidate briefing in full: Laura Fernández (Costa Rica)
Nairu Insights: AI-written topic briefings, El Salvador — insights.nairu.ai
A topic briefing in full: El Salvador's aid to Jamaica
01 / 13
5Presidential campaigns
3Mayoral campaigns
50K+Posts analyzed / day
200K+Comments / day

We started where the data was: a social-media monitoring system, built on Natural Language Processing and machine learning, that pulled in posts and comments and read them with NER, tagging and sentiment models, all searchable through Elasticsearch. Our first engagement put us inside the war room of the 2018 Costa Rica presidential runoff, delivering daily and weekly insight.

Over the years that data work sharpened into intelligence. We went from statistical dashboards to LSTM-based topic and sentiment models, then to generative reports, then to RAG over a vector index, and by 2025 to LangGraph agents that operate across our ocean of posts and comments to surface the patterns that matter.

We can fairly say we were a pre-AI company that already understood the ideas behind big data, machine learning and deep learning long before “AI” became the label. Today Nairu builds the insights and agents that let Central American clients decide on evidence, not guesswork.

Timeline

From data tooling to agentic AI.

  1. 2018

    Foundations

    Social monitoring on NLP + Elasticsearch; first Costa Rica presidential war room.

  2. 2019

    Visualization

    Power BI dashboards and mobile-friendly PDF reports; first brand clients.

  3. 2020

    Central America

    First El Salvador client; daily and weekly reports, sustained for years.

  4. 2022

    Deep learning

    LSTM topic and sentiment models; second Costa Rica presidential campaign.

  5. 2023

    Generative AI

    LLM-composed insight reports, moving from quantitative stats to validated subjective insight.

  6. 2024

    RAG

    Weaviate-backed retrieval and dynamic reports; first Panama campaign; deep on LangChain.

  7. 2025

    Agentic AI

    LangGraph agents over 50K posts / 200K comments per day; a modern stack of agentic AI, GraphRAG and retrieval working across the region's biggest campaigns.

  8. 2026

    Today

    Insights and agents for AI-backed decisions across the region.

The pipeline

From the ocean of posts to decisions.

  1. 01

    Listen

    Extract posts and comments at scale across social platforms: today, 50K posts and 200K comments a day.

  2. 02

    Process

    NER, tagging, topic and sentiment models: from early NLP to LSTM deep learning.

  3. 03

    Index

    Searchable at scale: Elasticsearch from 2018, vector retrieval on Weaviate since 2024.

  4. 04

    Reason

    RAG, then LangGraph agents operating over the indexed ocean to surface patterns.

  5. 05

    Deliver

    Daily and weekly insight to war rooms and governments: decisions grounded in validated, cited evidence.

Contact

Let’s build something that works.

Have a real problem where emerging technology might be part of the answer? We’d like to hear about it.