Make Me Ideas
Django media monitoring platform for the Aguas Firmes water sustainability program. Automatic collection with SerpAPI, analytical dashboard and AI-based tone classification.
This project came with real urgency: the program had years of coverage with no system to centralize it. The most complex challenge wasn't technical but methodological — designing a capture process that was exhaustive, reproducible and editorially reliable. I learned that in media monitoring, the search architecture is everything: with the wrong queries you can end up with hundreds of irrelevant entries or miss the most valuable coverage. Integrating AI not as a substitute for editorial judgment but as an amplifier was the most valuable lesson from this project.
Aguas Firmes is a public-private alliance between GIZ Mexico, Grupo Modelo and Industria Mexicana de Coca-Cola for aquifer recharge in Calera (Zacatecas), Apan (Hidalgo) and Mexico City. Make Me Ideas needed a tool to centralize all media coverage of the program — press, digital portals and social networks — with real qualitative context: who speaks, from which region, with what tone and representing which ally. Manual monitoring was unsustainable at the program's scale.
I designed a capture system with 5 parallel lanes operating monthly. Lane 1 uses SerpAPI + Google News to track mentions in media and press. Lane 2 deepens per published entry, searching for related articles of the same event. Lane 3 —on the roadmap— will cover source-based diffusion via Meta API and X API. Lane 4 searches directly on X, LinkedIn and Facebook via program queries. Lane 5 uses Claude AI via ZenReflex for contextual research: generates queries about actors, events and technical terms that conventional search engines don't index, discovering niche coverage invisible to previous lanes.
Each entry goes far beyond the title and URL. The platform classifies by content type (media, social post, milestone, event), lead ally, region or aquifer, program phase (1 or 2), source with its editorial typology and over 90 active sources mapped with their social accounts. Thematic tags, social media engagement metrics, program milestone numerical values and editorial notes are recorded. This qualitative layer is what transforms a press clipping into actionable communication intelligence.
Editorial tone classification —positive, neutral or negative— is performed by Claude AI integrated through ZenReflex, my own multi-model access platform. The process analyzes each mention's content in context: not just whether the article mentions the program, but how it positions it, what narrative it builds and which actors appear. 85% of analyzed coverage was positive. ZenReflex also feeds the Capture Lane 5, generating advanced contextual queries that discover coverage conventional searches can't reach.
The dashboard presents executive KPIs in three blocks: program totals (2020–2026), last closed month analysis with variation vs previous month and vs same month last year, and evolution charts with period selector. Each indicator is segmented by web coverage, social media presence and engagement metrics. Charts include annual coverage evolution, engagement evolution, editorial tone, top sources, distribution by ally and by region. The communications team can identify in seconds whether coverage grew, which ally generates the most mentions and whether tone remains positive.
The Timeline view organizes entries in a vertical accordion by year and month, with coverage statistics per period. A stackable filter system allows segmentation by type, ally, region, program phase, year and source — applied via AJAX without page reload. The custom administration panel includes a complete editorial workflow: review of automatically captured entries, confirmation, qualitative field completion, social media metrics loading and source management. Only what the editorial team validates gets published.
The technical expertise behind this monitor is available for new projects. I can develop custom media monitoring systems for any CSR program, public institution, communication campaign or brand that needs to centralize its media presence with real qualitative intelligence. Each development is independent and fully tailored: its own sources, its own indicators, its own identity. If you manage impact programs and need to know what the world says about your work, I can build exactly that for you.
The platform records over 300 published and verified entries covering the program's complete period. Over 90 active sources are monitored including digital media, press, governmental, academic and institutional portals. Social media coverage totals over 12,800 weighted engagement points across LinkedIn, Facebook and X. 85% of media mentions have a positive tone. The coverage peak was recorded in March 2026 with over 35 simultaneous mentions generated by a high-level official visit to Zacatecas — captured, classified and available in the system within 48 hours.
