A team of seven AI agents scans hundreds of sources every day to find what's actually working in social impact — before everyone else.
Every day, real innovations quietly reshape how the social sector works. A new financing model in Mali. A health AI pilot in Rwanda. A governance experiment in Estonia. Most of them never make it past a press release or a buried PDF report.
We built a pipeline of specialized AI agents — each with a different job — that continuously monitors news outlets, research repositories, UN agencies, NGO reports, academic databases, and policy trackers. They don't just collect links. They read, cross-reference, verify evidence, and flag the innovations that have real substance behind them.
Think of it as a daily intelligence brief for people who fund, build, and implement social impact.
Seven agents, each specialized. One innovation published daily.
Monitors 200+ sources across news, research, policy, and social media. Runs every 6 hours. Flags anything that looks like a concrete innovation — not opinion, not hype.
Catches the same story reported by multiple outlets. Merges signals, keeps the richest source. Prevents the same innovation from appearing twice under different names.
Scores each signal on novelty, evidence strength, geographic spread, and sector balance. Ensures the daily feed covers different sectors and regions, not just whatever's trending on Twitter.
Deep dives the top signals. Fetches primary sources, searches for independent evaluations, finds parallel implementations in other countries, and checks for counter-evidence. Kills weak signals here.
Gives every innovation a 2-4 word name that captures the pattern, not the organization. "Harvest-First Financing" instead of "AICCRA program." The name is the product.
Produces a ~120 word story: the problem, the innovation, the evidence. Numbers over adjectives. Active voice. No jargon. Every number sourced.
Final quality gate. Verifies every claim has a source. Checks that the name passes the "would you say this to a colleague?" test. Flags anything that feels off for human review.
TrendWatching ran "Innovation of the Day" for years with a team of human analysts covering consumer trends. They discontinued it in 2024. Nobody picked it up for social impact — the sector where innovation intelligence arguably matters most.
We're testing whether AI agents can do what editorial teams used to do — scan the landscape, spot patterns, verify evidence, and surface what matters — but across more sources, more languages, and more sectors than any human team could cover.
Everything here is AI-curated and human-reviewed. The agents find, research, and draft. A human checks the evidence and hits publish. AI as the engine, human judgment as the filter.
Built by Kaio Freitas as an open experiment in AI-powered content intelligence.