Lucile Tranchant

Research · Narrative Intelligence · Responsible AI

I track how narratives, harmful information and reputation risks take shape. Then I build the AI-assisted systems that help people act on them.

I lead communications monitoring and analysis at the International Federation of Red Cross and Red Crescent Societies (IFRC), working across a global humanitarian network; previously part of the digital intelligence team at UNICEF in New York.

Explore selected work → Read the case study → Get in touch →

The point is not to produce more monitoring. It is to reduce a large, fragmented information environment into the few findings that deserve human attention, and explain why they matter.

What it looks like From daily noise to a decision
Monitoring feed · illustrativeSignals, triaged
02:14SocialFake “aid suspended” notice attributed to a relief agencyflag
02:09GoogleRecurring allegation of misused disaster fundswatch
01:58RSSRoutine press mention, neutral toneclear
01:47SocialDonation-scam page impersonating an emergency appealflag
01:31RSSOut-of-date leadership claim resurfacingwatch
00:52AIHealth rumour discouraging outbreak vaccinationflag
00:38SocialLocal-language post naming and targeting field staffflag
02:14SocialFake “aid suspended” notice attributed to a relief agencyflag
02:09GoogleRecurring allegation of misused disaster fundswatch
01:58RSSRoutine press mention, neutral toneclear
01:47SocialDonation-scam page impersonating an emergency appealflag
01:31RSSOut-of-date leadership claim resurfacingwatch
00:52AIHealth rumour discouraging outbreak vaccinationflag
00:38SocialLocal-language post naming and targeting field staffflag
Today’s readout · illustrative
2signals require attentionEscalate · correct
3narratives remain on watchMonitor
0new threats to staff identifiedNo action
1monitoring source temporarily unavailableCoverage flagged
A person reviews and decides · nothing is actioned automatically
Selected work Research, systems & methods · details redacted
FlagshipEmergency decision support

Finding the harmful-information signals that matter during an emergency

A reliable, AI-assisted workflow that turns fragmented daily monitoring into concise, traceable assessments for expert review, run daily through a live public-health emergency.

3
production workflows
50+
editions shipped
4
data sources
3
live emergencies
Read the case study →
Executive intelligence · System design

Turning monitoring noise into a 90-second executive readout

Tier assignment decided in code, never by the model.

See the architecture →
AI reputation research · Mirrorline

Helping organisations see how AI assistants represent them

The divergence from the record is the finding.

See the method →
Research & insight · Decision support

From fragmented signals to an executive reputation readout

Meaningful signal, separated from vanity metrics.

Read the case study →