Outgoing Agent

Outgoing Agent is an AI agent for personalized event discovery. Mike shipped it to outgoing-agent.mikebapps.com on July 20, 2025. The repository is public. The product originated from a rapid sprint during the lu.ma Hackathon July 2025 build-up, documented in the WhatsApp thread with Achraf Adjileye.

Concept

Repository description: "AI agent to help find personalized event based on user profile." The stated technique is intent matching rather than keyword search: the system interprets what a user wants to do and surfaces events that fit the intent. The stack pairs TypeScript orchestration with a Python backend for the LLM calls.

Origin and Shipping

Outgoing Agent appeared on July 20, 2025, a few days into the hackathon preparation window. Achraf was the first tester and delivered structured feedback over WhatsApp audio the same afternoon: "Affichage pas responsive en mode portrait. Aucun event trouvé pour moi mais temps de réponse rapide du LLM. Par contre problème là je crois, il me dit un truc et son contraire en même temps." The responsiveness issue, the cold-start empty result, and the self-contradictory agent output, all flagged within hours of launch. The timeline shows the Mike and Achraf feedback loop operating quickly: ship, test, feedback, fix, iterate, all in audio.

Stack Decisions

TypeScript frontend, Python backend. In late-July audio with Achraf, the division of labor on a related integration was stated explicitly: "Tu vas mettre le deepsearch dans le meme repo en utilisant Python." The deepsearch pattern (a Python service called from the TypeScript interface) carried forward into later products.

Status

Listed as legacy work in the profile README. The repository remains public. Outgoing Agent sits in the same AI-agent cluster as SageVox, Realicheck AI, Retro AI, and Picture AI. The mikebapps.com deployment shares hosting with several other products.