How does AI decide who's interested in your product?

Most "AI outreach" stops at generating text. TokoAI runs three models in sequence, before, during, and after every send - each solving a different question.

Model 1: open-probability - who to contact

Before sending, the open-probability model scores which leads are likely to open at all. It learns from past opens across similar companies. Sending to a lead it scores low wastes the send and pollutes the signal - so it filters first.

Model 2: interest-density - what to say

Given a recipient, the interest-density model picks the angle and content most likely to resonate - based on their industry, role, and what similar recipients engaged with. One product, many valid messages; the model chooses the right one per recipient.

Model 3: signal-funnel - who to follow

After sending, the signal-funnel model ranks responses and behaviors - opened, clicked, forwarded, replied - into a priority order. You don't read a raw inbox; you get a ranked shortlist of who's worth talking to.

Why sequence matters

The three aren't independent. Model 1's filtering improves Model 2's training data; Model 2's content quality improves Model 3's signal clarity. They tighten each other every round.