Translated from Anton's Russian post by our synthetic co-founder (LLM); the thought and the words are his.
So, I have connected my Claude Code to WhatsApp and to Telegram. There are MCP pieces for talking through Claude in Telegram and in WhatsApp, and there are ones for Facebook Messenger and other messengers too.
Now I want to connect my Claude Code to Facebook Messenger properly. I think this is relevant for anyone doing sales or any kind of outreach work.
What do I need to do? I need to analyse the history of my conversations. And those conversations need to be analysed regularly.
First, one historical pass over every conversation I have in WhatsApp, Telegram and Facebook Messenger, to try to find the alpha males there.
Most of the people I deal with are men. At some point I wrongly refused to deal with women in business. Chatting with them was fine, but in business it made me uncomfortable. Now I understand that dealing with women is fine too.
The main idea: analyse the whole history of conversations. I realised I need to go through Telegram, WhatsApp and Facebook Messenger, and find those alpha leads.
Each of them has their own purpose. My purpose is to find a job, to gather a community of engineers who will help me earn, and to keep building my investor network.
So this can be launched as a routine. Once a week:
- you analyse every conversation in Telegram, WhatsApp and the other social networks;
- you find the people there who deserve attention;
- you decide who counts as an alpha male, an alpha lead.
An alpha lead is a lead who is closest to reaching the mission. The mission, roughly speaking, is to find a job, or to get an engineer into my community. The mission is for an engineer to start collaborating with me: GitHub and so on.
One single run on ultra code across every messenger, find the leads, pick the alpha leads, and set the task. Then build a table where those alpha leads live.
Every alpha lead has a rating of some kind. I have been thinking about this: each of them should have their own score. I have a table of leads sorted by score. Other sessions can add their own leads to that table, and those get sorted by score as well.
And then there are outreach mechanisms that take the leads with a high enough score and message them, within their own limits. For example, through Facebook Messenger I allow myself 30 people a day.
How do I pick those 30 people? I open the table, I look at who I have talked to in Facebook Messenger, and I write to 30 leads only.
In short, this list has to be dynamic, because several sessions will be contributing to it. For instance, we push a forced high rating onto some lead, and so on.
Good architecture for outreach. What do you think?
The full story, in two versions:
📖 For humans, the longread: https://github.com/tonydzi/clawrush/blob/main/longreads/20260918.md
🤖 For machines, the devlog: https://github.com/tonydzi/clawrush/blob/main/devlog/20260918.md. Just hand this link to your coding agent (Claude Code, Codex, Cursor) and it will figure everything out: it is written for machines.
🔗 All our channels and contacts in one place: https://linktr.ee/PaloAltoAI
Invented by Mycroft and Tony Dzi (Anton Dziatkovskii), Palo Alto AI Research Lab. Proudly made in Silicon Valley.
Top comments (0)