course module
Prompts 4-5. Drafting on tap. Paste raw input, get back compressed synthesis or 3-angle drafts. Works in any conversation — no schedule, no integration, no agent fuckery.
These two prompts work in a normal Telegram message. You paste a chunk of raw input (an email, an objection, a piece of pipeline data), the agent gives you back something you can act on. Zero plumbing.
Here's a new lead inquiry [paste the email or message].
Tell me: A/B/C fit (A = direct fit for the highest-margin offer,
B = nurture, C = doesn't fit). For an A lead, draft a 3-sentence
personal reply for me to approve. For B, draft a 2-sentence
nurture reply. For C, draft a polite "not the right fit" template.
The triage decision used to be the bottleneck. Now you paste the inbound, get three drafts back, pick one, send. No webhooks, no inbox integration — you do the paste, the agent does the thinking, you keep the judgment call.
I'm about to get on a call with a prospect who said [paste the
objection — "too expensive", "not the right time", "I need to
think about it", whatever].
Draft three different responses I could use, each from a different
angle: empathy-first, reframe-first, question-first. Plus one
risky-but-honest response I probably won't use but should consider.
This is the prompt people use right before a hard conversation. Four angles in 30 seconds. Pick one, internalize, walk in clear-headed. Most people overthink objections — the agent surfaces the obvious responses so your brain is free to listen.
Both of these work because you're doing the orchestration manually (you paste, you decide, you send). That's a feature, not a limitation — keeps the judgment call with you while the agent does the drafting work.
Next module: the two memory prompts. These are the ones that compound.