W15: Sales AI email engine, charter sync, and biznesam.ai identity
A week of ecosystem-wide progress — VT sales AI started generating personalized offers, charter price sync between CMS and Nerve Center, biznesam.ai shifted to a personal story, and the RAG wiki keeps growing.
- Sales AI response engine — personalized client emails
- Charter price sync CMS ↔ Nerve Center
- biznesam.ai identity: from 'we' to 'I'
- Weekly experience section (this one!)
- Ecosystem RAG wiki — continuous reindexing
- Authenticity > corporate language — a personal story resonates better than an abstract 'community'
- Sales AI + human review = the best combo. AI generates, human approves.
- Ecosystem memory (RAG) is infrastructure, not features — it grows in the background and delivers long-term value
- biznesam.ai positioned as 'expert community' — in reality it's one person
- CTO DECISIONS.md had to be updated multiple times — decisions weren't captured immediately
Ecosystem overview
This week saw activity across 10+ projects. Here are the highlights:
Vanilla Travel ecosystem
Sales AI email engine (vt-internal-services)
This week we built the Sales AI response engine — a system that generates personalized email offers for clients. Three main components:
- LLM client — unified interface for AI model calls
- Scoring engine — evaluates client requests and matches the best offer
- Response engine — generates ready-to-send email text for manager review
Core principle: AI generates, human approves. Never send automatically.
Charter price synchronization (vt-cms + vt-nerve-center)
Built a charter price sync pipeline between CMS and Nerve Center. Now charter prices from suppliers automatically flow into the central system and are available for sales AI and dashboards.
Marketing dashboard modularization (vanilla-travel)
Continued refactoring the dashboard server — splitting the monolith into smaller modules for easier maintenance and extension.
Email reactivation sequence
Developed a complete email reactivation sequence with copywriting rules — targeting clients who haven’t traveled recently.
Social media content
Facebook engagement posts about Greece — the organic content pipeline keeps running.
biznesam.ai
Identity shift — from “we” to “I”
The biggest conceptual change — moved from “we are an expert community” to “I’m a business leader who builds AI solutions”. Changed navigation, About section, all headings in both languages.
Experience section
Created this section — a weekly journal of the entire ecosystem. Accomplishments, discoveries, mistakes — unfiltered.
AI news format with LinkedIn
Daily AI briefings with built-in LinkedIn post copy functionality.
Infrastructure and memory
Ecosystem RAG (vt-ecosystem-rag)
The wiki and index keep growing automatically — 50+ reindexes this week. This is the ecosystem’s “brain” that ensures every AI agent knows what’s happening in other projects.
CTO decision tracking (claude-workspace)
Updated CTO DECISIONS.md multiple times — architecture and infra decisions now documented centrally.
Daily retrospectives (claude-desktop-support)
Automated daily retrospectives keep capturing each day’s work. This is the foundation from which weekly recaps are generated.
Insights
Sales AI + human review = the optimal formula. Full automation in client emails is risky — clients feel when text is machine-written. But AI that prepares a draft the manager reviews in 30 seconds — that’s 10x faster than writing from scratch.
Ecosystem memory is invisible infrastructure. RAG wiki, retrospectives, DECISIONS.md — it all looks like “overhead”. But when a new session immediately knows the context without 20 minutes of explaining — the payoff is massive.
Authenticity > corporate language. When a business leader looks for a partner, they want to see a person, not a “community”. The personal story resonates.
Mistakes and lessons
biznesam.ai identity. Initially positioned as an “expert community” — but in reality it’s one person with real experience. Lesson: don’t try to look bigger than you are.
Delayed decision capture. CTO decisions weren’t recorded immediately — resulting in multiple updates, and other agents not knowing about new decisions in between. Lesson: capture a decision the moment it’s made.