Agents judge, people approve, code protects: how Dylan is built
How Dylan splits the work: AI agents make the judgment calls, people approve what goes out, and plain code enforces the rules that must never bend.
- ai-agents
- compliance
- guardrails
How Dylan works, what the team behind him learns running outbound, and the rules he keeps.
RSS feedHow Dylan splits the work: AI agents make the judgment calls, people approve what goes out, and plain code enforces the rules that must never bend.
AI SDRs promised to replace sales teams. Most early rollouts disappointed. What survived: proven playbooks, signals, people at decision points, inbound.
Gmail's bulk sender rules get the headlines, but B2B cold email is decided by corporate gateways, Microsoft tenant limits and new-domain checks.
Only 1,138 US companies have 10,000+ employees. The outbound math for small markets, why precision beats volume, and what to do when the list runs out.
A brief decides who gets emailed, what they read and which meetings count. The seven inputs that matter, and why to edit a draft instead of writing one.
Booked, held and qualified are three different meetings. How Dylan defines a held meeting, what is never billed, and how disputes and the deposit work.
Retainers, seats and per-lead pricing leave you holding the risk. Why Dylan charges per held meeting instead, and where outcome pricing can go wrong.
Every first email from Dylan says he is an AI agent and who he works for. Why we disclose, what the EU AI Act and CAN-SPAM say, and how to stop him.