The head of business development at a web3 advisory firm needed to answer partnership and sponsorship emails faster without ever saying something about a token or a return that the firm must never put in writing. We set up a Claude based assistant inside his Gmail that researches every sender, drafts in the firm’s voice and waits for his approval on every send.
AIM (AIManagement Inc.) built the assistant for a New York web3 advisory, events and marketing firm. Its VP of Business Development handles inbound partnership requests, event sponsorships and outbound outreach from a single inbox, in an industry where one careless sentence can create legal exposure.
Reply speed was winning and losing deals, but a good reply takes research: who the sender is, what they have shipped, which of the firm's services fit. Generic AI drafts sounded generic, and some of them hinted at price expectations or investment outcomes the firm can never put in writing.
A seven file implementation package built on the Claude app and its native Gmail connector, with no outside middleware.
Claude handles the research, the drafting and the first compliance pass. The human handles judgment, relationships and the send button.
| Metric | Value |
|---|---|
| Emails sent without human review | 0, by design |
| Files in the implementation package | 7 |
| Steps in the email agent workflow | 7 |
| Middleware or custom code required | None |
The best AI email assistant is the one your compliance counsel will approve. Human approval on every send, backed by written guardrails, turns AI from a risk into a routine.
No. It drafts inside Gmail and waits for approval on every message. There is no autonomous send path in the design.
No. It runs in the standard Claude desktop or web app through the native Gmail connector, configured by AIM with prompts and guides written for the role.
Yes. AIM tailors each package to the client's role, industry and firm. The compliance guardrails change; the workflow does not.
Related: AI Automation · The Intelligence Node for a fractional advisory firm. See the rest of the work on the case studies index.
Bring one workflow, its owner, and what an incorrect result would cost. That is enough to scope the first engagement.