Teach your AI email drafts from your team's real edits

Compares AI-drafted support replies to what your team actually sent, then uses the differences to improve future drafts.

How the work actually flows

It branches. Exactly one path is taken; runs once per one per sent email.

Pattern: Exclusive Choice (4) · Multiple Instances with a priori Design-Time Knowledge (13)

flowchart TD trig(["scheduled check of sent emails"]):::trigtime s0[["match sent email to draft"]]:::mi s1["compare differences with ai"]:::svc s2[("save edit as example")]:::store trig --> s0 s0 --> s1 s1 --> s2 gx{"× does edit reveal a gap"}:::gate s2 --> gx p00["update knowledge base entry"]:::task gx -->|"knowledge gap found"| p00 p10["no further action"]:::task gx -->|"no gap found"| p10 p00 --> out p10 --> out out[/"improved drafts and updated knowledge base"/]:::out pay{{"ai keeps improving without extra work"}}:::pay out --> pay classDef task fill:#e7f6fe,stroke:#34b8f0,color:#2c2a29 classDef svc fill:#f6f8fa,stroke:#7c8795,color:#2c2a29 classDef mi fill:#e7f6fe,stroke:#0079a8,color:#2c2a29,stroke-width:2px classDef human fill:#fff,stroke:#0079a8,color:#0079a8 classDef store fill:#f6f8fa,stroke:#0079a8,color:#2c2a29 classDef trig fill:#00a4eb,stroke:#0079a8,color:#fff,font-weight:bold classDef trigtime fill:#00a4eb,stroke:#0079a8,color:#fff,font-weight:bold classDef trigdata fill:#8ad4f5,stroke:#0079a8,color:#06314c,font-weight:bold classDef gate fill:#fff,stroke:#e8a23d,color:#6b4708,font-weight:bold classDef out fill:#1f9d6b,stroke:#167a53,color:#fff,font-weight:bold classDef pay fill:#06314c,stroke:#021f33,color:#fff
Starts itAn outside serviceRuns once per itemA record or sheetOne path onlyResultPayoff
Build size
Advanced

A larger build with multiple systems, AI reasoning, and custom rules.

Business functions
AI Agents & Autonomous SystemsEmail AutomationSpreadsheet & Database OpsEducation & TrainingCustomer Support & Ticketing
Connects
GmailOpenAIPostgreSQL

The problem it solves

Your team edits AI-drafted email replies every day, but those corrections disappear the moment the email is sent. The same mistakes and gaps in the AI's knowledge keep showing up, because nothing captures what your team actually changed.

Who it fits

Support teams already using an AI draft-reply system for their support inbox who want it to keep improving.

How it works

  1. On a set schedule, the system checks Gmail for support emails your team has sent
  2. Each sent email is matched to its original AI-generated draft
  3. AI compares the two and identifies what was changed and why
  4. Genuine edits are saved as examples so future drafts can reference real human corrections
  5. If the edit reveals missing information, the relevant knowledge base entry is updated automatically
What you get

Drafts that keep sounding more like your team

Your AI-drafted replies keep getting closer to how your team actually writes, learning from every real edit your team makes.

What you get

A growing library of real edits that makes future AI email drafts more accurate, plus an updated knowledge base.

What you need

A Gmail account, an OpenAI API key, and a PostgreSQL database.

We can build this. But should you?

The hard question is not how to build it. It is whether this is the right thing to build first.

That is what a Fractional Chief AI Officer figures out with you, before anyone writes a line of code.

Let's Talk Strategy

Related automations

Back to the AI Playbook