Feature 02 — AI Engine
Conversations that fool spam filters. Because they look real.
Static warmup templates have a massive detection footprint. SimplyWarmup uses LLMs to generate subject lines, body copy, and reply threads that are unique per send — every time, automatically.
Generated conversation thread
Thread depth: 3 turns · Model: Gemini
Following up on our last call — the engineering team flagged some concerns about the API rate limits. Can we schedule 20 minutes this week to align on the revised spec?
Thursday at 2pm works on my end. I'll pull up the v2 schema beforehand. Do you need the sandbox credentials again or do you still have them from last sprint?
Thursday works. I'll resend the credentials — ours expired in the rotation. See you then.
Why static warmup templates don't work anymore.
Google and Microsoft train their spam models on warmup traffic. They know what it looks like.
Static "Got it, thanks!" templates
Traditional warmup tools rotate through a small set of generic reply templates. Modern spam classifiers can identify this pattern in under a second using embedding similarity and vocabulary diversity scores.
LLM-generated unique threads per send
Every email in the pool is independently generated using Gemini. Subject line, opening, body, and any replies are all unique per conversation thread. Vocabulary diversity and contextual coherence pass ML filters.
How the AI engine generates a warmup cycle.
Each daily warmup cycle runs through a four-stage generation pipeline before a single email is dispatched.
Plan generation
The orchestration service computes the day's send targets based on current pacing level, health scores, and business-hour slots. It selects conversation partners from the pool — never repeating recent pairs.
Subject & body generation
Gemini generates a subject line and a business-context body for the outbound email. Contexts include planning meetings, technical reviews, vendor evaluations, and project follow-ups — rotating to prevent footprint clustering.
Reply candidate pre-generation
For 2–4 turn thread depth, Gemini pre-generates reply candidates that logically continue the generated thread. Replies reference the content of earlier messages to ensure contextual coherence that ML classifiers validate as human.
Randomised business-hour dispatch
Send times are randomised within 09:00–17:00 local business hours. Send intervals are varied to avoid mechanical regularity. Pacing advances by 2 emails/day per warmup stage, capped at 40.
FAQ
Common questions about the AI engine
Can Google and Outlook detect AI-generated warmup emails?
Modern ESPs evaluate engagement signals — reply rates, open timing, conversation depth — rather than reading email content for AI detection. What triggers spam filters is pattern repetition: the same template rotating across thousands of sends. SimplyWarmup generates each warmup email uniquely via Google Gemini, producing varied topics, lengths, and conversation threads. The result is indistinguishable from authentic business correspondence at the signal level that ESPs actually evaluate.
What is a multi-turn warmup conversation and why does it matter?
A multi-turn warmup conversation is a 2–4 message back-and-forth exchange where each reply contextually references the prior message. Unlike single-shot warmup emails, multi-turn threads mimic real business correspondence and generate richer engagement signals — longer read times, more replies, higher thread depth. These are the exact signals that distinguish legitimate business email from bulk sends in ESP reputation models.
Does email content matter during warmup?
Yes. Warmup emails containing spam-trigger words or marketing-style copy can receive spam reports from pool participants. More critically, multi-turn conversation content must read as authentic business correspondence — our Gemini-generated threads cover plausible vendor reviews, project planning, and scheduling contexts that don't pattern-match to spam.
Is the same warmup system used for Google Workspace and Microsoft 365?
Yes, but the protocols differ. Google Workspace requires a minimum 15-day warmup period and weights domain reputation and engagement patterns heavily. Microsoft 365 Premium inboxes can begin cold sending after 3–5 days minimum (10–14 days recommended). SimplyWarmup automatically applies appropriate pacing and conversation strategies for each inbox's provider — you don't need separate tooling for each.
Industry data: Google Workspace inboxes require a minimum of 15 days warmup before cold sends produce reliable inbox placement. Microsoft 365 Premium requires 3–5 days minimum, 10–14 days recommended. Rushing these timelines causes 23% more spam folder placements in the first month.