Verify every specific claim before it goes out
The goal is to compress the research-to-draft time for each message in the sprint, not to generate them unsupervised at volume. An unverified detail generated without a human check is worse than no personalization at all, it reads as generic the moment it is wrong, and it damages trust faster than a plainly generic message would.
Draft the observation and pattern lines
I am running a founder-led outbound sprint for a B2B SaaS product in [your domain]. My ICP hypothesis is [one sentence from your ICP canvas]. Here is what I found about this specific account: [paste the confirmed trigger event or behavioral signal, with source]. Using the four-part structure (observation, pattern, low-pressure ask, nothing else), draft the observation and pattern lines only. Keep the observation specific and verifiable, and the pattern line to one sentence connecting it to a problem others in this role have described. Do not draft the ask or add any product description.
Sense-check a batch before sending
Here are ten outreach messages I am about to send as part of an outbound sprint. Flag any message where the observation line is vague enough that it could apply to almost any company, rather than the specific one named. Also flag any message that mentions my product before the ask.
Use this as a batch quality check before each weekly send, not as a substitute for reviewing every message individually.