Pre-Send Email Checks: The 15 Automated Guardrails Every Professional Needs in 2026
Missing attachments, wrong recipients, confidential leaks, reply-all disasters. A complete breakdown of the 15 pre-send checks that prevent the most common — and most expensive — email mistakes.
The Case for Pre-Send Checks
Every professional has a story. The pricing sheet sent to a competitor by autocomplete. The internal Slack export attached to a client-facing recap. The "please find attached" without the attached. The reply-all that turned a private observation into a company-wide incident. These are not writing mistakes. Grammar-checkers cannot see them. Spell-checkers cannot see them. Only a pre-send check — an automated pass over the composed email a fraction of a second before it leaves your outbox — can catch them.
In 2026, pre-send checks are becoming the default expectation for professional email tooling, in the same way that seatbelts became the default expectation for cars. This guide breaks down the 15 checks that account for the vast majority of preventable email disasters — what each one does, why it matters, and how to implement it with Presend or roll your own.
Category 1: Attachment Checks
1. Missing Attachment Detection
The classic. The email body says "attached is the report" and there is no attachment. Presend scans your draft for phrases in over 30 languages — "attached," "enclosed," "see the file," "PFA," "the deck below," "screenshot" — and blocks send when the attachment picker is empty.
2. Wrong Attachment Warning
You dragged in `Q4-Financials-CONFIDENTIAL.xlsx` when you meant `Q4-Public-Summary.pdf`. Filename-based confidentiality tokens (`confidential`, `internal only`, `draft`, `do not share`) trigger a warning if the recipient list contains external domains.
3. Oversized Attachment Guard
Attachments over the recipient's likely inbox limit (usually 25MB for Gmail, 20MB for Outlook) fail silently on the send side. A pre-send check catches these before you assume the message was delivered.
Category 2: Recipient Checks
4. External-Recipient Warning on Internal Threads
You are on an internal thread called "engineering-leadership-2026-comp-review" and you type an address that resolves to an external domain. Pre-send blocks with a red interrupt, not a yellow warning. This one check has saved careers.
5. Recipient Autocomplete Ambiguity
You typed "Sara" and Gmail resolved to `sara.jenkins@vendor.com` instead of `sarah.jenkins@yourcompany.com`. When the two addresses only differ by one character, the pre-send layer asks "did you mean the internal Sarah?" with both contact cards side by side.
6. Reply-All Blast Guard
Threads with more than 10 recipients get a soft check on reply-all. If your message text uses second-person singular ("your latest slide," "your question about the API"), the tool nudges you toward a direct reply.
7. BCC Sanity Check
Two rules: (a) if BCC has more than 3 people, ask if the sender intends a broadcast; (b) if BCC + CC + To includes an address that is also in BCC, warn about the duplicate.
Category 3: Content Sensitivity Checks
8. Confidential Content Detection
Regex + LLM patterns for credit cards, SSNs, API keys (AWS, OpenAI, Stripe, GitHub tokens), passwords, and JWT tokens in the email body. Blocks send by default; user can override with a confirmation.
9. Financial-Terms Guardrail
Emails mentioning `salary`, `equity`, `ARR`, `MRR`, `runway`, `comp`, `valuation`, or `termsheet` to external domains prompt an extra confirmation. Not blocking — most of the time the send is legitimate — but forcing an intentional click.
10. PII Leak Detection
Names, addresses, phone numbers, and IDs pulled from your CRM that appear in the email body and match customers other than the recipient. Prevents the "I forwarded your account details to another customer" nightmare.
Category 4: Structural & Formatting Checks
11. Empty Subject Line
Gmail and Outlook both allow empty subjects. Recipients dislike them, filters penalize them, and follow-ups become impossible to find. Pre-send blocks empty subjects and suggests a subject based on the email body using a small LLM.
12. Broken Link Verification
A quick HEAD request against every URL in the body. Broken links are flagged with the specific 404 or DNS error. Optional: verify UTMs are attached to marketing links.
13. Merge-Tag Leftovers
If your template still contains `{{first_name}}` or `[NAME]` or `{customer}`, the pre-send check catches the un-rendered merge tag before your recipient reads a message addressed to "Hi {{first_name}}!"
14. Formatting Sanity
- Trailing quoted content bloat — Reply threads that have ballooned to 30+ nested quotes get a "trim history" suggestion.
- Inline images missing alt text — Accessibility warning that ships as a soft nudge.
- Signature duplication — Two signatures at the bottom trigger a warning.
Category 5: Tone & Intent Checks
15. Tone Match with Thread History
The full-body model reads the last 3 messages in the thread and rates whether your tone matches. A one-line "sounds good" reply to a client's serious escalation gets a warning: "The previous message tone is Formal/Concerned. Your reply reads as Casual. Continue?"
Building Your Own vs. Using Presend
You can build most of these checks yourself. A Gmail add-on or Outlook manifest can hook the compose event, run local regexes for the structural checks, and post the body to an LLM for tone and intent checks. Costs:
- Time — ~4-6 engineering weeks for the first 6 checks, another 8 weeks for the LLM-backed intelligence checks.
- Infra — LLM calls at ~$0.001-$0.01 per pre-send check, depending on your model.
- Ongoing — Model drift, regex maintenance, new attack patterns (fresh API key formats, new PII types).
Alternatively: install Presend. Every check on this list runs client-side or through the Presend API on every draft in Gmail and Outlook, with per-workspace policy configuration (block vs. warn vs. log), enterprise-grade audit logging, and no LLM cost passed to the user.
Rollout Strategy for Teams
For teams larger than 5, roll out pre-send checks in three phases:
1. Observe — Enable all checks in "log-only" mode for two weeks. Collect data on how often each check would have fired.
2. Warn — Turn the highest-value checks (missing attachment, external recipient on internal thread, confidential content) into soft warnings.
3. Block — Once the team is trained, promote the most damaging checks to hard blocks that require an explicit override.
This gradual approach avoids the "the tool is too annoying" backlash that kills most email-safety rollouts. By the time you promote a check to blocking, everyone on the team has seen enough real-world catches to know why the block exists.
The Bottom Line
Pre-send checks are not about paranoia. They are about acknowledging that human attention is finite, email is high-frequency, and a single miss can undo hours of work. Fifteen automated guardrails, running in a fraction of a second per send, quietly prevent the class of mistakes that historically make it to the front page of a company's post-mortems.
If you send more than 20 emails a day, pre-send checks pay for themselves the first time they catch a wrong recipient on a sensitive thread. If you send more than 100, they pay for themselves before lunch.