Conversation & Attachment Limits

Conversation & Attachment Limits

SlyReply is designed to give you fast, accurate AI replies. To keep responses high-quality and affordable, there are some built-in limits on how much context is sent to the AI for each reply.

Conversation History

SlyReply does not store your conversations. When you reply to an agent, the context it sees is whatever your email client quotes below your message — the "On, X wrote:" block with the >-prefixed lines. The agent works from your new message plus that in-band quoted history.

What this means for you:

• Use your email client's Reply button so the quoted history travels with your message — that's what gives the agent continuity.
• Very long threads carry a lot of quoted text. The AI focuses on your newest message; a few mobile clients trim the quote chain aggressively, in which case the agent sees only what you just wrote.
• Changing topic or starting a new question? Compose a new email rather than replying deep in an existing thread. This gives the AI a clean slate and produces better results.

Attachment Limits

SlyReply reads PDF and image attachments on Pro and Power accounts (free accounts get text-only conversations). Other document formats — Word (.docx), Excel (.xlsx), CSV (.csv) and similar — are not read yet: if you attach one, you'll get an explicit notice in the reply naming the file(s) that weren't processed, so a paying customer never silently loses an attachment. Server-side extraction for those formats is on the roadmap as a follow-up to issue #707. To keep processing fast and costs manageable, there are limits on what gets through. The size cap is the total of all attachments on a single email, not per file:

What happens when limits are exceeded:

There are two distinct cases, and they behave differently:

• Too many files (over the per-email count limit). If you attach more files than the limit, the agent still answers — it processes the first files up to the limit, in the order they appear in your email, and skips the rest. The reply you get back includes a --- Note: line at the bottom telling you exactly how many attachments were skipped and what the limit is, so you know to resend the remainder in a follow-up email. The skipped files are not silently lost without a trace — but do read that note, because the agent's answer only reflects the files it actually saw.
• Attachments over the total size cap. If the combined size of your attachments exceeds your plan's cap, the email is not answered at all. Instead, the account owner gets a short notice email explaining the email was over the size limit, with an upgrade link. Nothing is processed in this case — it's a clean reject, not a partial answer.

Tip: if a reply seems to ignore a document you sent, scroll to the > bottom and check for a skipped-attachments note. That note is the > signal that the agent didn't see everything you attached.

Tips for attachments:

• Compress images before attaching — a 1 MB photo gives the AI just as much to work with as a heavier file, and you stay well under the cap.
• Split large requests — if you have 6 images to analyse, send them across two emails.
• Use text when possible — pasting text into the email body is always more efficient than attaching a document.

Reply Length

Each AI reply has a maximum length set by the platform. This is typically enough for a detailed email response (several paragraphs). If you need a longer answer, try asking the AI to continue in a follow-up email.

These Limits Exist to Help You

These limits ensure that:

• Replies stay fast — less context means quicker AI processing
• Replies stay accurate — the AI focuses on what's relevant now, not messages from weeks ago
• Costs stay predictable — your per-reply pricing stays consistent regardless of how long the thread gets

If you have questions about these limits, contact support or check our other documentation pages.