AI for Epistolary Fiction Perspectives

As of 2026-10-19 · the AZET team. We operate azet.io and offer this educational guide for fun and reference; the correspondence, characters, and writing exercises are entirely hypothetical.

A letter has two audiences: the person receiving it inside the story and the reader encountering it on the page. Those audiences rarely need the same explanation. The recipient may already know a shared history that the reader must infer from a careful omission.

For this exercise, epistolary fiction means a story presented through correspondence. Ask AI to help explore how the same incident changes when reported to different recipients. The goal is a set of purposeful documents, each written because someone wants something from someone else.

Give the letter a destination

Our hypothetical setting contains a community observatory and a dismantled brass telescope. An apprentice, Iva, has moved its parts into numbered trays. The instrument's owner, Sol, is away. A second apprentice, Bren, knows Iva was following an unsigned instruction slip.

Before drafting, specify three letters:

Sender and recipient	Immediate purpose	What the recipient already knows
Iva to Sol	Ask permission to continue assembly work	The telescope was difficult to turn
Iva to Bren	Ask who wrote the instruction slip	Bren was present when the trays arrived
Bren to Sol	Request that Iva keep her workshop place	Sol dislikes changes made without discussion

The incident remains the same. The rhetorical task changes. Iva might sound precise to Sol and impatient to Bren without becoming two unrelated characters. The difference follows from the relationship and the request.

These are invented motivations, not deductions about how real apprentices behave. Write them into the exercise so the model has a specific basis for its choices.

Compare openings with different obligations

Here are two authored hypothetical openings from Iva:

To Sol: “The turning ring is on tray four. I have kept every screw with its matching bracket. May I continue with the adjustment you mentioned before leaving?”

To Bren: “You saw the slip. Did those arrows look like Sol's writing to you? I have already filled four trays.”

The first opening foregrounds order and invokes an earlier conversation. The second foregrounds uncertain authority. Neither says everything Iva knows. The omissions become interesting because the recipient changes what she needs to establish.

One danger is making the private letter explain the whole situation for the reader: “As you know, we both work at the observatory owned by Sol.” Bren already knows that. If background information is necessary, attach it to a live disagreement or a request that makes mentioning it natural.

Treat delivery as part of perspective

For this illustrative sequence, Iva writes to Sol on Monday, writes to Bren on Tuesday, and receives Sol's reply on Friday. Bren's letter reaches Sol on Thursday. These dates are fictional author choices.

Sol's Friday reply can respond to Bren's appeal if you establish that it was read before the reply was written. Iva's Tuesday note cannot react to Friday's answer. More subtly, Sol might answer a concern Iva never expressed, making her realize that someone else has written.

Choose reading order separately from writing order. Showing Bren's appeal after Sol's reply can temporarily leave the reader uncertain about why Sol mentions Iva's workshop place. That is an intended information gap if the later document resolves it. It becomes muddle if the author has never decided which letters arrived.

A prompt for purposeful correspondence

Use the supplied hypothetical incident and correspondence table.
Draft two letters from Iva: one to Sol and one to Bren, 100 words each.
Keep the physical facts identical. Change emphasis according to recipient,
purpose, and shared knowledge. Do not make either writer omniscient.
After each letter, identify the request, one deliberate omission, and one
phrase that depends on the relationship. Use only the delivery timeline
provided. Flag missing information rather than inventing a received reply.

Review the commentary separately from the letters. An explanation that “Iva is hiding uncertainty” does not establish that the actual prose conveys it. Underline the exact phrases doing that work. If you cannot find them, revise the document rather than relying on the model's interpretation.

Make a human reader follow the exchange

Read only the letters, without the planning table. Mark where a pronoun lacks a clear referent, where a request disappears under exposition, and where a writer knows too much.

Then ask a reader to describe what each sender wants. Their answer need not match every private motive. It should identify enough immediate purpose to make the next letter matter. If all three letters merely summarize the telescope incident, give each a decision it is trying to influence.

FAQ

Must every letter have a different vocabulary?

No. People in the same workshop can share technical terms. Distinguish what they emphasize, how directly they ask, and what they assume needs defending. Constantly changing vocabulary can conceal a weak difference in purpose.

Can one letter be misleading?

Yes, if the misleading account is an authorial choice. Keep a private record of what happened and what the sender claims. Decide whether the sender is mistaken, withholding information, or deliberately making a false statement; those choices invite different later consequences.

Should I include every reply?

Only if it contributes to the reading experience. A missing reply can create pressure, but establish whether it was never sent, never received, or simply omitted from the presented sequence. You can withhold that answer from readers while knowing it yourself.

Product context

This is a general illustrative exercise, not a documented AZET capability. Our maker introduction described an AI assistant waitlist on October 3, 2026. That dated product baseline does not verify today's live status.

For the AZET waitlist, visit azet.io.