How to use AI to pitch a book to agents

Short answer. AI helps with the four mechanical parts of pitching agents, finding the right agents, building a comp set, drafting a personalized query and synopsis, and tracking submissions, but it cannot want your book the way you do, and an obviously AI-written query gets rejected fast. Use AI to do the research and the first drafts, then make every word sound like a writer who has read the agent's list. WriteLoom's Pitch studio is built around this exact sequence.

Step 1: Find agents who actually represent your book

The most common querying mistake is sending to agents who don't represent your genre. AI is genuinely good at this research because it's a filtering problem: given your genre, comps, and tone, which agents have a track record in this space and an open list? WriteLoom's AI agent and publisher search filters to agents whose recent deals match your book, which beats querying from a static spreadsheet that went stale two seasons ago. Agent interests change; a query to an agent who stopped taking fantasy last year is wasted.

Step 2: Build an honest comp set

Comps (comparable titles) tell an agent where your book sits on the shelf. A strong comp set is recent (roughly the last two to four years), genre-appropriate, and honest, not a list of mega-bestsellers you wish you'd written. AI can surface candidate comps you haven't read and explain why each one fits, which is faster than trawling Amazon also-boughts by hand. WriteLoom's comp curation does this inside your project, so the comps flow straight into the query. (We wrote a separate field guide on picking five good comps.)

Step 3: Draft the query and synopsis

A query letter has a tight job: hook, mini-synopsis, comps, bio, and a personalized line about why this agent. A synopsis is the whole plot, ending included, in one to two pages. Both are exactly the kind of structured, conventional writing AI drafts well. WriteLoom's Pitch studio drafts a personalized query and a synopsis from your project, so the first draft already knows your characters, comps, and bio.

But the first draft is a starting line, not a finish line. Here's what to do next.

Step 4: De-genericize everything

This is the step that separates queries that work from queries that get auto-rejected. Agents read hundreds of queries a week and recognize AI boilerplate instantly. Take the AI draft and:

• Rewrite the hook in your own voice; it should sound like the book.
• Replace any generic praise of the agent with a specific, true reason you're querying them, a client, a deal, an interview, a manuscript wish-list post.
• Cut anything that sounds like it could describe a hundred other books.
• Read it aloud; if it sounds like a press release, keep cutting.

Step 5: Track submissions

Querying is a numbers game spread over months, and the admin sinks people, who queried whom, when, what they sent, who replied. AI-adjacent tooling helps mostly by keeping this organized. WriteLoom's submission organizer tracks each agent, the date, the materials sent, and the response, so you don't accidentally re-query someone or lose track of a partial request.

What AI cannot do here

The summary: AI removes the research drudgery and the blank page, which is most of the friction. It does not, and should not, send a letter that sounds like a machine. Use it to get to a strong draft fast, then spend your human attention on the 20% that actually persuades an agent, the voice and the personalization.