Your first chat

A chat is the fastest way to get something done in StickyPrompts: describe what you need, pick who answers, and send. This page takes you from an empty composer to a finished answer you can reuse, using a launch email for Northwind Analytics as the example.

app.stickyprompts.com Screen recording

Send your first message

1. Start a new chat

   Select New chat at the top of the sidebar, or just start typing in the composer on the home screen.

2. Write the prompt

   Say what you want, who it is for and what a good result looks like. The more context you give, the less you need to edit afterwards.

   Draft a launch email for Northwind Analytics, our new self-serve
   dashboard for mid-size retailers. Audience: existing customers.
   Give me 5 subject line options, a preheader, a short body with
   3 bullet points and a call to action. Warm, plain English.

   By default, Enter sends the message. If you prefer Enter to start a new line, switch on Add new line to text under On enter in your preferences; you then send with the Send button.

3. Choose the model

   The model button shows which model will answer. Select it to choose:

   • An auto-select mode: Best quality, Balanced or Fastest. StickyPrompts then picks the best model for the job for each message.
   • A model you recently used.
   • Choose model to open the full selector with every model you are allowed to use.

   Not sure? Keep your preferred model or use Best quality. See Choosing a model.

4. Set the reasoning effort

   The button next to the model shows the reasoning effort: Low, Medium or High. Higher effort lets a reasoning model think longer before it answers, which helps with analysis, planning and tricky logic. For a quick rewrite, Low is plenty. Models that do not reason ignore this setting.

5. Pick a style

   Style sets how the reply is shaped. Choose Prompty for concise, to-the-point answers, Chatty for more conversational ones, or a response style you created yourself. See Response styles.

6. Switch on snippets (optional)

   The puzzle button, just before Style, opens your snippets: short pieces of context about you, your company, a brand or a client. Tick the ones that fit, select Apply, and the AI knows those facts without you typing them into the prompt.

7. Attach a file (optional)

   Select Attach to add a file from your computer, for example last year’s launch email to match its tone. To let the AI search a whole set of documents instead, pick a knowledge base with the book icon.

8. Send

   Select Send. The answer streams in, and the chat appears at the top of your recent chats in the sidebar with a title generated from the conversation.

Read and reuse the answer

app.stickyprompts.com/conversations/0VxTk5Pc 1 Copy 2 Export raw text 3 Branch from here 4 Share the chat

Replies are formatted with headings, lists and tables, so they paste cleanly into email, documents and slides. Under each reply you’ll find a row of icons:

Icon	What it does
Copy (1)	Copies the reply to your clipboard.
RAW (2)	Exports the reply as raw text, without the formatting.
Branch (3)	Starts a new chat from this point. The original stays as it was, so you can try a different direction without losing it.

The header at the top of the chat shows the chat’s title, the model, the response style (here Prompty), Share (4) and Chat info. You can also move a chat into a project, so it sits with the rest of that work. To keep going, type in the Reply… box at the bottom. You can change the model, style or reasoning effort between messages.

Good habits

• Give the model a role and an audience. “You are writing for store managers who have 30 seconds” gets a better email than “write an email”.
• Ask for the format you need. Tables, numbered steps, a fixed word count, JSON: say so in the prompt.
• Save prompts that work. When you find yourself typing the same request again, save it to your prompt library with {{variables}} for the parts that change.
• Check facts that matter. For answers grounded in your own documents, use a knowledge base: replies then cite the file each claim came from.

Next steps