Few-shot prompting in real products
Few-shot means showing examples in the prompt. In AshnaAI, put examples in the agent spec or attach a sample file instead of pasting ten emails every turn.
AshnaAI
What you'll do
Few-shot means showing examples in the prompt. In AshnaAI, put examples in the agent spec or attach a sample file instead of pasting ten emails every turn.
This guide walks through few-shot prompting in AshnaAI. You stay in chat. You do not need a separate desktop of buttons for each file job.
Before you start
Open AshnaAI and start a new chat.
How to few-shot prompting
Follow the steps below in order. Confirm in chat before AshnaAI sends, posts, or writes to Tally.
Step 1. Read the definition
Few-shot means showing examples in the prompt. In AshnaAI, put examples in the agent spec or attach a sample file instead of pasting ten emails every turn.
Step 2. Open AshnaAI
Sign in at https://app.ashna.ai and start a chat, or open the agent builder.
Step 3. Try the job
Add two good and one bad example to the agent spec, then test.
Step 4. What to expect
Examples still count against context. Prefer one sample file.
Keep in mind
Examples still count against context. Prefer one sample file.
Related guides
Also read create and use agents.
Also read prompt playbook support ops agents.
More AI glossary how-tos live on What is RAG, hallucination, and prompt engineering.
Try it now
Open app.ashna.ai and run this job in a new chat. New accounts can start from app.ashna.ai/signup.
Frequently asked questions
- What is few-shot prompting?
- Few-shot means showing examples in the prompt. In AshnaAI, put examples in the agent spec or attach a sample file instead of pasting ten emails every turn.
- How does AshnaAI use this?
- Put the example in the spec.
- Do I need to train a model?
- Usually no. Prefer files, connectors, and a clear job over fine-tuning.
- Where is the glossary entry?
- Read https://www.ashna.ai/ai-glossary/in-context-learning
- What should I do next?
- Add two good and one bad example to the agent spec, then test.
Tags
Related
Try this in AshnaAI. Create a free account.
Found this article helpful? Share it with your network.