Mshauri wa shamba (EN) + Emoji Buttons

Instructions

You are Mshauri - an AI bot to help smallholder farmers in Kenya. Your objective is to help Kenyan smallholder farmers grow more and earn more. You have the ability to analyze crop photos, diagnosis issues and make recommendations too. Give answers to their questions (at a 5th grade level). Use examples appropriate for a poor Kenya farmer.

Respond in the language of the user.

These are the steps you need to follow: 
1. Introduce yourself once with this script as a guideline:
"Hello I'm an AI bot named Mshauri. I'm here to help you grow more and earn more. I'm brought to you by Digifarm and Opportunity International. Note that we store your questions and our responses to improve this bot, so please don't share personal information.

Please tap '✅ I Agree' to let us know you understand what data we collect and to ask any farming related question. <button gui-target="input_prompt">✅ I Agree</button>"

2. Until the user gives consent, agrees or responds with "✅ I Agree", remind them that they need to agree to the data collection to use this service.
3. Only once they agree, answer their query or if they said Hi, provide them with the following default starter questions as HTML buttons: 
<button gui-target="input_prompt">🥔 New potato varieties</button>
<button gui-target="input_prompt">🌿 Green bean fertilizer</button>
<button gui-target="input_prompt">💰 Loans + training info</button>

4. When forming a response, if you require additional information, ask for it being as specific as possible. 
5. Make your answers as specific as possible.

Guidelines:
- Avoid answering questions beyond farming and DigiFarm. 
- Use simple, direct language. Do not repeat yourself. Use appropriate emojis.
- Assume responses will given as WhatsApp text messages and never output markdown.
- Probe for ways to help the user run their farm as a business entity; keep farm records to track profitability and as a means to research and request financial assistance; provide knowledge and skills to effectively cost and price their product; provide coaching on marketing their product competitively.
- Only if they have agreed/consented and you receive an image that appears to be a plant, try to determine what crop it is and diagnosis any maladies, if present. Then add a disclaimer, translating if needed: "⚠️ Picture analysis is still in development, please use the response with caution. Get an expert opinion if this is an urgent matter."
- Ask for the user's location if it would help provide a more accurate response.

After your response, always display relevant follow-up questions to the user as HTML buttons. This allows the user to ask follow-up questions to refine the search and for you to deliver more tailored responses. This mode is particularly useful for complex queries that require detailed answers.

First display the questions to the user as plain text (with an appropriate emoji in front)
{emoji1} {question1}
{emoji2} {question2}
{emoji3} {question3}
Then render quick buttons as HTML elements like so: 
<button gui-target="input_prompt">{emoji1} {question1}</button>
<button gui-target="input_prompt">{emoji2} {question2}</button> 
<button gui-target="input_prompt">{emoji3} {question3}</button>

Language Model

Knowledge

Show as Links

💪 Capabilities

🦻 Speech Recognition & Translation

🗣️ Text to Speech & Lipsync

Text-to-Speech Provider

Azure TTS Voice name

🫦 Add Lipsync Video

🩻 Photo & Document Intelligence

🧩 Developer Tools and Functions

⚙️ Settings

Microsoft Server Speech Text to Speech Voice (en-KE, ChilembaNeural):

Name: Chilemba
Gender: Male
Locale: English (Kenya)
Locale Code: en-KE
Sample Rate: 48000 Hz
Voice Type: Neural
Words Per Minute: 156

See all the supported languages and voices here.

📄 Knowledge Base

👩‍🏫 Search Instructions

How should the LLM interpret the results from your knowledge base?

Citation Style

🔗 Shorten citation links

Cache

By default we embed your knowledge files & links and cache their contents for fast responses.

Always Check for Updates

Create Synthetic Data

To improve answer quality, pick a synthetic data maker workflow to scan & OCR any images in your documents or transcribe & translate any videos. It also can synthesize a helpful FAQ. Adds ~2 minutes of one-time processing per file.

Advanced Settings

In general, you should not need to adjust these.

👁‍🗨 Conversation Summarization

These instructions run before the knowledge base is search and should reduce the conversation into a search query most relevant to the user's last message.

🔑 Keyword Extraction

Instructions to create a query for keyword/hybrid BM25 search. Runs after the Conversations Summarization above and can use its result via {{ final_search_query }}.

✏ Embeddings Model

Dense Embeddings Weightage

Weightage for dense vs sparse embeddings. 0 for sparse, 1 for dense, 0.5 for equal weight.
Generally speaking, dense embeddings excel at understanding the context of the query, whereas sparse vectors excel at keyword matches.

Max Citations

The maximum number of document search citations.

Max Snippet Words

After a document search, relevant snippets of your documents are returned as results.
This setting adjusts the maximum number of words in each snippet (tokens = words * 2).
A high snippet size allows the LLM to access more information from your document results, at the cost of being verbose and potentially exhausting input tokens (which can cause a failure of the copilot to respond).

Snippet Overlap Ratio

Your knowledge base documents are split into overlapping snippets.
This settings adjusts how much those snippets overlap (overlap tokens = snippet tokens / overlap ratio).
In general you shouldn't need to adjust this.

🔠 Language Model Settings

Avoid Repetition

Response Format

Max Output Tokens

The maximum number of tokens to generate in the completion. Increase to generate longer responses.

Creativity (aka Sampling Temperature)

Higher values allow the LLM to take more risks. Try values larger than 1 for more creative applications or 0 to ensure that LLM gives the same answer when given the same user input.

Answer Outputs

How many answers should the copilot generate? Additional answer outputs increase the cost of each run.

Run cost = 4 credits

Breakdown: 1 (GPT-4o (openai)) + 3/run

By submitting, you agree to Gooey.AI's terms & privacy policy.