Skip to main content

AI Assistant

Eyesotope includes Tako, a built-in AI assistant that can analyze your farm data, answer questions, generate reports, and provide actionable recommendations — all in a real-time streaming chat interface.

Accessing the AI Assistant

  1. Navigate to AI in the left sidebar
  2. The chat interface opens with a conversation panel on the left and the chat area on the right
  3. Click New to start a fresh conversation, or select a previous one from the sidebar

What Can Tako Do?

Farm Data Analysis

Query your fields, inventory, equipment, operations, contracts, deliveries, and partner data using natural language.

Smart Recommendations

Get suggestions for crop management, optimal planting times, resource allocation, and operational improvements.

Report Generation

Generate summaries and reports about your farm operations, seasons, and financial data.

Farm Tools

Tako has access to specialized tools that query your farm database in real-time — fields, seasons, equipment, inventory, weather, and more.

File Analysis

Upload images, PDFs, spreadsheets, and text files. Tako can analyze documents, extract data from photos, and process CSV files.

Web Research

Enable web research to let Tako search the internet for up-to-date agricultural information, market prices, and best practices.

Chat Interface

The chat interface features a modern two-row input design:
  • Top row — Text input area (auto-expanding, supports multiline)
  • Bottom row — Toolbar with action buttons:
    • 🌐 Web — Toggle web research on/off
    • ✨ Pro — Toggle deep reasoning mode
    • 📎 Attach — Upload files (images, PDFs, text, CSV)
    • ➤ Send — Send your message

Real-Time Streaming

Responses stream in real-time, token by token, with a phased status indicator:
  1. Thinking… — The AI is processing your request
  2. Calling tools… — The AI is querying your farm data (with a live tool count)
  3. Processing results… — Tool results are being analyzed
  4. Generating response — The answer is streaming in with a smooth fade-in effect
After the response completes, a meta row shows the model used, token count, duration, and number of tools called. Click the Debug button to expand detailed diagnostics.

@Mentions — Tag Your Farm Data

Type @ anywhere in the message to open the mention picker. You can tag specific farm entities to give Tako precise context:
  • @Fields — Tag a specific field by name
  • @Contracts — Reference a contract
  • @Equipment — Reference equipment
  • @Deliveries — Tag a delivery
  • @Inputs — Reference an inventory item
  • @Seeds — Reference a seed variety
  • @Seasons — Tag a season
  • @Partners — Reference a partner
Tagged entities are passed as structured context to the AI, enabling more precise and relevant answers.

File Attachments

Attach files to your messages for analysis:
  • Supported formats — JPEG, PNG, WebP, GIF, PDF, plain text (.txt), CSV
  • Limits — Up to 3 files per message, 10 MB each
  • Upload methods:
    • Click the 📎 Attach button
    • Drag and drop files onto the chat area
    • Paste images from your clipboard (Ctrl+V / Cmd+V)
When images are attached, Tako uses a specialized vision model to analyze photos, extract text via OCR, and identify objects. PDFs and text files are automatically parsed and their content is included in the conversation.
Upload a photo of a soil analysis report or a pest-affected plant and ask Tako to identify and recommend treatment.

Web Research

Toggle the 🌐 Web button to enable web research. When active, Tako searches the internet before responding, providing:
  • Current market prices and agricultural news
  • Up-to-date pest and disease management practices
  • Weather forecasts and climate advisories
  • Regulatory and compliance information
Web research results are automatically integrated into the AI’s response as additional context.

Deep Reasoning (Pro Mode)

Toggle the ✨ Pro button to activate deep reasoning mode. This triggers a two-stage analysis pipeline:
  1. Stage 1 — The standard AI agent gathers your farm data using tools
  2. Stage 2 — A specialized reasoning model performs deeper analysis on the gathered data
Deep reasoning is ideal for complex questions like:
  • “Analyze the profitability trend of my wheat fields over the last 3 seasons”
  • “What are the risk factors for my current crop plan?”
  • “Compare my operational efficiency with best practices”
Deep reasoning takes longer but provides significantly more thorough analysis. The reasoning model and tokens used are shown in the debug panel after completion.

Conversation History

All conversations are automatically saved and accessible from the sidebar:
  • Sidebar — Toggle the conversation list with the menu button
  • Search — Find previous conversations quickly
  • Delete — Remove conversations you no longer need
  • Resume — Click any conversation to continue where you left off

Example Questions

  • “What is the total area of all my wheat fields?”
  • “Show me the operations planned for this season”
  • “Which equipment has been used the most this month?”
  • “What’s my current inventory of fertilizers?”
  • “Summarize the deliveries for the past quarter”
  • “Analyze the NDVI trends for @Field Name over the last 6 months”
  • “What’s the weather forecast for my farm this week?” (with web research enabled)
  • “Review this soil analysis report” (with PDF attached)

AI Credits

The AI assistant uses a credit-based system:
  • Each farm has an AI credit balance
  • Credits are consumed per interaction (based on tokens used)
  • Your current balance is shown in the user dropdown (top-right corner)
  • Credit balance is color-coded:
    • Green — More than 100 credits remaining
    • Yellow — 20–100 credits remaining
    • Red — Fewer than 20 credits remaining
Deep reasoning mode and web research consume more credits per interaction. Monitor your balance in the user dropdown. Contact your farm administrator to top up credits when running low.

Debug Panel

After each response, click Debug to see detailed information:
FieldDescription
ProviderThe AI provider used (e.g., Nebius AI Studio)
ModelThe primary model and its ID
TokensInput, output, and total token counts
DurationResponse time in seconds
Tool CallsNumber of tool invocations
ToolsList of tools used (e.g., get_fields, get_weather)
Web ResearchWhether web search was used
AttachmentsNumber of files attached
Vision ModelThe vision model used for image analysis (if applicable)
Reasoning ModelThe deep reasoning model used (if Pro mode was active)
Reasoning TokensToken consumption for the reasoning stage
A detailed Tool Log shows each tool call with its success/failure status, latency, and output size.

Access

The AI module is typically available to Admins only by default, but farm administrators can grant AI access to any member from the Teams page under module access settings.