Agricultural Landscape
🌱 Seed Stage — Building the Future of Farming

The AI Agronomist for 120 Million Farmers

Designed around Claude by Anthropic — AgriSense is an early-stage concept being designed to deliver crop disease diagnosis, market pricing, and localised agronomic advice directly to smartphones in Southeast Asia, in the farmer's own dialect. AgriSense is a product of NhieuTruyen.

Request Early Access
Claude-first MVP · Anthropic API (planned)

The Problem Is Enormous. The Window Is Now.

Southeast Asia loses an estimated $50 billion annually to preventable crop disease and poor agronomic decisions — largely because many smallholder farmers have limited access to timely expert guidance. AgriSense is being designed to help close that advisory gap.

$4.7B

Total addressable market for precision agri-AI in Southeast Asia by 2030 (Source: McKinsey Global Institute, 2024).

14.2% CAGR 2024–2030
Pre-Seed Funding Stage
eco

SDG 2 Aligned

The planned pilot will test whether faster, easier crop advice can help smallholders reduce crop loss and support food security.

120M+

Smallholder farmers in Southeast Asia, majority without access to professional agronomic advice — the direct addressable audience for AgriSense. (FAO, 2023)

Designed for
Claude.

visibility

Multimodal Crop Diagnosis

Claude's image and text reasoning will be evaluated on crop photos, symptom descriptions, and transcribed voice notes. AgriSense plans to use this capability to validate crop diagnosis workflows across priority crops such as rice, maize, cassava, and sugarcane.

travel_explore

Source-Backed Guidance — Designed for Review

AgriSense plans to pair Claude with retrieval from vetted agronomic references and market data, with citations and agronomist review.

language

Localized Dialect Support

AgriSense plans to use Google Cloud Speech-to-Text and agricultural vocabulary prompts to support local farming language, starting with Vietnamese dialects before broader regional expansion.

AI Interface
Product Walkthrough

How AgriSense Works

A concept walkthrough for the planned MVP: capture a crop issue, send the crop photo and text or voice transcript to Claude, receive structured guidance, then validate the result with real users during pilot testing.

1
photo_camera

Capture

Farmer opens the planned AgriSense app and takes a photo of the affected crop — or records a voice message describing the symptoms in their local dialect.

Planned: offline capture · sync when connected
2
psychology

AI Diagnosis

Claude would interpret the image, symptom description, voice transcript, and basic location context. A separate retrieval step would supply selected agronomic references for review.

Target: fast field response
3
clinical_notes

Treatment Plan

The farmer would receive a structured draft answer: likely disease, confidence level, suggested next step, estimated input cost, and sources to review.

Planned: source citations
4
monitoring

Track & Follow-up

The planned product would let farmers log treatment progress over time, upload follow-up photos, and build a simple crop health history.

Planned: crop health history
terminal Illustrative AI Diagnosis Output ILLUSTRATIVE 
{
  "diagnosis": {
    "disease": "Rice Blast (Magnaporthe oryzae)",
    "confidence": 0.94,
    "severity": "moderate",
    "affected_area_pct": 35
  },
  "treatment": {
    "primary": "Apply Tricyclazole 75% WP at 0.6g/L",
    "dosage": "300L spray solution per hectare",
    "application_method": "Foliar spray at tillering stage",
    "estimated_cost_usd": 12.50,
    "alternative": "Isoprothiolane 40% EC at 1.5mL/L"
  },
  "sources": [
    "IRRI Rice Knowledge Bank — Rice Blast Management",
    "Local pest advisory source — placeholder",
    "FAO Crop Protection Compendium"
  ],
  "nearby_input_dealers": [
    { "name": "Vật Tư Nông Nghiệp Mekong", "distance_km": 3.2 },
    { "name": "AgriMart Cần Thơ", "distance_km": 7.8 }
  ],
  "follow_up": "Re-photograph affected area in 7 days for recovery assessment"
}

Illustrative Use Cases

AgriSense is designed for everyday farming challenges. The examples below are illustrative scenarios for the planned MVP, not completed pilot results.

grass
bug_report

Disease Detection

Rice Blast Outbreak — Mekong Delta

A farmer in Cần Thơ notices yellowing leaf tips on a rice paddy and photographs the affected plants with the planned AgriSense workflow.

schedule Pilot metric: time-to-guidance
check_circle Example output: possible rice blast with confidence score
savings Validation goal: estimate avoided crop loss

The pilot will test whether earlier guidance can reduce waiting time before a farmer speaks to an extension officer or agronomist.

store
price_check

Market Intelligence

Harvest Timing — Cassava Price Optimisation

A cassava farmer in Tây Ninh could ask by voice: "Should I harvest now or wait two more weeks?"

trending_up Example advice: compare current and expected prices
database Data source: Vietnam Commodity Exchange + weather
savings Validation goal: estimate income impact per hectare

This workflow depends on future integration with commodity price and weather data sources.

water_drop
water_drop

Input Optimisation

Fertiliser Over-application — Maize in Thailand

A maize cooperative in Thailand could use AgriSense to review leaf colour photos and input records for possible fertiliser overuse.

analytics Example finding: possible over-application pattern
eco Validation goal: reduce unnecessary input cost
public Future metric: estimate avoided emissions

A future cooperative dashboard could support bulk review and seasonal planning for member farms.

Execution Roadmap 2026–2027

Planned milestones for moving from concept to MVP, then to a small field pilot.

1

Now — Concept & Research

Problem and Architecture Definition

Define the farmer workflow, Claude-first architecture, candidate public datasets, and the validation plan for a Claude-powered MVP.

2

Next — MVP Build

Lightweight Claude Demo

Build a minimal workflow for crop photo upload, voice/text symptom capture, grounded response generation, and basic farmer feedback.

3

Pilot — Field Validation

300–500 Farmer Target

Test usefulness, accuracy, response quality, retention, and cost per diagnosis with farmers, cooperatives, and agronomist review.

4

2027 — Exploration

Regional Exploration — Thailand & Indonesia

Use Vietnam pilot learnings to decide whether regional localisation for Thailand and Indonesia should be pursued.

Pilot Plan & Validation

What We Plan to Validate

AgriSense is pre-product and pre-revenue. The next milestone is a small MVP pilot with farmers, cooperatives, and agronomist reviewers.

300–500

Pilot Farmer Target

To validate after MVP build

MVP

Current Build Goal

Photo + voice + grounded answer

TBD

Diagnosis Accuracy

To be measured with agronomists

$

Cost per Diagnosis

Estimated after real usage

insights Validation Questions

  • arrow_right Can farmers capture useful crop photos and voice notes in real field conditions?
  • arrow_right Do agronomists agree with the top diagnosis and treatment guidance often enough for safe first-line support?
  • arrow_right Does voice-first interaction improve usability for farmers who do not want to type?
  • arrow_right What is the real infrastructure cost per diagnosis after measuring Claude API, hosting, storage, and sync usage?
  • arrow_right Which features are useful enough to justify a future paid tier or cooperative dashboard?

science Pilot Evidence to Collect

  • checklistAgronomist review notes for sampled diagnoses.
  • checklistFarmer feedback on clarity, trust, and ease of use.
  • checklistLatency, cost, and error logs from actual Claude API and hosting usage.
  • checklistBefore/after examples of follow-up photos where available.

Integration Ecosystem & Partners

AgriSense is designed to plug into the existing agricultural value chain — not replace it.

account_balance

Government Agricultural Departments

Subsidised access programmes for extension services

groups

Agricultural Cooperatives

Bulk diagnosis API for member farms with seasonal reporting

assured_workload

Agri-Insurance Companies

Risk assessment data feed for crop insurance underwriting

storefront

Input Suppliers & Dealers

Marketplace integration for treatment product purchases

Frequently Asked Questions

Common questions from startup programme reviewers, partners, and early pilot collaborators.

How accurate is AgriSense's crop disease diagnosis? expand_more
AgriSense has not yet completed a formal field accuracy study. Accuracy is a core pilot metric: each sampled diagnosis will be reviewed against agronomist feedback and trusted crop protection references before we claim production-level performance.
Does it work without internet access? expand_more
Offline-first support is part of the planned MVP design. The first version will prioritise offline capture and later sync; Claude-based analysis will run once the device reconnects.
Which crops and diseases are supported? expand_more
The planned coverage roadmap starts with common rice, maize, cassava, and sugarcane issues in Southeast Asia. The 240+ condition list is a target taxonomy built from public references and must be validated before being described as supported production coverage.
What languages and dialects are supported? expand_more
AgriSense is planned as a voice-first product. The initial MVP will focus on Vietnamese farmer language, then test broader regional dialect support using Google Cloud Speech-to-Text and agricultural vocabulary tuning.
How does the revenue model work? expand_more
AgriSense is currently pre-revenue. The revenue model is a hypothesis to validate after the pilot: free basic access for farmers, possible low-cost premium features, cooperative dashboards, and later partner API or public-sector programmes. Unit economics will be measured from actual Claude API and hosting usage.
How is this different from existing agri-tech apps? expand_more
AgriSense's intended differentiation is to combine a smartphone-first workflow, crop photos, voice input, Claude-powered image and language reasoning, vetted reference retrieval, and local agricultural context for smallholder farmers. This positioning still needs to be validated through MVP testing.
What technology will power AgriSense? expand_more
The planned architecture is Claude-first: Anthropic Claude API for image analysis, reasoning, and structured guidance; vetted source retrieval for references; and supporting services for speech transcription, maps, hosting, and storage. Integrations and costs will be validated during MVP testing.
What is the current funding status? expand_more
AgriSense is currently bootstrapped and pre-product. The immediate goal is to secure cloud/startup programme support, build the MVP, and prepare a small validation pilot before any larger fundraising plan.

Ready to Back the Future of Food Security?

AgriSense is seeking Anthropic API support, technical guidance, and early collaborators to build a Claude-first MVP and validate the concept with farmers.

Register below to receive the concept brief, technical architecture, and pilot plan.

For startup programme reviewers, mentors, strategic collaborators, and early validation partners. We will respond within 48 hours.