How to become a contracted field partner for Appen, Scale AI, and Macgence — a practical guide for Indian agencies

The global AI training data market is worth over $2 billion annually and growing at 25% per year. Companies like Appen, Scale AI, Telus International, and Macgence spend a significant portion of this paying field agencies to collect real-world human behaviour data — walking, driving, shopping, working — to train AI models. India, with its diverse population, varied geographies, and relatively low field operation costs, is one of the most attractive data collection markets globally.

Here is how an Indian field agency gets into this pipeline.

Understanding what these platforms actually need

These are not research companies — they are logistics companies for data. They have contracts with AI companies (Google, Meta, autonomous vehicle manufacturers, surveillance technology firms) that need specific types of human behaviour data captured in specific ways. Your job as a field partner is to execute the data collection protocol they define, manage the field team, ensure quality, and deliver clean datasets on time.

The data types they most commonly need in India:

• Behavioural video data: People walking, sitting, using phones, driving, cooking — recorded in various environments to train activity recognition models
• Egocentric (first-person) data: Field agents wearing body cameras or smart glasses recording their daily activities from a first-person perspective — primarily for robotics and augmented reality training
• Speech and language data: Audio recordings in regional Indian languages — a massive and growing need
• Image annotation: Labelling objects in photographs and video frames for computer vision models
• Structured observation: Field agents systematically recording what they observe in specific environments — retail stores, roads, public spaces

What they require from a field agency partner

These platforms distinguish between individual crowdsourced workers and agency partners. As an agency, you are taking on larger contracts that require managing teams of 20–200 field workers across multiple geographies. The requirements are:

• Legal entity: A registered company with GST and PAN. They need to issue contracts and pay via bank transfer.
• Data privacy compliance: Understanding of GDPR principles even for India operations, since many end clients are European or American. You must be able to obtain informed consent from data subjects.
• Equipment capability: For video data, you need to source or rent appropriate recording equipment. For egocentric data, body cameras or compatible smartphones. For annotation, simply computers and trained annotators.
• Quality management: A demonstrated QA process — how you verify data quality, handle rejections, and maintain consistency across a distributed field team.
• Data security: How you store, transfer, and delete data. They will ask about your data handling procedures in detail.

The onboarding process — step by step

Each platform has its own process, but the general flow is consistent:

1. Supplier registration: Submit your company details, capabilities, and equipment list through their vendor portal or a dedicated agency application form.
2. Qualification assessment: They may send a test project — a small data collection task with specific requirements. Your ability to execute this correctly and on time is the real qualification test.
3. Contract negotiation: Once qualified, you negotiate a framework agreement covering rates, quality standards, rejection policies, and payment terms.
4. Pilot project: Before large-scale work, they assign a pilot — typically 50–500 data points. This is your chance to demonstrate operational capability.
5. Scale up: Successful pilots lead to ongoing contracts, often with increasing scale as trust builds.

What Blue Projects is building toward

We are currently in the application process with Appen, Scale AI, and Macgence. Our competitive advantage is our existing field infrastructure in Karnataka — trained supervisors, digital data collection protocols (ODK/KoBo), GPS-enabled field teams, and real-time quality monitoring systems built for our government survey contracts. The same infrastructure that runs an election survey can run a computer vision data collection assignment with relatively minor adaptation.

If your company has field capability and wants to explore data collection contracts for global AI platforms, we are open to discussing sub-agency arrangements or partnership models.