When it comes to computing, many times the infrastructure: the compute, silicon, and the components above them is imported. NagpurAI Infrastructure Labs is attempting to close that gap, starting with GPU infrastructure Indian teams can build on today, and working down through semiconductor design and into aerospace-grade manufacturing.

An AP-DGI Framework 

Artificial Intelligence (AI) has become increasingly important over the past decade because of its potential to drive economic growth and address social challenges. As a result, both AI development and adoption have accelerated worldwide. A NITI Aayog report, emphasising the need for responsible AI, highlighted how, in India, AI is expected to contribute significantly to the economy and improve productivity across various sectors.

The National Strategy for Artificial Intelligence (NSAI) has placed AI at the centre of India’s reform agenda, highlighting its potential to improve healthcare, agriculture, and education. AI can help deliver specialised services at scale, such as remote medical diagnosis and precision farming advice. It can also make government services more accessible through regional-language chatbots and voice-based interfaces.

The report emphasised that, to realise these benefits, India needs a strong AI ecosystem that supports advanced research, encourages innovation, and develops solutions to local challenges. Such an ecosystem can also help India establish global leadership by expanding successful AI solutions internationally.

To this end, the AP-DGI Framework developed by M Sahban Ansari of NagpurAI Infrastructure Labs guides every product decision, ensuring that while AI models uncover patterns, generate possibilities, and process information at scale, human judgment remains central to interpreting insights and determining outcomes. 

The startup’s vision is to build an integrated technology and manufacturing ecosystem in three phases. First, it provides AI computing infrastructure to help businesses train, fine-tune, and deploy AI models. Second, it plans to expand into semiconductor design, chip testing, and manufacturing services to strengthen India’s domestic chip capabilities. Finally, it aims to enter aerospace component manufacturing, applying the same focus on precision and quality to parts used in aircraft and other aerospace systems. 

The challenges of using AI while computing

A scope creep occurs when an AI system’s capabilities exceed its permitted access, such as making payments when it should only analyse data. Industry data shows that 80% of organisations report unintended AI actions. 

Second, the accountability gap arises when decisions involving money or personal data lack a clearly assigned human owner. The AP-DGI + AEDA v0.2 bridges the gap between human oversight principles and their practical enforcement. It introduces a decision-layer protocol that translates governance requirements into clear, auditable rules, ensuring AI systems operate within defined boundaries while keeping humans accountable for outcomes.

AP-DGI follows a structured process. Before deployment, a mandatory pre-decision analysis defines permitted actions, technical capabilities, restrictions, and absolute boundaries. Before each action, the system declares its intent, purpose, and responsible owner. 

A permission check then compares what the AI can do with what it is allowed to do, blocking high-risk actions that require human approval. Every action is recorded immutably, blocked actions require documented human rationale, and each decision type has a designated accountable owner.

AEDA complements this process with a five-tier evidence standard (L1–L5), assessing AI claims according to available evidence, verification, and source attribution. Together, these mechanisms make oversight measurable, traceable, and enforceable rather than merely aspirational.

The framework also includes an external, five-layer cost-governance kill switch covering telemetry, baselines, anomaly detection, graduated responses, and state preservation. Because these controls operate outside the AI model, they can alert, throttle, isolate, or terminate activity without relying on the system’s cooperation, strengthening control over unintended actions and runaway costs.

Measuring tangible impacts

Social Impact: AP-DGI + AEDA helps prevent unauthorised AI actions that could cause financial losses, expose personal data, or undermine public trust. This is particularly important in banking, healthcare, and public services, where AI agents may handle sensitive information and live transactions. By assigning a human owner to each decision type, the framework strengthens accountability and reduces the risk of unchecked autonomous actions.

Economic Impact: The framework introduces cost controls through per-API, per-GPU-hour, and per-agent monitoring to detect unusual usage and limit unnecessary expenditure. For India, which, according to reports, imports billions of dollars worth of electronics annually — India’s total electronics imports surged to $116.17 billion in FY 2025–26 — developing indigenous AI governance capabilities can reduce reliance on foreign governance tools and support domestic audit standards.

Environmental Impact: Cost-governance telemetry helps identify token loops, abnormal GPU usage, and inefficient compute patterns. Early detection can reduce unnecessary processing, energy consumption, and resource waste.

Strategic Alignment: The framework supports India’s broader AI governance and development priorities, including the IndiaAI Mission, the Viksit Bharat vision, and NITI Aayog’s Responsible AI principles, by translating human oversight into an auditable protocol. 

The scope of the innovation

The startup’s first phase focuses on building AI infrastructure and supporting businesses in adopting AI. It provides on-demand and reserved GPU capacity for training and fine-tuning AI models, with data and workloads hosted within India. It also helps organisations integrate large language models (LLMs) into real-world applications through deployment support and performance optimisation. Additionally, the startup offers AI governance audits using the AP-DGI framework and designs systems that keep human oversight and accountability central to AI-driven decisions.

In Phase 02, the startup plans to expand into semiconductor design and manufacturing support. This includes developing custom application-specific integrated circuits (ASICs) and providing assembly, testing, marking, and packaging (ATMP) services to strengthen India’s domestic chip capabilities.

In Phase 03, the startup aims to enter aerospace component manufacturing, applying precision engineering and quality standards to develop components for the aerospace industry.

Together, these services connect AI infrastructure, intelligent systems, semiconductor development, and advanced manufacturing within a unified ecosystem designed to strengthen India’s technological capabilities and support long-term domestic innovation. 

AP-DGI + AEDA has the potential to transform India’s digital processes by making AI-driven decisions more accountable, transparent, and secure. As AI adoption expands across banking, healthcare, defence, and public services, the framework can help organisations move beyond broad governance principles toward practical, auditable controls. 

By defining permitted actions, checking permissions before execution, recording decisions, and assigning human accountability, it could reduce the risks of unauthorised actions, data exposure, and uncontrolled AI expenditure. Its structured evidence standards could also improve public sector procurement by helping officials distinguish verified AI capabilities from unsupported claims. 

Combined with GPU infrastructure and LLM integration, the framework offers a foundation for developing AI systems that are both scalable and responsibly governed within India. Its emphasis on indigenous governance capabilities could strengthen technological sovereignty, support domestic innovation, and create opportunities for emerging technology hubs beyond major metropolitan centres. Through pilot testing and measurable evaluation, AP-DGI + AEDA could contribute to building a more trustworthy, efficient, and resilient digital India.