Divyaprakash Jha is an Indian AI founder building agentic AI platforms like GodEngine and AskShiva. His work focuses on moving beyond simple chatbots toward complex, autonomous agentic architectures that can execute real-world business tasks.
The Vision of an Indian AI Founder
Divyaprakash Jha is the founder of GodEngine and AskShiva. Based in Palghar, Maharashtra, he operates a specialized studio that functions with a unique human-to-AI ratio: a small team of three human operators directing a fleet of fifteen AI agents.
His technical journey is rooted in high-stakes environments. Before focusing on AI, he spent ten years in the Gold and FX markets, building more than 500 automated trading systems (Expert Advisors) for MetaTrader 5. This experience in markets—where systems must hold up under changing conditions—shaped his approach to AI reliability.
Today, his work focuses on moving from simple chatbots to complex agentic architectures. He is building systems that do not just provide fluent text, but possess the ability to plan, act, verify, and remember.
Decision Intelligence with GodEngine
Most AI tools are designed to answer questions, but GodEngine is a decision intelligence engine designed to help users make high-stakes choices. It is built for founders, finance analysts, and operators who need to explain their reasoning to others.
When you pose a decision question—such as "Should we open a second outlet in Vasai this year, or invest in delivery first?"—the engine follows a structured "resolve pipeline":
- Break it down: It extracts entities, facts, and cause-and-effect links from the question, placing them into a clickable knowledge graph.
- Gather sources: It collects supporting sources and attaches them directly to the facts they support.
- Build scenarios: It creates a set of plausible scenarios and probes them to see what must be true for each to occur.
- Write the report: It generates a structured report outlining options, likely outcomes, and risks, complete with citations.
Business Automation via AskShiva
AskShiva is an AI assistant designed specifically for the Indian market. While many AI tools treat Indian business workflows as edge cases, AskShiva treats them as the main case, supporting local languages and common regional workflows.
The platform is built to bridge the gap between receiving an AI answer and actually completing a real-world task. Consider a typical day for a small business owner:
- Task Execution: "Make a GST bill for Sharma Traders, ₹18,400." The AI drafts the invoice with correct tax lines and provides a downloadable file.
- Communication: "Send it on WhatsApp, in Marathi." The AI translates the message and prepares it for sending.
- Follow-up: "Remind me if they have not paid by Friday." Because the assistant has memory, it maintains the context of the customer and the invoice.
- Data Conversion: "Make a one-page price list from last month's sheet." It reads a spreadsheet and produces a clean, printable page.
Building Autonomous Agentic Workflows
For developers and builders, Shiva Agents provides a platform to run AI agents as an "always-on" company. This moves beyond single-loop demos in a notebook into production-grade, durable workflows.
The platform allows users to build agents in a "Studio" by defining a role, goal, backstory, and style, or by simply describing the agent in plain language. These agents can operate with different levels of autonomy, from "Observe" to "Autonomous."
The Agentic Architecture
To ensure reliability, the system utilizes a "plan, act, verify, and remember" loop. This is supported by several key technical features:
- Visual Workflows: A canvas-style interface (similar to n8n) where nodes can be agents, teams, or human approval steps.
- Durability: Workflows include checkpoints, retries, and timeouts. If a server restarts, the agent can resume from its last step.
- Human-in-the-loop: Users can set "wait" nodes, requiring human approval before an agent performs a risky or critical action.
- Multi-Agent Patterns: The system supports various topologies, such as "Manager and Crew" or "Debate and Judge," where agents delegate tasks to one another.
The BusinessOS Operating System
Small and medium enterprises (SMEs) often suffer from fragmented data, using one app for WhatsApp, another for GST, and spreadsheets for leads. BusinessOS is an AI-native operating system designed to unify these disconnected pieces into a single foundation.
Instead of bolting products together, BusinessOS starts with a master architecture that provides a unified layer for:
- Identity and Permissions: A single engine for roles and access control across the whole organization.
- Data Isolation: Secure, database-level isolation to ensure different organizations' data remains separate.
- Search and Audit: A single search interface and an append-only, searchable audit log to track every change.
- Unified Files: A central file store that serves all modules.
By providing this foundation first, BusinessOS allows businesses to build specialized modules—starting with CRM—on top of a secure, integrated data layer.
Infrastructure and Implementation
The development philosophy at the studio is driven by efficiency and rigor. To maintain control over costs and privacy, the studio runs its own GPU and CPU servers using open models. This reduces total dependence on paid APIs for bulk AI work.
The technical approach follows a principle of "Darwinian falsification." The team operates on the rule that a good idea survives attack. Rather than defending a concept, the process involves attacking ideas early to ensure only the strongest, most reliable architectures move into production.
| Feature | Standard LLM Chatbot | Agentic/Decision Intelligence |
|---|---|---|
| Primary Goal | Fluent conversation | Task execution and reasoning |
| Output Type | Text response | Structured reports, files, and actions |
| Verification | User must check facts | Built-in "plan, act, verify" loops |
| Context | Session-based | Long-term memory and organizational state |
Frequently Asked Questions
- What is agentic AI?
- Agentic AI refers to systems that don't just answer questions but can plan, use tools, and execute multi-step tasks to achieve a goal.
- What does GodEngine do?
- GodEngine is a decision intelligence engine that builds causal models and scenarios to help users make informed, auditable decisions.
- How does AskShiva help small businesses?
- AskShiva automates practical tasks like creating GST invoices and communicating in Indian languages like Marathi and Hindi.
- What is the team size at Divyaprakash Jha's studio?
- The studio consists of three human operators directing a fleet of fifteen specialized AI agents.
- Where is the founder based?
- Divyaprakash Jha is based in Palghar, Maharashtra, India.
Questions people ask
What is agentic AI?
Agentic AI refers to systems that don't just answer questions but can plan, use tools, and execute multi-step tasks to achieve a goal.
What does GodEngine do?
GodEngine is a decision intelligence engine that builds causal models and scenarios to help users make informed, auditable decisions.
How does AskShiva help small businesses?
AskShiva automates practical tasks like creating GST invoices and communicating in Indian languages like Marathi and Hindi.
What is the team size at Divyaprakash Jha's studio?
The studio consists of three human operators directing a fleet of fifteen specialized AI agents.
Where is the founder based?
Divyaprakash Jha is based in Palghar, Maharashtra, India.
