Start with the work
We begin with the decision, workflow, and outcome—not a model looking for a use case.
About BodhiContextAI
Research-driven AI work, proven by outcomes.
We build a cloud-native Context OS for Telecom and Retail, and run an outcome-focused bootcamp that makes engineers production-ready—for enterprise agent systems, Forward Deployed Engineering roles, and beyond. Research in context engineering, retrieval, and agent evaluation only becomes product after it is validated against those outcomes.
AI becomes useful when it understands both contexts.
BodhiContextAI began with a simple observation: an enterprise can have excellent models and still fail to produce dependable outcomes. The gap always sits in the same two places.
The people building AI need production engineering context—architecture, evaluation, operations, security—to ship a system that survives real traffic instead of stalling after the demo. The AI systems themselves need enterprise context—entities, relationships, policies, real-time state—to answer correctly instead of confidently guessing.
That's why we build both, not one: our bootcamp turns engineers into people who can ship the first kind of context, and our context layer gives their agents the second—so the same production standard holds on both sides of the handoff. The payoff shows up in production: capstones that ship instead of stall, and agents that cite the order or policy behind every answer instead of guessing at one.
Founder
Amit Kayal
Amit spent two decades building enterprise applications inside manufacturing plants, telecom OSS/BSS stacks, and retail commerce systems—close enough to the work to know where the exceptions live and why another dashboard never fixed them. That history is why BodhiContextAI starts with these three domains: the pain points are already understood, and agentic AI is finally capable enough to turn that operational knowledge into faster, more automated workflows.
He now builds AI-first, cloud-native enterprise platforms and turns complex systems into products people can understand and trust—spanning agentic AI, enterprise architecture, and AWS-native systems, including production agent fleets, model-routing cost optimization, and AI-led workflow automation.
At BodhiContextAI, he leads the product and learning direction—keeping every engagement close to a real enterprise workflow and a measurable outcome.
Connect with Amit on LinkedIn ↗
Co-founder
Aditeya Kayal
Aditeya is co-founding BodhiContextAI while pursuing Electronics and Instrumentation at BITS Pilani—the same campus where two students once built a robot for the college cafeteria and turned it into GreyOrange, a global robotics company. That builder-first culture, backed by one of Asia's oldest student-run E-Cells and an incubator that has spun out 300+ startups, is the training ground for tackling a genuinely enterprise problem: the disconnected context trapped inside manufacturing, telecom, and retail systems. His interests in AI agents, robotics, hyperautomation, and cloud computing point straight at that gap, and he puts them to work in Python and data science, turning context-engineering research into product.
Connect with Aditeya on LinkedIn ↗
Advisor & mentor
Avishek Das
Avishek is an AI Quality Architect working in agentic AI, LLM and RAG evaluation, AI reliability, and LLM observability on AWS. He advises BodhiContextAI on exactly that: guardrails, hallucination detection, prompt regression, and outcome-focused roadmap decisions—pressure-testing what we build against whether it holds up in production, not just a feature list.
Connect with Avishek on LinkedIn ↗How we work
AI-native in method, measured in production.
We translate recent AI research into testable product hypotheses, then use domain evaluations and operational evidence to decide what improves the solution.
Research, then evaluate
We test advances in context engineering, retrieval, memory, and agent security against domain-specific benchmarks.
Make improvement auditable
Quality, provenance, feedback, evaluation, and operational signals stay visible so teams can trust and extend the system.