Dipankar Sarkar

Author, The GenAI Playbook

Dipankar Sarkar is the author of The GenAI Playbook — an open-source, executive-grade guide to implementing Generative AI in organizations. He writes about practical AI adoption: strategy, security, data foundations, people analytics, and measurable ROI.

The Playbook is published in 10 languages under the Apache 2.0 license and is used by CXOs, IT leaders, and innovation managers worldwide to navigate Generative AI adoption — from core concepts to future-proofing. A companion Agentic AI Playbook extends the guide into autonomous agents, tool use, the Model Context Protocol (MCP), orchestration, evals, and production deployment.

What this site is

What Generative AI is an open-source reference for turning Generative AI from a demo into durable business value. Its recommendations are vendor-agnostic: they start from your workflow, not from a product. It is deliberately practical: every chapter ends with steps you can act on, and the writing is calibrated for decision-makers who need depth without academic jargon. The two playbooks together span 21 chapters covering strategy, data foundations, security and compliance, people analytics, developer productivity, agent architecture, and the road ahead.

Prefer to browse by concept? Start with the GenAI & Agentic AI glossary for plain-English definitions, then dive into the full chapter index.

Independence and disclosure

The playbooks, guides and decision tools are free and stay free. For organisations that want hands-on help, paid engagements are contracted and delivered by Neul Labs, which Dipankar Sarkar founded and leads. That relationship is disclosed on every commercial page.

Recommendations stay vendor-agnostic, and we do not publish supplier rankings.
When a need is outside our capacity or competence, we say so. Any referral to another provider is disclosed, including any referral fee.
A paid assessment may conclude that you should use the software you already own and not commission custom AI. That is a valid outcome.

About the author

Dipankar Sarkar is a technologist and author. He wrote AI for Everyday Automation (Packt Publishing, 2026), a practical guide to automating knowledge work with ChatGPT, Claude, and Perplexity. He has spent his career building and shipping software, and writes to help organizations adopt AI responsibly and effectively.

Expertise

Generative AI strategy and enterprise adoption
Agentic AI: agents, tool use, MCP, orchestration, and evals
AI security, compliance, and governance (GDPR, EU AI Act, OWASP LLM Top-10)
AI-powered people analytics and HR transformation
Data engineering and RAG for GenAI applications
Measuring ROI of AI initiatives

Editorial approach

Every chapter and post on this site follows the same rule: describe the mechanism, not the hype. Where a claim needs a number — a benchmark score, a pricing figure, a release date for a third-party model — the preference is to link to the primary source or describe the shape of the capability rather than assert a figure that will be stale within months. Comparisons between tools and frameworks are written to name where each genuinely wins, not to declare a single winner, because the right choice depends on the task, the team, and the constraints — not on which vendor is loudest that quarter.

The site is intentionally vendor-neutral. It covers OpenAI, Anthropic, Google, and open-source tooling side by side, and updates chapters as the underlying tools and standards (like MCP) mature, rather than anchoring the guidance to one vendor's roadmap.

How this site is built and maintained

What Generative AI is a static site built with Astro, with content authored in Markdown under an open-source (Apache 2.0) license — the full source, including every chapter and translation, is public on GitHub. Translations into ten languages are maintained alongside the English source so the Playbook stays useful to teams working outside English-language markets.

If you spot an error, a stale claim, or a chapter that needs updating, an issue or pull request against the repository is the fastest way to get it fixed — this is a living reference, not a one-time publication.