Copyright & disclaimers
Last updated 14 October 2026.
Copyright
© 2026 Kapil Gupta. All rights reserved. From One Call to a Platform, its text, diagrams and tables, and the content of keshavaai.in are protected by copyright. No part may be copied, reproduced, scraped, redistributed, resold or republished, in any form, without written permission. Reading access is personal and non-transferable; sharing your account or access with others is not permitted.
What you may do
Read online, and quote short excerpts with clear attribution and a link to keshavaai.in, for commentary, review or teaching, within the limits of fair dealing / fair use that apply where you live. For anything more (training materials, group licences, translations), please get in touch.
Illustrative scenarios
The companies, platforms, incidents, numbers and people in the book's worked examples (including the helpdesk assistant and the multi-agent operations platform) are composite and fictional, created to teach engineering ideas. Any resemblance to a real organisation or person is coincidental. Views are the author's own and do not represent any employer, client or past employer.
Not professional advice
The book and this site are provided for education and general information. They are not legal, medical, clinical, financial, regulatory, security or compliance advice. The chapters on clinical decision support, contract intelligence and fraud detection describe engineering patterns, not approved clinical, legal or financial practice; do not rely on them to make decisions affecting health, rights or money. Consult qualified professionals and your own legal, security and compliance teams before building or deploying such systems.
No warranty
Content is provided "as is", without warranty of any kind. AI systems and the tools around them change quickly; details may be incomplete or out of date. To the fullest extent permitted by law, the author is not liable for any loss arising from use of, or reliance on, the book or this site.
Tools, comparisons and opinions
When the book or this site describes or compares tools and approaches, it reflects the author's understanding of how they are commonly used for a particular use case, as of the date shown. These are not rankings, endorsements or purchasing advice, and nothing here claims one tool is better than another in general. Where an opinion is given, it is the author's own and is presented as an opinion. Tools may be open source, source-available or commercial, and their licenses and terms change; each tool name in the book links to its official website, which is the authority on its features and license. Please check it before adopting any tool.
Third-party names and sources
Product, framework, standard and company names mentioned (for example cloud, database, observability and agent frameworks) are trademarks of their respective owners and are used only to identify them; no affiliation, sponsorship or endorsement is implied. Where the book relies on a specific public study, report or news item, it is cited in the text. If you believe something has been attributed incorrectly or should be removed, email keshavaai.in@gmail.com and it will be reviewed promptly.
Prior work and credits
The book is the author's own writing and diagrams, but many of its ideas and terms build on public work by others, which deserves credit: retrieval-augmented generation (Lewis et al., 2020); the "lost in the middle" effect (Liu et al., 2023); ReAct (Yao et al., 2022) and chain-of-thought prompting (Wei et al., 2022); service-level objectives, error budgets and the four golden signals (Site Reliability Engineering, Beyer et al., Google, 2016); the DORA delivery metrics (the DORA research programme at Google Cloud, and Accelerate, Forsgren, Humble and Kim, 2018); the "paved road" (Netflix) and "golden path" (Spotify) platform ideas; the Model Context Protocol (Anthropic) and Agent2Agent protocol (Google); and ISO/IEC 42001:2023.
Real-world cases described in the book (including Uber's QueryGPT, Moffatt v. Air Canada, New York City's MyCity chatbot, Klarna's support assistant, Zillow Offers, the Replit incident, and the EXAM federated-learning study) are summarised from public reporting and cited in the text. They are described as reported at the time; rights in the original works remain with their authors and publishers.
Report an error, a missing credit or a rights concern
If you believe something here uses your work without proper credit, describes something inaccurately, or should not be published, please email keshavaai.in@gmail.com (or use the contact form and choose "A question or message") with the chapter or page and what you would like changed. I aim to respond to every message within 15 business days. Missing credits and factual errors are corrected promptly; if material may have been copied, the passage is taken offline while it is reviewed, and then credited, rewritten or removed.
Corrections and updates
- 14 October 2026 — Pre-launch review of the book's sources and wording. Real-world case summaries were rewritten from their original sources, unsupported statements were removed, and chapter counts and cross-references were corrected.
External links
Links to other sites (such as the sources cited in the book and the websites of the tools it mentions) are for convenience. Their content, availability and terms are outside my control.