From manuscript to marketplace.
This is a guided tour of the Packt publishing platform for everyone who isn't an engineer. Ten short chapters follow a piece of content through the system: how it arrives, who may use it, how the platform understands it, and how it reaches the channels where customers buy. Press play on any diagram to walk through it step by step.
One platform, seven jobs, and a river of events.
The platform is a small set of focused services. Each one does a single job well, and none of them call each other in secret: when something meaningful happens, a service announces it as an event, and whoever cares reacts. Hover any box for a plain-English job description; click it to jump to its chapter.
A manuscript walks in the door.
Content arrives from the places people already work — SharePoint, Google Drive, Dropbox, or a straight upload. Ingestion's job is to notice it, check it, keep a permanent copy, and turn whatever format it arrived in into one clean internal format the rest of the platform can rely on. Nothing is ever lost, and running the same import twice never duplicates anything.
For the curious
Manuscripts aren't the only thing that arrives. The platform ingests audio, video, images and code repositories too — anything with speech is transcribed and its scenes narrated, so even the slides in a recorded conference talk end up as searchable, quotable text. And the media itself is understood natively: ask a question, get the right clip back.
For the curious
Who may do what, where, and until when.
Every piece of content the platform holds is governed by agreements — with authors, translators, and licensors. The Contracts service is the system of record for those agreements. The rights they grant are worked out once, held on the Work (the intellectual property itself), and copied onto every piece of content that belongs to it. One golden rule runs through everything: silence denies — if no contract grants a permission, the answer is no.
For the curious
One lake, every drop accounted for.
The Content Lake is the platform's single store of content. It doesn't hold "files" — it holds structured, addressable content: every chapter, paragraph, image and code sample has its own identity, its own history, and its own copy of the rights that govern it. This chapter is the anatomy lesson: five layers, from the intellectual property at the top to the raw bytes at the bottom.
For the curious
The platform reads what it stores.
The moment a new version of any content is snapshotted, a crew of automatic workers wakes up. They read the content, link every mention of a technology or person to the Knowledge Graph, polish the images, analyse the tables and code, and label everything for discovery. None of this changes the author's words — enrichment is written alongside the content, and everything it learns flows straight into search.
For the curious
Finding the right paragraph, not just the right book.
Platform search doesn't return "documents that mention Kubernetes" — it returns the exact paragraphs — or, from video and audio, the exact clips — that answer the question, ranked by how well they answer it, and filtered so you only ever see what you're allowed to use. It works by reading the question three ways at once, then narrowing half a thousand candidates down to a handful of great answers.
For the curious
One Work, many ways to sell it.
Publishing has a beautiful old idea at its heart: the book and the copies of the book are different things. The platform models this precisely. A Work is the intellectual property. An Expression is that Work realised in a medium and language. A Manifestation is a sellable embodiment with its own ISBN. And a Product is the commercial wrapper a customer actually buys.
For the curious
When content has children.
The future of Packt isn't just publishing Works — it's transforming them: courses compiled from books, study guides distilled from chapters, assessments built on source material. The platform draws one crucial line. Does the change create new intellectual property? Then it's a new Work — derived, permissioned, and linked to its sources forever. Same IP, just realised differently? Then it's a new Expression of the Work you already have.
For the curious
A course compiled from three books. Needs DERIVE permission; carries its own rights from then on.
The French translation, the audio narration, the corrected revision — same Work, same contracts. (A new edition is different: substantial new content makes a new Work, and almost always a new contract.)
The paperback beside the ePub — new ISBN, nothing else changes.
You decide. Six real scenarios — click what you think each one creates.
We translate Advanced Kubernetes into French.
same intellectual property, realised in a new language. A translation creates a new Expression under the same Work — rights still flow from the same contracts.
We compile chapters from three books into a "Kubernetes Security" course.
new intellectual property. A compilation derives a new Work — it needs DERIVE permission on every source, and carries its own rights from then on.
We record an audiobook narration of the existing text.
same IP in a new medium (audio). Narration creates a new Expression under the same Work.
We release the existing text as a paperback alongside the ePub.
nothing about the content changed — this is a new sellable embodiment of the same text Expression, with its own ISBN.
We write a certification exam based on the book's material.
an assessment is its own Work — a Certification product wraps an assessment Work. New IP, derived from the source, with its own rights basis.
We publish a 3rd edition, updated for the latest Kubernetes release.
a new edition is substantial new authored content — a new Work that supersedes the prior edition's Work, almost always under a new contract (authors and royalties can change between editions). Only a small same-content correction — a revision — stays under the same Work and contract.
One product, dressed for every channel.
Between the products Packt makes and the channels that sell them sits the catalogue: a curated collection that decides what's on offer, where. Some titles are picked by hand; others arrive by rules that quietly re-evaluate themselves as rights and lifecycles change. And catalogues are where channel-specific presentation lives — Distribution never ships the original product record, it ships the catalogue entry.
An editor adds a title to the "Spring Sale" catalogue — deliberate, one product at a time.
"Every published Kubernetes title we may sell in the EU." Membership re-evaluates itself whenever rights or lifecycles change.
Every entry and exit is published — product_added /
product_removed — so Distribution reacts the moment
membership changes.
- Title
- Advanced Kubernetes
- Blurb
- Production-grade cluster security, from first principles.
- Price
- $49.99
overrides
- Title
- Advanced Kubernetes, Second Edition
- Blurb
- Production-grade cluster security, from first principles.
- Price
- $39.99 launch promo
The base product record stays untouched. The same product can sit in many catalogues at once, dressed differently for each channel — and when it leaves a catalogue, Chapter X shows what happens next.
Reaching the shelves, and leaving them gracefully.
Published products reach customers through channels — Amazon US, Amazon DE, O'Reilly, packtpub.com. A channel isn't just a name: each partner is stored with a declared way of using content, which makes the channel boundary the final gatekeeper of rights — nothing ships anywhere it isn't allowed to be used. And when things change, the platform reacts as a chain of events with no one pressing buttons. Watch the finale — the ripple.
For the curious
Words you'll hear.
The intellectual property itself — "the book", independent of any format. Rights live here.
A unit of content in the Lake: a chapter, an image set, a video, even a code repository.
A frozen, unchangeable snapshot of a Resource. Products pin the version they were built from.
The platform's atom of content — one paragraph, table, image or code sample, individually addressable and citable. PCF is the Packt Content Format.
The raw bytes behind media — stored once no matter how many times they're used, shared via short-lived links.
The record of an agreement — parties, territories, formats, usage, royalties, duration. The source of every permission.
The Work's resolved rights, copied onto each Resource so every read already knows what's allowed.
A Work realised in a medium and language — the French translation, the audio narration, the 2nd edition.
A sellable embodiment of an Expression — the paperback, the ePub — each with its own ISBN.
Creating new IP from existing Works — a course from three books. Needs DERIVE permission; the result is a new Work with its own rights.
The commercial wrapper customers buy: Book, Audiobook, Video Course, Course, Interactive, Certification.
A real-world thing — Kubernetes, Python, an author — that content mentions and links to.
A service's public announcement that something happened. Other services react; nobody polls.
A curated collection of products with self-evaluating rules — the unit Distribution ships to channels.
A partner that takes products to customers — stored as a profile of how it uses content (formats, territory, access model). The final gatekeeper of rights.
The provenance ledger: who or what created, used or derived each piece of content, forming a full family tree.
A business unit on the shared platform (Packt, ExpertEdge, CVP). Each tenant's data is architecturally separate.