AI-ready product documentation published on release day across every supported version

A Docs-as-Code documentation platform for a graph database platform company, structuring multi-product, multi-version documentation as a knowledge layer for both people and AI systems.

Outcome

Complex product workflows are easier to discover and follow, documentation ships with each release through automated builds and AI-assisted review, and product knowledge sits in a structured layer that supports search today and provides a foundation for LLM-powered assistants and agentic workflows.

See the results

Talk to an amp'd specialist

At a glance

Client & industry

Graph database platform company, with an enterprise graph database and analytics product line

Engagement

Staff augmentation and managed documentation engineering

Scale

Multi-product, multi-version documentation spanning the server, query language, visual graph studio, insights, cloud platform, tools, APIs, and enterprise capabilities

Business owner

Engineering and Product teams, throughout the product lifecycle

Maps to below services

Data Engineering & Advanced Analytics, Agentic AI & AgentOps

Where it started

A large, fast-evolving documentation estate that needed to work as structured product knowledge for people and AI systems alike.

Business problem

The client's technical documentation covered multiple products, releases, APIs, enterprise features, cloud capabilities, installation, configuration, security, upgrades, and troubleshooting, and it kept growing. As it grew, product knowledge needed to be structured, discoverable, version-aware, and accessible to both humans and AI systems.

Technical barriers

Documentation was distributed across multiple repositories and product areas, and several product versions had to be maintained at once, with every change backward-compatible across supported versions. Complex topics needed continuous restructuring as products evolved, search and navigation had to keep pace with new and reorganized content, and content had to be structured so AI agents could retrieve product knowledge accurately.

Agivant's role

Technical documentation engineering, documentation platform engineering, information architecture, release documentation, and AI-readiness of product knowledge

What Agivant built

Agivant's documentation team works directly with Engineering and Product teams throughout the product lifecycle, on a Docs-as-Code platform built on Git, AsciiDoc, and Antora. Each change moves from requirement through technical analysis, information architecture, implementation, automated validation, engineering review, search indexing, and publication. Value in days and weeks.

Docs-as-Code platform

Git repositories, AsciiDoc, and Antora, with versioned branches and automated builds feeding a documentation site searchable by people and AI systems.

Task-oriented information architecture

Complex areas are redesigned around what customers need to accomplish, starting with customer task analysis before any page is rewritten.

Backward-compatible restructuring

Page aliases and stable URLs keep existing links and references working when large documentation areas are reorganized.

Release documentation

Documentation is maintained as part of the product release lifecycle, with release notes, bug fixes, CVEs, and version-specific updates published on release day.

PR Review Agent

An AI review layer checks content structure, consistency, formatting, cross-references, and implementation quality before a change reaches engineering review.

Search and discoverability

Algolia indexing for newly published and reorganized pages, plus SEO improvements to the Antora platform and documentation-site content.

Machine-readable knowledge layer

Consistent structures, semantic cross-references, and version-aware content, so the same product knowledge serves documentation search and provides a foundation for LLM-powered assistants and agentic workflows.

Feature coverage audits

Feature inclusion is audited across release lines to identify missing or inconsistent coverage.

How it is built

Source control	Authoring	Site generation	Versioning	Search	Quality	LLM discoverability
Git, GitHub	AsciiDoc	Antora, Antora UI	Versioned documentation branches, cross-references, page aliases	Algolia	Automated documentation builds and validation, AI-assisted PR review	Dynamic llms.txt, under investigation

Before

Documentation maintained as a collection of individual pages across repositories, with navigation and search trailing newly reorganized content.

After

A structured, task-oriented, versioned, and searchable knowledge platform built as Docs-as-Code, with automated builds, AI-assisted review, and content ready for both human and machine consumption.

Proof in production

Query operations documentation was reorganized into two task-based areas, one for creating and installing queries and one for running and managing them, with page aliases keeping every existing link working. The same approach now runs across backup and restore, REST API, and RBAC documentation. A recent release's documentation, including release notes, bug fixes, CVEs, and version-specific updates, was published on the product release date, and every change passes through the PR Review Agent before engineering review, so fewer avoidable iterations reach engineers.

Measured results

Complex product workflows are easier to discover and follow, and documentation development scales with the release cadence.

Release day

Documentation published with the product release

Every change

Reviewed by the PR Review Agent before engineering review

Multi-version

Supported releases documented in parallel

Preserved

Existing URLs and references, through page aliases during restructuring

Indexed

New and reorganized pages searchable through Algolia

The repeatable pattern

Reusable asset

The engagement established a reusable AI-ready documentation blueprint, covering Docs-as-Code architecture, multi-version management, task-oriented information architecture, automated builds and validation, search indexing, AI-assisted review, and machine-readable content, for any enterprise product whose technical knowledge base needs to become more structured, searchable, maintainable, and AI-ready. Every build amplifies the next.

Challenges overcome

Restructuring complex documentation without breaking existing URLs and references, resolving Antora build and resource-ID conflicts during large reorganizations, validating major UI changes across different product documentation experiences, and introducing AI-assisted review while keeping human technical and engineering validation.

Enterprise movement

Foundational

Operational

Path to autonomous