tldr: Software development is evolving rapidly. The industry is moving from manual coding to AI-assisted development and automation. This creates unprecedented documentation needs. Documentation is shifting from a “nice-to-have” to a critical requirement for product success.
Here are five reasons why the future of software development is less coding, more documentation.
1. The move towards platform engineering and reducing cognitive load
Large enterprises are moving away from traditional DevOps towards platform engineering. Internal teams build “Internal Developer Platforms” (IDPs). These platforms provide self-service infrastructure, CI/CD, testing, security and observability.
The goal is to reduce cognitive load on developers. If the platform is difficult to use, multi-million pound investments fail.
Platform engineering treats internal platforms as products. This requires documentation with the same rigour as external product documentation.
Technical Authors create “golden path” documentation. These are step-by-step, frictionless guides. They show developers exactly how to deploy services without understanding underlying infrastructure. This is core to developer experience (DevEx).
In continuous delivery environments, the challenge shifts. It moves from “write a manual” to “continuously maintain a living knowledge system”. This system spans product usage, operational runbooks, troubleshooting and change communication.
The 2024 DORA findings emphasise that documentation quality connects meaningfully to performance and user-centricity. This contradicts industry slogans sometimes misinterpreted as de-prioritising documentation.
Documentation is shifting from a milestone to an integrated control. It moves from waterfall-style “end phase” to continuously updated content. This reduces variability and risk in delivery and operations.
2. API-first development and advanced specifications
APIs are the primary product for many companies. The 2025 Stack Overflow developer survey reported 90% of developers preferred API/SDK documents as their primary learning resource.
Simple Swagger documentation might not be enough. New standards like the Arazzo specification and AsyncAPI require expert architectural writing. Arazzo documents multi-step API workflows. AsyncAPI handles event-driven architectures.
Technical Authors optimise “time to first call”. This measures how long third-party developers take to make their first successful API call. It’s a key metric for API sales and developer adoption.
3. AI governance, compliance and ethics requirements
The EU AI Act entered into effect in August 2025. Various US state laws also apply. Software companies face unprecedented regulatory pressure.
High-risk AI systems must have technical documentation before market release. This documentation must be kept up to date. Compliance requires technical transparency documentation.
Required documentation includes risk assessments, model cards and data lineage reports. Model cards explain how models were trained. (See Annex IV of the AI Act for complete requirements.)
Technical Authors bridge the gap between legal requirements and technical reality. They ensure AI products are legally cleared for market. This is auditable technical writing.
4. Software supply chain security and transparency
Security documentation is no longer optional. Transparency is the new standard. Around 90% of organisations fall under one or more regulatory requirements.
The EU Cyber Resilience Act requires documentation of control frameworks. From 11 December 2027, companies must document ICT risk management and incident reporting. Similar requirements exist for financial entities.
Technical documentation must include software bills of materials (SBOMs). It must also include dependency tracking and secure development evidence. These must be audit-ready documentation sets.
Vibe coding and shadow application development
An increasing amount of code will be developed by AI. And an increasing number of apps will be developed by people who don’t know how to code.
AI-assisted development increases the volume of shipped changes. It also increases system complexity. Both must be understood and operated safely.
Generative AI tools enable non-developers to build applications. Security teams struggle to gain visibility over “shadow” apps. Marketing or HR departments might build these apps.
Low-code/no-code platforms democratise development. However, they still require extensive documentation:
- Templates and component libraries
- Best practices for citizen developers
- Integration guides with traditional code
- Governance and security policies
Simple automation scripts could expose API keys or personally identifiable information (PII). They require clear documentation of security policies and approved toolchains.
5. The rise of agentic workflows
The industry is making greater use of autonomous AI agents. These agents perform tasks, call APIs and make decisions. They’re moving beyond simple chatbots.
Unlike humans, AI agents are “power readers”. They need documentation formatted for machine consumption. This means highly structured “system instructions” and prompt libraries.
Ideally, AI systems use componentised content with metadata and taxonomies. This makes documentation easier for both users and AI to find. Semantic AI within content management systems can understand relationships and enable reuse.
Technical Authors must curate “ground truth” knowledge bases. This prevents AI models from hallucinating. Retrieval-Augmented Generation (RAG) uses metadata to find relevant content and generate answers with citations.
Who should write the documentation?
Different documentation types require different expertise. The table below suggests ownership for various documentation types.
| Documentation type | Recommended owner |
| API reference documentation | Technical Author with developer |
| System instructions for AI agents | Technical Author with AI engineer |
| Internal developer platform guides | Technical Author |
| AI technical documentation (Annex IV) | Technical Author with compliance team |
| Software bill of materials (SBOM) | Developer with Technical Author |
| Security recipes and playbooks | Technical Author with security engineer |
| RAG knowledge base curation | Technical Author |
| Code documentation and comments | Developer |
| Incident response playbooks | Technical Author with Operations team |
| Low-code/no-code platform guides | Technical Author |
Can’t AI write the documentation?
AI can generate some documentation. Generative AI assists with authoring, ensuring consistency and translation.
However, AI produces high volumes of content with variable accuracy. Companies need human-in-the-loop editors. This prevents documentation rot: where stale AI content leads to broken code.
Technical Authors provide content lifecycle management. They audit, prune and “hallucination-check” entire documentation libraries. Human writers transition from sole Authors to editors and information architects. They focus on structure, voice and analytics.
The ROI of documentation
Defensive value
Documentation reduces operational risk and mitigates support load. It improves onboarding and change safety. It helps organisations satisfy compliance and audit expectations.
Offensive value
Documentation is a primary product surface. API and developer documentation directly influences developer adoption, conversion and retention. This is particularly true in API-first and platform business models.
Conclusion
These developments change documentation in three ways:
- They increase the amount of change that needs documenting
- They provide new ways to generate and personalise documentation
- They create governance and trust problems that documentation must solve
As a result, Technical Authors are becoming essential partners in software delivery.
Ready to transform your documentation?
Contact us to learn more about how Cherryleaf can help you with your documentation needs.
Sources
2024 Stack Overflow Developer Survey. Available at: https://survey.stackoverflow.co/2024/
2025 DORA Report: AI Capabilities Model Report. Google Cloud. Available at: https://cloud.google.com/resources/content/2025-dora-ai-capabilities-model-report
AI Act Service Desk – Article 11: Technical documentation. Available at: https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-11
Announcing the 2024 DORA report. Google Cloud Blog. Available at: https://cloud.google.com/blog/products/devops-sre/announcing-the-2024-dora-report
CNCF Annual Survey 2024. Cloud Native Computing Foundation. Available at: https://www.cncf.io/
Software Bill of Materials (SBOM). NIST. Available at: https://www.nist.gov/itl/executive-order-14028-improving-nations-cybersecurity/software-security-supply-chains-software-1

Leave a Reply