DevOps promised to break down silos between development and operations, improve speed to delivery and enable a sustainable ‘you build it, you run it’ model. While this worked for many organisations, the complexity of modern toolchains, cognitive overload and duplication of operational tasks by many teams has led to the rise of platform engineering. Platform engineering is the practice of forming dedicated platform teams to develop an Internal Developer Platform (IDP). This enables the best parts of DevOps to be productized and ensures the DevOps philosophy can be practiced in a sustainable way. Below are 5 ways that this change is impacting software engineering.
Productising Autonomy
While autonomy is a great goal for DevOps teams, it works well for small teams who share a mutual context. Once organisations grow past a few dozen engineers autonomy can quickly become knowledge hoarding within tribal teams, tickets and inconsistent environments. Platform teams help by making the internal platform a product. Product teams can then utilise golden paths, self serve service catalogs and curated workflows to provision environments, deploy services and manage secrets all without raising tickets.
Developers truly have autonomy here. Whilst they retain ownership of their services, the platform team will own the reliability, compliance and patterns used. This helps organisations regain the speed of DevOps by reducing the hidden costs of having every team create their own deployment pipelines and cloud setups. As previously stated by Gartner, around 80% of large engineering teams will have dedicated platform teams by 2026.
Substantially Reduces Cognitive Load
A clear issue with not scaling DevOps is the hidden costs for application developers. Many studies and measurements in large companies have shown that developers can spend 30-40% of their time working on infrastructure, tooling and toil. Platform engineering helps solve this by abstracting complexity. Analysis shows that by using an IDP, developers can reduce their operational surface by up to 75% in mature implementations.
By having consistent ways to provision, observe, manage secrets and deploy, developers are spending more time solving issues in their domain. This has additional benefits, including increased deployment frequency, reduced lead times, increased retention of senior engineers and less cognitive load. Organisations who prioritise developer experience by measuring metrics such as time-to-first-deploy and platform usage see additional gains.
Golden Paths Reduce Complexity
Having everything consistent can cause friction in DevOps teams that value autonomy. If rules are too strict this causes resentment within the team and creates shadow IT. If there isn’t enough consistency then there’s a risk of configuration drift, lack of security and duplication of effort. Platform engineering helps by providing ‘golden paths’. These are opinionated defaults which provide security, reliability and DevOps best practices but also provide escape hatches for valid reasons.
Teams who follow the golden paths will be faster as they won’t have to solve complex issues like networking, identity, cost control and compliance scanning. Teams who need to deviate will do so clearly. The platform now becomes your enforcement point for consistency across your organisation. This means you won’t have hundreds of services with inconsistencies, something DevOps was trying to avoid.
Security, Compliance and Cost Control Built In
Traditional DevOps implementations often had security and compliance being an afterthought. Platform engineering helps solve this by embedding security into your development process. Policy as code, scanning, identity and access control and FinOps are all built in. If a change is made by either a human or AI agent, it will still be controlled.
This means there is less friction between speed and security. Cost visibility is clear when a resource is created. Compliance evidence is generated constantly. The platform team becomes your control plane. Product teams can work faster knowing they’ll still meet the requirements of the enterprise. Many mature platforms have these features as standard.
AI Friendly Platform
With the rise of AI coding assistants, agent workflows and remediation, we’re seeing the limitations of disparate tooling. Platforms who have consistent APIs, service catalogues, ownership graphs and policy engines can be used as a platform for AI. Developers (and agents) can find services, understand dependencies, request new compliant infrastructure and deploy.
Leading organisations are moving towards AI friendly IDPs. These platforms provide services that can be called by machines, with correct identity. This helps reduce any risks of rogue AI. By having consistent ways to select tests, generate configurations, identify anomalies and even resolve simple incidents you’ll see your AI working faster. As you’ve invested in the platform to improve developer experience, you’re getting more out of your AI investment. Organisations with mature platforms are seeing better returns from their AI investments.
Conclusion
Platform engineering is not a fad. It is a discipline to ensure that DevOps principles can be sustained for organisations who have grown beyond a certain size. The most successful implementations will see the platform as a product. Measuring how many people are using it, how happy developers are, reducing toil and business metrics such as time-to-market.
As more organisations become mature, the competition will come down to not just having DevOps, but consistently, safely and quickly providing DevOps across your whole engineering team. Those who get the platform right will give their developers more time to focus on what truly differentiates them.
