Anthropic Shows What Most Developers Are Missing

What really separates the developers who will thrive from those who will get replaced? Ever since AI entered the mainstream, it has started automating a lot of things for us, completely transforming our workflows. As you saw in our previous video that Claude has become an orchestrator of agents, software developers were the first to adopt it heavily because much of their work involved repetitive codew writing which often became inefficient. Now AI is a main part of every developer's workflow. And if you still use AI the same way you did 6 months ago, you are not keeping up. Upon this scenario, Anthropic released an article discussing trends in software development. As we were talking about it within our team, we found that it was something that was too inherent in our workflow and resonated with us. The software development life cycle is changing dramatically. The cycles that used to take weeks or months are now happening in hours because of AI. The traditional life cycle looked like this. weeks worth of planning and design, implementation and testing, review and then the cycle repeated. That changed with AI. Now you only express your intent and the agent produces an implementation. The only parts where humans are still involved are the review and expressing the intent. The rest is handled by AI agents. This changes what engineering means entirely. Software development doesn't mean writing code anymore. It means orchestrating agents that write code, providing strategic direction, and making sure the system works as intended. Even the onboarding has collapsed from weeks to hours. AI can explore the codebase and onboard new joiners immediately. And now as our focus is on directing agents, everyone is becoming a full stack engineer rather than a specialist in a single domain. Engineers can work with just basic knowledge of their stack and AI fills the gaps while they lack knowledge. This enables tighter feedback loops and faster learning. Weeks of cross team coordination become a single working session. This matches exactly what Linear's CEO predicted in his article that the middle of the software workflow has been replaced by AI. And if you're still spending your time in the middle phase, you're working against this shift. And this brings us back to the same principle we keep talking about. You need to be effective in your planning and express your intent in clear terms. The skill that matters most now is clarity. That is describing exactly what you need and making agents deliver the best product. Before we move on, team AIABS is attending the web summit 2026 being held in Doha, Qatar. If you are attending or are nearby, this is your chance to meet with the team, connect with us and learn from us. Looking forwards to seeing you there. We've evolved from single agents to multi- aent systems. We have already covered this in our previous video that Claude code has now implemented multi-agent architecture inside their product. Earlier the structure used to handle development with a single agent having a single context window and managing all the tasks by itself. The problem was a single context window got bloated fast because there was too much information in its working memory making it lose focus. Now there's an orchestrator agent that acts like a project manager and delegates tasks to specialist agents. Each of the agents has its own context window and then integrates the output to produce the final result. Even though Claude handles agent spawning and management on its own, we create our own agents for specialized tasks. We use these agents because they were tailored according to our preferences using different Claude models based on the tasks difficulty and containing instructions to guide the agent. Sub agents have gotten better because you can now let them run in the background, handling different aspects of the application simultaneously, speeding up the whole process. Longunning agents will become more capable. Agents have gone from building feature by feature to being able to build complete systems on their own. This started emerging by late 2025. Ever since models like Opus 4.5 and GPT 5.2 were released. In 2026, agents will be able to work for days at a time with minimal human intervention. Previously, agents were handling small parts of an application. Now, they're building and testing entire applications and systems while verifying if the system is working before moving on to the next feature. We have dedicated a video explaining how to make long-running systems more effective, which you can check out on the channel. With the right tools and workflows, agents are able to plan, iterate, and they recover from failure at scale. This changes the economics of development. In big companies, software accumulates years of technical debt that nobody had time to address. Now, agents can actively work through the backlog. This also opens up a path for entrepreneurs. Previously, the main struggle was the skill gap and time. People had ideas but lacked resources to build them. With autonomous agents, startups can now build and deploy products in days. We also use long-running agents for our tasks. Our workflow for longunning tasks uses claude.md containing instructions. We guide Claude to test after each implementation. For a feature to be well completed, Claude needs a way to verify it's working. We test using agents internally and for visual testing, we use Claude Chrome. Once testing is completed, from both the agents perspective and visual verification, we commit the changes to Git with descriptive messages. This matters because agents tend to modify code and tests that we didn't ask them to. Git lets us roll back easily. We always ask Claude to document the decisions that were made so commits are clean and ready to be shipped. To maximize time, we ask Claude to break down tasks into smaller independent units and assign agents to work on them simultaneously. If you want this claude.md and the agents so that you can use them for your own projects, you can find them in AIABS Pro. For those who don't know, it is our recently launched community where you get ready to use templates, prompts, all the commands, and skills that you can plug directly into your projects for this video and all the previous videos. If you found value in what we do and want to support the channel, this is the best way to do it. Links in the description. Human oversight is scaling through intelligent collaboration. As agents are getting better, they can review outputs much faster than we can. We cannot review large scale outputs that the models are producing at the same speed as agents do. So we are relying on them for all kinds of reviews like security vulnerabilities, architecture consistency and quality issues. Going through a codebase that you didn't write is draining. Agents handle that. Now agents are also learning to ask for help rather than blindly attempting tasks. They know when human input is required and ask questions as part of a team. Our team has already noticed this pattern in Claude. When we said that the output looked bad, it asked clarifying questions about what didn't meet our expectations and how it could improve. Oversight is shifting from reviewing everything to reviewing what matters. We only need to review the exceptional cases where problems might emerge. This also addresses the question of AI replacing developers. Even though AI capabilities are expanding, the role of humans remains central. The main change is the shift from writing code to reviewing code and guiding AI outputs. One of anthropics engineers said the best practice for working with AI is to use it when you know what the right answer should look like. The people who know the answer are those with real software engineering experience who have learned programming concepts the hard way. And how do you know the right answer? When you know which method to use for which purpose. For example, for testing, you need to use specific approaches. We've already shown you how to use the testdriven approach, whitebox testing, and blackbox testing. We've also covered visual testing using tools like the Claude Chrome extension and Puppeteer MCP. Also, if you are enjoying our content, consider pressing the hype button because it helps us create more content like this and reach out to more people. Agentic coding is expanding to new services and users that we have never seen before. Earlier in 2025, AI coding was mostly effective for popular frameworks and often struggled with systems that used legacy languages or frameworks that are not commonly used. That's why the best working applications were react-based because these were the frameworks model was primarily trained on. By 2026, agentic coding will expand into contexts that traditional development tools couldn't reach, including support for legacy languages like cobalt, forran, and other domain specific languages. This will make maintaining legacy systems easier by eliminating the need to navigate through old documentation. AI has made development accessible for non-developers, opening up opportunities to non-traditional developers in fields like cyber security, operations, and data science. The release of co-work is already signaling progress in this direction. The barriers separating people who code and people who don't are becoming increasingly invisible as AI progresses. For example, someone on a security team can use AI to understand unfamiliar code in order to find issues. Research teams have been using it to build front-end visualizations for their data and nontechnical employees are already using AI in areas unfamiliar to them like networking and data analytics. This is something our team has already been doing. One of our team members wasn't familiar with Golang, but was tasked to make a backend for a chat application. They turned on plan mode and created a whole plan by answering the questions about the app. Claude built the entire server in one shot, working exactly as intended. This eliminated the need for wasting time learning a new language for just one task. Productivity gains will reshape software development economics. We already mentioned how timelines have been compressed because agents handle the difficult parts. Three factors reinforce each other. Agent capabilities, orchestration improvements, and human experience. Together, they compress timelines and change what's viable to build. Projects that were once considered too difficult are now viable, allowing products to enter the market more quickly. Agents help teams work with fewer people. Project timelines are shorter, letting us achieve faster returns on investment. Features that used to take much longer can now be built in a smaller time frame. But before we move forwards, let's have a word from our sponsor, Luma AI. If you've messed around with AI video before, you know the frustration. It usually feels like a slot machine. But Luma AI's new model, Ray 3 Modify, actually changes the game by giving us the modify capabilities that developers have been waiting for. Instead of just prompting and praying, you can now take a video and completely restyle it, swapping environments or lighting while keeping the original motion and physics fully locked. It respects your input data. It's not just generating random noise. It's video to video that maintains structural integrity. Plus, with character reference, you can finally keep your subject consistent across different shots, which is usually impossible. It's the first time an AI video feels like a controllable tool rather than just a toy. Make small productions feel huge. Scan the QR code on the screen or check the link in the pinned comment and try Ray 3 in Dream Machine. Today, there is an increase in the number of non- tech use cases across organizations. Teams in sales, marketing, legal, and operations can now use AI coding to automate workflows and build tools without any engineering team support. AI agents can operate directly by their guidance and develop systems. People with domain expertise and deep understanding of the problems they face use agents to initiate solutions themselves. For example, someone working in accounting or other departments understands the problems they face better than anyone else. They can instruct agents and have a working solution without waiting for dev team. Our team has already been using Claude in our workflow. We automated the boring non-development work like documentation, ideation, and research by using claude code, letting us focus on the interesting and creative part of our work. Agentic coding improves security defenses and offensive uses. Security and AI are a double-edged sword. The same AI that can navigate your codebase and help with onboarding is also capable of exploiting its vulnerabilities. Security knowledge is not limited to security engineers. Any engineer can act as a security reviewer handling hardening and monitoring of systems. Since security engineers are domain specialists, they still need to be consulted. But combining AI with their knowledge, it becomes easier to build, harden, and secure applications. While security engineers can defend the applications, there will be offensive use cases too. Last year, we saw a coordinated attack carried out using clawed code and its tools. This means agentic capabilities will evolve the types of attacks we see, making them more intelligent and harmful than ever. Securing systems is going to become increasingly crucial, and engineers will need to focus on security from the start. AI agents will play a growing role in cyber defense systems, enabling responses that match the speed of offensive attacks. We need to prepare before attacks happen. We also expect a rise in zeroday attacks, making proactive preparation even more important. When our team creates an app, we use specialized agents for security. These agents handle code review, testing, and serverside security, the layer where we control access. Securing applications can be done using different combinations depending on the application, whether it's built-in skills, reusable commands for build purposes, or tools from external MCPs. It is better to use an external tool like Code Rabbit because they're built to catch known vulnerability patterns early. That brings us to the end of this video. If you'd like to support the channel and help us keep making videos like this, you can do so by joining AIAS Pro. As always, thank you for watching and I'll see you in the next one.