[2/2] How AI is Transforming Work at Anthropic (AI column)
Second half. parsing It focuses on what managers can learn from the Anthropic report.
ROI and productivity The use of AI has a tangible economic effect. An internal survey showed ~50Percentage increase in productivity per employee, and real metrics confirm this: for example, the number of daily code changes (PR) He grew up on an engineer. 67% after the introduction of Claude. In other words, thanks to AI, the team is doing significantly more in the same time. Plus ~27Percentage of tasks that Claude helps solve have never been done before (lack of resources) Now these improvements and experiments increase the quality of the product and open up new possibilities.
AI does not replace, but strengthens Despite the jump in productivity, people are still needed. Most engineers use Claude on a daily basis, but they can only automate it completely. 20Percentage of work. The rest requires human participation: setting tasks, monitoring and editing. AI speeds up the execution of routine parts and provides rough solutions, and experts finalize the result. Claude is an accelerator, not an autonomous worker.
Changes in the team AI tools are changing the working dynamics. The developers now ask Claude first, and then from colleagues. On the one hand, it reduces the burden on experienced employees. (Less distracting questions about small things) It allows you to focus on more complex problems. On the other hand, team interaction and mentorship suffer: beginners are less likely to turn to older people for help, relying on AI. Without dedicated efforts, this can lead to a failure of the transfer of experience. Managers should consider this effect and perhaps formalize mentoring. (As AI takes over some of the learning, we need to find new ways to develop younger colleagues.).
Risk of skills loss Engineers are expanding their profile with AI, but there is a danger that basic skills will become “rusty” with rare use. Some employees are already admitting that yes, they are less practised in manual coding and the intricacies of algorithms, although this is not a big obstacle. There are even those who consciously sometimes solve problems without the help of Claude, so as not to lose shape.
Personnel planning and training There are new accents in the profile of engineers. Many actually turn into managers of AI-agents – control and direct the work of several copies at once. The work goes to a higher level of abstraction: less manual labor, more review and coordinating functions. As one teamhead joked, now his job is to "responsible for work." 5 or 100 Copies of Claude instead of one developer. In the future, the profession may shift to systems design and AI mentoring, and the ability to correctly set tasks and test answers will become a golden skill.
Uncertainty and adaptation Strategically, it is important for management to prepare for different scenarios. The long-term trajectory of the teams is still unclear: even the engineers themselves find it difficult to say how their role will change. 3-5 years. Many people have mixed feelings: today everything is fine, and tomorrow, you see, AI will take even more tasks. Some enthusiasts are confident that the industry will adapt – improved “fences” for AI, training will become part of the tools, and people will grow with machines. But the common denominator is this: you need to be as flexible as possible.
How Anthropic prepares The company is already thinking about how to cope with these challenges. Discuss new regulations for working with AI, how to encourage cooperation and knowledge sharing in the era of AI, how to support the professional growth of employees. Even structural steps are considered: to create new career paths, reskilling programs within the organization as models grow. In addition, Anthropic is expanding research on the impact of AI beyond engineers and helping external partners tailor curricula for the AI future.
#Engineering #Software #Processes #Productivity #Economics