Brief IA

Speak Revolutionizes Engineer Recruitment with AI at Its Core

💼 Business & Startups·Tom Levy·

Speak Revolutionizes Engineer Recruitment with AI at Its Core

Speak Revolutionizes Engineer Recruitment with AI at Its Core
Key Takeaways
1Speak, led by Andrew Hsu, is transforming its engineer recruitment by integrating AI into the interview process.
2The company has eliminated traditional coding questions, preferring to assess the use of agents like Claude Code and Codex.
3Engineers are now evaluated on their ability to collaborate with AI agents to maximize productivity and quality.
💡Why it mattersThis approach could redefine the skills required for software engineers, influencing the entire tech industry.
Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisis ton rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

📄
Full Analysis

Speak Redefines Engineer Recruitment with AI

Andrew Hsu, co-founder and CTO of Speak, a language learning application utilizing artificial intelligence, has announced a revamp of the company's engineer recruitment process. Speak, which employs around 150 people, including 60 engineers, has decided to abandon traditional coding questions during interviews.

The company now emphasizes what it calls an "agentic mindset," in addition to the usual core skills. This approach aims to recruit candidates who can work effectively with AI agents, which handle up to 80% of coding tasks.

A Paradigm Shift in Software Engineering

Since last December, coding agents have gained prominence within engineering teams. This shift has transformed the role of engineers, leading them to communicate primarily through these agents. Speak has thus reevaluated its interview method, focusing on agentic engineering skills.

Traditional coding tests have been eliminated, as AI models are now advanced enough to answer these questions. Candidate evaluation now centers on their ability to use tools like Claude Code or OpenAI's Codex. Candidates must demonstrate their ability to build features, solve problems, and debug with these agents while maintaining a high level of quality.

The Emergence of the Agentic Engineer

Speak is looking for engineers with an agentic mindset, capable of leveraging AI agents to optimize their work. The company distinguishes between two types of engineers: the "Engineer One," who uses agents to speed up their work without changing their way of thinking, and the "Engineer Two," who adapts their approach to fully integrate the capabilities of AI agents.

Engineers who succeed in this environment are those who are proactive and intellectually curious. Speak values builders, those who focus on solving problems for users rather than on the beauty of the code. During interviews, emphasis is placed on candidates' ability to design and architect systems, skills that have become even more crucial.

The tech industry is evolving towards this new norm, and those who adopt the agentic mindset will be best positioned to thrive. Hsu noted that the gap between engineers who understand this new dynamic and those who do not is likely to widen. Speak has always sought engineers who are more builders than programmers, who do not tie their identity to the beauty of the code but to the ability of what they build to solve a problem for users.

In conclusion, Speak has not revised its staffing goals for the current year. The company does not plan to reduce its engineering team but is instead looking for engineers capable of working in this new style. Hsu clarified that the entire industry is moving in this direction, and those with an agentic engineering mindset will be the ones to prosper.

The Transformation of Technical Interviews

Speak has always avoided algorithmic coding tests like LeetCode, preferring to pose more realistic coding problems. Today, all traditional coding questions have been eliminated, as AI models are capable enough to answer them. The real test now is whether a candidate can build a feature using agents like Claude Code or OpenAI's Codex.

Candidates must demonstrate that they can discuss and solve problems, plan, and debug through these agents. They must do so in a way that amplifies their output while maintaining a high level of quality. Speak continues to assign take-home projects to candidates, but they now utilize coding agents to the fullest. The on-site process has also been revamped to incorporate agentic coding in a live setting.

Candidates are then invited to explain what they have built and why they made certain choices or compromises. This process allows Speak to ensure that the engineers it recruits are not only technically competent but also capable of adapting to a work environment where AI plays a central role.

The Future of Engineering at Speak

Speak has found that engineers who succeed in this new world of agentic engineering are those who are extremely proactive and intellectually curious. The company has always sought engineers who are more builders than programmers, who do not tie their identity to the beauty of the code and system, but to the ability of what they build to solve a problem for users.

During interviews, Speak wants to see that the candidate, in combination with all the traditional strengths of a human software engineer, is fully engaged in building with agents. Emphasis is also placed on traditional skills, such as system design and architecture. The high-level aspects of building a technical system are now even more important.

The entire industry is moving in this direction, and those with an agentic engineering mindset will be the ones to thrive. Speak has not revised its staffing goals for this year and does not plan to reduce its engineering team. The company is simply looking for engineers capable of working in this new style, convinced that this approach is key to remaining competitive in a constantly evolving sector.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.