How are natural language interfaces changing software engineering?

All answers agree that natural language interfaces shift engineers' roles from direct coding to orchestrating, reviewing, and designing workflows for agents, fundamentally changing traditional software development tasks.

5 answers

Answers

  • Shawn Wen

    CTO and Co-Founder · PolyAI

    Enabling multi-task workflow

    Shawn Wen says agents like voice agents let him do email triage, video editing and coding at once, changing his workflow to a breadth-first search one because those domains are specified well enough for agent autonomy.

    The answer starts at 57:31.

    Machine Learning Street Talk (MLST) · Listen on Apple Podcasts
  • Thariq Shihipar

    Member of Technical Staff · Anthropic

    Turning workflows into loops

    Thariq Shihipar says the change is turning every part of the software development lifecycle into a routine or loop, which he thinks is where things are headed.

    The Developing Dev · The Developing Dev · Read
  • Patrick Debois

    DevRel · Tessl

    Engineers become orchestrators

    Patrick Debois says the developer eventually becomes more of a conductor and orchestrator of agents, an evolution he considers fair.

    AI Engineer · Watch on YouTube
  • Anders Hejlsberg

    Technical Fellow · Microsoft

    More reviewing, less typing

    Anders Hejlsberg says the craft is changing from typing lines of code to having agents type them while engineers review the agents' work, so they do more reviewing and less typing.

    Ryan Peterman · Watch on YouTube
  • 🦄 Peter Steinberger

    Founder · Amantus Machina

    Design pipelines for agents

    Peter Steinberger says the work is now designing a pipeline from what you want to the outcome, part of software engineering—engineering how his agents can do their best job—because better pipelines produce better outcomes.

    a16z speedrun · a16z speedrun · Read

More Software Engineering questions

View all →