Slack Code

Original Article ↗ Hacker News Discussion ↗

Slack’s new “Slack Code” feature aims to turn the chat app into a collaborative hub for AI-assisted software development, letting teams work with coding agents directly inside channels. Commenters are split: some see it as an inevitable move as work, project tracking, and AI agents converge in Slack, potentially enabling non-engineers to prototype and ship small changes. Others criticize it as buzzword-heavy SaaS bloat, question its real ergonomics versus traditional IDE- and Git-based workflows, and express fatigue with yet another “agentic coding hub” that may complicate pricing without solving core developer needs.

Overall reaction

  • Many are underwhelmed; several expected simple improvements like native syntax-highlighted code blocks or better workflows, not a full “coding in Slack” experience.
  • Some see it as an obvious move: Slack wants to be the central place for work, including AI agents and coding, especially in an enterprise context.
  • Others view it as “yet another coding agent” and marketing-heavy rather than substantively new.

Perceived value and use cases

  • Supporters think it could:
    • Let product/operations/PMs prototype or triage issues without waiting on engineers.
    • Bring code discussions, diffs, and agent activity into the same place teams already chat.
    • Fit naturally with Salesforce/Slack org metadata and other tools already living in Slack.
  • Skeptics argue serious work still belongs in an IDE/GitHub, seeing Slack as better suited to communication, not as a primary dev environment.

AI coding and workflow trends

  • Multiple comments describe workflows where AI generates most code, tests, and even reviews, with humans guiding and approving.
  • Some teams reportedly ship much faster with agentic setups, with clients accepting that engineers don’t fully understand every implementation detail.
  • Others still prefer “open editor, type, run make,” feeling overwhelmed by constant tool churn but not obviously worse off for ignoring it.

Concerns about code quality and reliability

  • One camp reports “it’s fine”: similar bug levels as before, just faster delivery.
  • Another camp worries:
    • LLMs create increasingly tangled, hard-to-reason-about codebases.
    • Engineers may become dependent on LLMs to understand their own systems.
    • Long-term costs and access limits (token budgets, vendor control) could collide with atrophied skills.

Tool and SaaS fatigue

  • Several express “ecosystem fatigue” with overlapping AI products (Slack Code, Buzz, Claude Code, IDE agents, Jira/Atlassian tools, Teams/GitHub).
  • There is strong criticism of SaaS pricing and upsell tactics, with hopes that AI plus open source will make self-hosted alternatives more viable.

UI and feature wishes / unclear aspects

  • Some just want:
    • Better syntax-highlighted code blocks (not yet easily accessible in the UI).
    • Less painful Slack workflows (e.g., branching/rejoining paths).
    • Simple agent abilities like reading/editing messages and canvases.
  • How Slack Code actually wires to GitHub/infra is seen as unclear; some suspect it mostly fronts existing cloud-hosted agents rather than providing deep new capabilities.