Augment, a GitHub Copilot rival, launches out of stealth

Original Article ↗ Hacker News Discussion ↗

A heavily funded GitHub Copilot rival called Augment has emerged from “stealth” with backing from Eric Schmidt, but with few public details beyond a press release and a waitlist. Commenters question why such an obvious product category warrants stealth mode, how Augment can meaningfully differentiate from the many existing AI coding assistants, and whether any of these tools have a durable moat given falling switching costs. The thread also surfaces mixed but often positive experiences with current assistants like Copilot, along with concerns about long‑term economics, GPU costs, and the broader sustainability of the AI coding boom.

Launch & basic info

  • Augment, an AI coding assistant positioned as a GitHub Copilot rival, emerged from “stealth” with a very large funding round (over $200M, article says ~$252M; company blog shows $227M).
  • It claims to use fine‑tuned “industry‑leading” open models but provides few concrete technical or product details publicly.
  • Some commenters joined the waitlist; at least one reports “OK” suggestions from the VS Code plugin, similar to Copilot.

Stealth mode, demo absence, and marketing

  • Many are puzzled by “stealth” for such an obvious product category with many existing competitors.
  • Strong criticism that the public launch has no demo video, limited UX detail, and “schedule a demo” vibes instead of open access.
  • Some speculate stealth and timing may be about hype, press cycles, or hiding GPU demand; others dismiss this as mostly ego and branding.

Differentiation in a crowded field

  • The market is already saturated with coding assistants from big clouds and many startups.
  • Commenters question how Augment can meaningfully differentiate if it uses similar underlying LLM tech and offers the same “better autocomplete” value proposition.
  • Skepticism that any one tool is “head and shoulders” above others; current tools mostly help with boilerplate and configuration.

Economics, GPUs, and business viability

  • Discussion of reports that Copilot may be heavily subsidized, with true costs possibly far above current pricing.
  • Some expect costs to fall with more efficient models and chips; others doubt the strength of moats and lock‑in when switching assistants is easy.
  • Debate on whether a $200M+ war chest matters in the GPU market, with disagreement on how much a single buyer affects prices.

Developer experiences with coding assistants

  • Many strong coders describe assistants (often Copilot) as:
    • Smart autocomplete that saves typing, especially for repetitive code, boilerplate, configs, and unfamiliar APIs.
    • More useful in well‑represented languages (e.g., Go, Python, Rust, JS) and for configs like Terraform/Kubernetes.
    • Less reliable for complex logic or debugging; some explicitly avoid using them for bug fixing.
  • Others find assistants more annoying than helpful: slow, intrusive UI, error‑prone suggestions, and limited value for the subscription price.

Ethical and ecosystem concerns

  • Criticism of the lead investor’s past involvement in anti‑competitive hiring practices and worry that LLMs will reduce developer headcount and professionalism.
  • Some concerns about copyright, training on proprietary code, and data logging; a few hope Augment might be more IP‑respecting, though this is unclear.
  • One commenter complains Augment collected an email then denied access, calling it unethical.

AI bubble and long‑term outlook

  • Several view this as more air in an AI bubble: many near‑identical assistants chasing huge valuations, uncertain paths to profitability, and risk that future frontier models will commoditize current startups’ advantages.
  • Others argue that wide experimentation (“Monte Carlo method”) is normal for emerging tech and that significant workflow shifts (toward higher‑level task/requirements tools) may still be ahead.