IdeaWell

Project ideas from Hacker News discussions.

IBM Bob

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

Theme 1 – Mockery of the product’s name and perceived low quality
Many commenters ridiculed IBM’s “Bob” branding, comparing it to the infamous Microsoft Bob and dismissing the offering as sub‑par or embarrassing.
- “I never thought the day that IBM would make complete slop.” – colesantiago
- “It would have been better to release nothing than this.” – saberience

Theme 2 – Skepticism about IBM Bob’s actual value (viewed as a thin harness/wrapper)
A common view was that Bob is merely a routing layer or harness that forwards queries to other models while adding cost or limiting user choice.
- “It routes to the model best suited for the ask… you can’t even choose which model!” – kbwal7
- “except they tack on surcharges to all token calls to Claude and GPT” – getatme32

Theme 3 – IBM’s pattern of late‑to‑market, legacy‑driven attempts to stay relevant
Several observers framed the launch as another example of IBM chasing trends (LLMs) after missing earlier waves, leaning on legacy branding like Watson rather than delivering genuine innovation.
- “IBM desperately clinging to relevancy by hopping on the LLM bandwagon.” – xyst
- “IBM already have IBM Watson which is probably named after the CEO.” – vablings

🚀 Project Ideas

Generating project ideas…

ModelChoice: Transparent AI Model Selector for VS Code

Summary

  • A VS Code extension that lets developers pick any LLM provider (OpenAI, Anthropic, IBM Granite, local models) and see real‑time token usage and cost in plain USD, eliminating hidden “Bobcoins” and forced routing.
  • Core value proposition: full model choice, transparent pricing, and no vendor lock‑in for AI‑assisted coding.

Details

Key Value
Target Audience Developers frustrated with opaque AI integrations like IBM Bob who want control over which model powers their IDE assistant.
Core Feature Model picker dropdown, per‑request token counter, cost estimator, and custom routing rules (e.g., use Granite for COBOL, GPT‑4 for general code).
Tech Stack TypeScript, VS Code Extension API, LiteLLM proxy for unified routing, Tailwind CSS for UI.
Difficulty Medium
Monetization Hobby (open‑source); optional Revenue‑ready: hosted telemetry dashboard with team‑based pricing ($5/user/mo).

Notes

  • HN users complained “IBM Bob…auto-routes your queries with no way to change this” (EagnaIonat) and “Bobcoins…obfuscate costs” (villish); ModelChoice gives explicit model selection and USD pricing.
  • Provides a practical alternative to the “VS Code fork + agent panel” criticism (taspeotis) by extending the existing editor rather than replacing it.
  • Could spark discussion on open‑source AI tooling and drive community contributions for new model adapters.

LegacyCodeAI: AI‑Powered Modernization Suite for IBM i (COBOL/RPG)

Summary

  • A SaaS toolkit that uses fine‑tuned IBM Granite models augmented with RAG over IBM documentation to translate, refactor, and generate unit tests for legacy COBOL/RPG codebases, with clear per‑token pricing.
  • Core value proposition: accelerates mainframe modernization while avoiding the “slop” perception of IBM Bob and giving enterprises transparent cost tracking.

Details

Key Value
Target Audience Enterprise teams maintaining IBM i applications (COBOL, RPG, CL) seeking AI‑assisted migration to modern platforms.
Core Feature Batch code translation, impact analysis, test generation, and a model‑choice UI that shows token usage and cost in USD.
Tech Stack Python, FastAPI, LangChain, IBM Watsonx.ai SDK, Docker, PostgreSQL for metadata.
Difficulty High
Monetization Revenue‑ready: subscription tiered by lines of code processed per month (e.g., $0.001 per 1k tokens + platform fee).

Notes

  • Commenters noted “IBM has models trained for older mainframe usage (RPG II, RPG/400, COBOL, etc)” (jenkstom) and wished for a useful AI tool for those languages; LegacyCodeAI directly leverages that strength.
  • Addresses the frustration that IBM Bob “is just a harness that uses proprietary models…you can’t even choose which model!” (kbwal7) by exposing model selection and cost.
  • Could generate practical utility by reducing manual refactor effort and stimulating discussion on AI for legacy systems in the HN community.

CarbonChat: Polished AI Chat UI Built on IBM Carbon Design System

Summary

  • A reusable React component library that implements a clean, professional chat interface (markdown, code blocks, tool usage display) using IBM’s Carbon Design System, allowing any LLM backend to be plugged in without the cringe branding of IBM Bob.
  • Core value proposition: gives product teams a high‑quality, design‑system‑consistent AI chat widget quickly, improving user trust and reducing dev time.

Details

Key Value
Target Audience UX designers and product teams building internal AI assistants or customer‑facing chatbots who want a non‑cringe, enterprise‑grade UI.
Core Feature Pre‑built chat window with theming, message avatars, code syntax highlighting, token usage badge, and easy API connector hooks (REST/WebSocket).
Tech Stack React, TypeScript, Carbon Design System, Tailwind CSS, Vite for bundling.
Difficulty Low
Monetization Hobby (MIT‑licensed); optional Revenue‑ready: premium theme packs or enterprise support contracts ($99/yr).

Notes

  • HN criticism: “The landing/marketing page for this looks like something someone promoted/vibe coded including the Bob mascot.” (absoluteunit1) and “They have a decent component system with Carbon…they could have easily pointed their LLM of choice to it.” (nadnad); CarbonChat satisfies that suggestion.
  • Provides a concrete alternative to the “late 2024 energy” and “VPN‑like” feel of IBM Bob, giving teams a polished UI they can trust.
  • Likely to spark discussion on design‑system adoption for AI tools and could be showcased in Show HN threads.

Read Later