IdeaWell

Project ideas from Hacker News discussions.

DevDay 2026 Recap

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

1. Jev vs. OpenAI Decision API – pricing and competitive threat
Many commenters framed OpenAI’s new “Decision model API for Luna” as a direct response to TypeSafe/Jev, focusing on cost differences and market impact.
- “They say it's built on Luna, which costs $0.10M/in, vs Jev which only costs $0.04M/in … Based on how fast they rushed out a competitor … it seems that Jev is much more of a threat to OpenAI than they want to admit.” – HarHarVeryFunny
- “I suspect Jev was always intended to be an acquisition play and instead OAI said 'nah, we're good' … investors/founders are horrified to discover there is no industrial‑scale moat.” – brandall10
- “I did some internal benchmarking … GPT 6 Luna … was about 1.6x faster at 1.2x the cost of Jev … Though this was only useful for offline processing.” – CharlieDigital

2. Split verdict on the usefulness/novelty of the announced features
Opinions diverged sharply on whether the new offerings (Cloud Agents, Dots, Ultrafast, etc.) were genuinely valuable or mostly hype.
- “Cloud Agents are awesome, super useful, I can't go back to working locally only … Dots have a cheesy name, but they remind me of a more polished version of Cursor Projects.” – senordevnyc
- “This could and probably is me projecting, but the enthusiasm was tepid … The reset button stuff was an embarrassment … Everything that was released today … was a failure.” – prodigycorp
- “I’m struggling to find any use cases for them … The 'Super Fast' addition is nice … but overall, it feels like they're missing the mark.” – itzikkatz

3. Hype, crowd reaction, and skepticism about employee‑driven enthusiasm
Several users pointed out that the exuberant audience reaction seemed driven by OpenAI staff rather than genuine public excitement, questioning the authenticity of the hype.
- “The crowd lost their minds before Sam even told them what Dots were … Yeah, the front dozen rows were reserved for OpenAI employees.” – rolymath & simonw
- “Aren’t these crowds mostly people that work at the company? So you’re really cheering for the thing you built?” – easton
- “If that's true that restores my faith in humanity maybe they're cheering for stuff their friends were working on. Otherwise, the consumerism has reached a point where giving up on society … seems like a reasonable option.” – rolymath (reflecting on the dynamic).

🚀 Project Ideas

DevValue Meter – AI Tool ROI Dashboard

Summary

  • Tracks usage, time saved, cost, and correlates with engineering metrics to quantify ROI of AI coding assistants.
  • Core value proposition: turns vague productivity feelings into hard numbers, justifying subscription spend.

Details

Key Value
Target Audience Engineering teams using Codex, Cursor, Cloud Agents, or similar AI dev tools
Core Feature Ingests IDE/plugin telemetry (via open SDK), aggregates token usage, computes time‑saved estimates via commit/PR data, shows cost vs. saved engineer‑hours
Tech Stack TypeScript/Node backend, React frontend, Postgres, optional OpenTelemetry for telemetry
Difficulty Medium
Monetization Revenue-ready: SaaS tiered pricing (free for small teams, $9/user/mo for advanced analytics)

Notes

  • HN commenters lamented: “If OpenAI can show that a $6k tool nets $50k in productivity… they have no problem selling it.” (onion2k)
  • Provides concrete data for internal debates and vendor negotiations, sparking discussion on AI‑tool ROI measurement.

DecisionBench – Open Benchmark Suite for Classifier APIs

Summary

  • Provides reproducible benchmarks (latency, cost per million tokens, accuracy) for decision/classifier APIs like Jev, Luna, OpenAI Decisions API.
  • Core value proposition: lets teams pick the cheapest, fastest model that meets their accuracy SLA without guesswork.

Details

Key Value
Target Audience ML engineers, product teams building business‑automation workflows that need classification/decision models
Core Feature Runs a standard dataset (e.g., intent classification, fraud rules) against multiple APIs, reports throughput, cost, confidence intervals; includes batch‑size tuning wizard
Tech Stack Python (FastAPI) for benchmark runner, Docker for isolation, ClickHouse for results storage, Grafana for dashboards
Difficulty High
Monetization Revenue-ready: hosted service with free tier (limited runs) and paid plans ($49/mo for unlimited private benchmarks)

Notes

  • HN users complained: “No benchmarks or price comparison… likely means it doesn't compare that well.” (dudus) and asked for internal benchmarking.
  • Enables transparent model selection, a frequent topic in discussions about Jev vs. Luna pricing wars.

AuthBridge – OpenAI Subscription Gateway for Third‑Party Apps

Summary

  • Acts as a bridge that lets any third‑party app authenticate with a user’s OpenAI/ChatGPT subscription and securely forward token usage to the provider for billing.
  • Core value proposition: eliminates the need for custom sign‑in flows and gives indie apps a simple way to leverage users’ existing AI quotas.

Details

Key Value
Target Audience Indie SaaS developers, power users who want to use their OpenAI subscription inside non‑Codex tools (e.g., note‑taking, design, automation)
Core Feature OAuth‑like flow that exchanges an OpenAI‑issued access token for a scoped usage token; proxies API calls, meters consumption, and reports back to OpenAI for billing
Tech Stack Go (or Node) proxy server, Redis for rate limiting, PostgreSQL for user‑app mappings, OpenAPI spec for compatibility
Difficulty Medium
Monetization Hobby (open‑source) – can be self‑hosted; optional hosted version with revenue‑ready pricing ($5/mo per connected app)

Notes

  • HN commenters exclaimed: “Sign in with ChatGPT: how is this not major news? They're launching with partnerships…” (ac29, senordevnyc) showing demand for broader subscription use.
  • Solves the friction of managing multiple AI subscriptions and encourages cross‑tool innovation, likely to generate discussion on API‑level subscription sharing.

Read Later