Opportunities › Multi-LLM orchestration for developers › Proposed app
Railyard
Route each LLM request to the right model by cost, speed or skill, with one spend view
- Score
- 41
- confidence 58
- Evidence
- 7
- items on the page
- Rivals
- 6
- named in sources
- First version
- 30 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2023
- Emerging market · AI estimate
ScreensConcept
Desktop app: Developers set routing rules, review cost data, and configure LLM request logic at a desk during development and platform maintenance work. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
This month
Under budget$84.20
About this appAI proposal
Railyard gives developers one endpoint for many language models. You add your own provider keys, write rules such as "architecture goes to the strongest model, small edits go to the cheapest", and Railyard sends each request down the route you chose.
It is for developers who already juggle several providers and track prices by hand. Every call is logged with its model, latency and cost, so one dashboard shows what you spent across all providers.
Unlike a fixed model list, the rules are yours and visible. When a provider changes its price, you see the effect on your own traffic and can change a rule in a minute. It is a PWA, so you can check spend from your phone.
- Tagline
- One key. Every model. The cheapest one that can do the job.
- Voice
- PlainPreciseDry
- Why this name
- Railyard: where routes are set and cars are sent down the right track. Two plain syllables, easy to say, and it says what the product does with requests.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Rule-based routing
Write rules by task type, prompt size or tag. Send each to the cheapest, fastest or strongest model you pick.
No member yet
Bring your own keys
Connect your existing provider accounts. No extra subscriptions per model; your keys stay encrypted.
No member yet
One cost view
Spend per provider, model, project and rule in one place, with daily and monthly totals.
No member yet
Price change alerts
Update a provider's rates and see how your recent traffic would have cost under the new price.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#0F4C5CDeep Teal · primary#F6F3ECPaper · surface#1B2224Ink · text#E8A317Signal Amber · accent#4C8C5AMoss · data: savings
Typography
Aa
Railyard
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
One key. Every model. The cheapest one that can do the job.
PricingAI proposal
The material gives no competitor routing prices. Deepseek's page lists one paid plan at CNY 0.02 by usage, so usage stays with the provider via your keys. These tiers are a starting point to test.
Free
Free
- 2 providers
- Basic routing rules
- 7-day request log
Solo
$12 a month
- Unlimited providers
- Unlimited rules
- Fallback chains
- 90-day log and cost view
Team
$39 a month
- Up to 10 seats
- Budget caps per project
- Dry-run replay
- Price change alerts
Would you pay Solo · $12 a month?
Your answer tells PWAS which prices people accept. It changes no score.
No answers yet.
What it could be worthEstimate
What a software business making this much sells for today, from real deals. Pick a monthly revenue.
If it makes
Worth about
$194,000
Sale range $132,000 to $263,000
$120,000 a year in revenue
From 87 software businesses listed and sold on Empire Flippers in the last year, revenue vetted by the broker: they sell for about 19× their monthly revenue, most between 13× and 26×.
Measured on profit, the middle deal is 32× monthly net profit: about 2.7 years of profit.
An estimate, not an offer or financial advice. A real price depends on profit, growth, churn, how much of the work is the owner's, and the buyer.
How it would growAI proposal
How the product would reach new customers through its own use, and how the first ones would be found. A plan to test, not a result.
- Built-in loop
- No loop in the product. Railyard runs behind the developer's own tools and is not seen by their end users. It reaches customers through developer content, comparisons of routing setups, and word of mouth among developers.
- Do they become customers?
- Reached developers are the buyers, so content and community posts can convert directly. To test: whether solo developers pay for routing and cost tracking, or just use provider keys directly.
- First customers
- Solo developers and teams of 2 to 10 engineers who already pay for two or more LLM providers and use them for coding assistance or app features.
- Where to reach them
- Hacker News: Post a Show HN with a worked example of routing rules and the cost difference measured on real traffic.
- Developer communities: Offer setup help in subreddits and Discords where people compare coding models and their costs.
- Open-source rule templates: Publish a public repo of routing rule sets for common coding workflows that link back to Railyard.
- First milestone
- Ten developers route real traffic for four weeks, and five of them pay for a paid tier.
Why people already pay for this
- Public claimDevelopers are actively spending money on multiple LLM subscriptions and APIs to optimize their workflows
- Public claimCost efficiency and pricing are major decision factors when choosing between LLM models
- Public claimClaude's pricing and costs are seen as prohibitively expensive by developers
- Public claimQwen's pricing structure creates friction and customer acquisition barriers
- Public claimDevelopers actively rotate through and manage multiple LLM models for different specialized tasks
Tech stack
Every proposal builds on the same PWA base, installable from the browser on any phone or computer. Dashed boxes are what this app adds.
- AppNext.js and TypeScript
- PWAManifest and service worker
- UITailwind CSS
- Data and authSupabase Postgres
- PaymentsStripe
- EmailResend
- HostingVercel
- This app · Request proxyEdge streaming proxy for LLM APIsRequests must stream through low-latency routes with provider adapters, beyond normal page serving.
- This app · Secret storageEncrypted key vault (KMS-backed)Provider API keys need envelope encryption, not plain database columns.
RisksAI inference
- PlatformProviders may change terms, rate limits or key rules, and some (like Qwen per the material) have awkward sign-up, which could limit which models can be connected.
- CommercialDevelopers may prefer calling providers directly or writing a small router themselves, so the paid value has to be the cost view and rule testing.
- TechnicalSitting in the request path adds latency and becomes a point of failure; streaming and provider API differences take real effort.
- DataStoring provider keys and logging prompts is sensitive. Prompt content may need to be off by default, with only metadata kept.
- AIRules based on task type depend on classifying prompts, which can misroute hard tasks to weak models and hurt output quality.
Everything marked AI proposal or AI inference was written by a model from the evidence on the opportunity page. The proposed name must be checked against trademark registers before it is used.