Opportunities › API rate limits and pricing optimization for enterprises › Proposed app
Spillway
One gateway that routes LLM traffic across providers to avoid limits and cut spend
- Score
- 38
- confidence 37
- Evidence
- 2
- items on the page
- Rivals
- 1
- named in sources
- First version
- 45 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: Engineering teams configure routing rules, set budget caps, and review spend dashboards—dense, keyboard-heavy infrastructure work done from a desk. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
Headroom
All pools healthy68%
About this appAI proposal
Spillway is a gateway for teams that run many AI workloads on several model providers. You bring your own provider accounts and keys. Spillway sits in front of them, sends each request to a model that has headroom and fits the budget, and moves traffic over when a limit is hit.
Platform and finance leads get one view of spend and rate-limit headroom per team, workload and provider. Caps and routing rules are set in a dashboard, not in each service's code. Engineers change one base URL and keep their existing client libraries.
It differs from using a single cloud's model catalogue: the rules span providers, so a limit or price change at one place becomes a routing decision, not an outage or a surprise invoice.
- Tagline
- When one provider says wait, your traffic keeps moving.
- Voice
- PlainPreciseCalm
- Why this name
- A spillway lets excess water leave safely instead of breaking the dam. Spillway does that for request volume that exceeds one provider's limit.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Limit-aware routing
Tracks remaining requests and tokens per provider key and shifts traffic to another model before a request fails.
No member yet
Fallback chains
Set an ordered list of models per workload. On a 429 or timeout, the next one answers the same prompt.
No member yet
Budget caps per team
Daily and monthly spend caps per team or workload. Soft alerts first, then a hard stop or a downgrade to a cheaper model.
No member yet
Cost-aware model rules
Send simple tasks to cheaper models and hard ones to stronger ones, using rules you write and can inspect.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#0F3D47Deep Channel · primary#F5F2EAPaper · surface#1B2A2FInk · text#E8A33DSignal Amber · accent#2E9AA6Flow Teal · data#C4492FAlert Clay · warning
Typography
Aa
Spillway
PricingAI proposal
The material gives no competitor prices. These tiers are a flat-fee starting point to test against a share-of-spend model; providers are billed to the customer's own accounts.
Team
$499 a month
- Up to 3 provider accounts
- Routing, fallbacks and budget caps
- Dashboard and CSV export
- 30-day audit log
Business
$1999 a month
- Unlimited provider accounts
- Pooled keys across regions
- SSO and role-based access
- 1-year audit log
- Priority support
Enterprise
Contact us
- Volume pricing on routed spend
- Private deployment option
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. Spillway runs inside a company's backend and its end users never see it. Growth comes from sales, content on rate limits and cost control, and engineer-to-engineer word of mouth.
- Do they become customers?
- Teams that see it are already buyers or influencers. Test whether a platform lead who trials it on one workload expands to other teams within a quarter.
- First customers
- Platform engineering leads at companies with 500 to 5,000 staff that already use two or more LLM providers and have hit rate limits in production.
- Where to reach them
- Hacker News and engineering blogs: Publish a practical write-up on surviving rate limits, with a free routing-rule template and a calculator.
- Direct outreach on LinkedIn: Message platform and AI infrastructure leads who have posted about limits or LLM cost.
- AI engineering meetups: Run live demos of failover under load and collect trial sign-ups.
- First milestone
- Five paying teams, each routing a production workload through Spillway for 30 days.
Why people already pay for this
- Public claimLarge enterprises face rate limits and high costs with Claude and OpenAI that drive them to alternative providers
- Public claimEnterprises are actively provisioning alternative LLM providers to solve the problems of rate limits and pricing
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 · Gateway runtimeLow-latency edge or container proxy serviceStreaming LLM requests need a long-lived, low-overhead proxy that Vercel functions do not suit well.
- This app · Metrics storeTime-series databasePer-request token, cost and limit counters need fast aggregation for dashboards and caps.
RisksAI inference
- PlatformProviders may restrict resale or proxying of keys, or change limit headers and terms. Pooling accounts could breach some provider terms; this needs a legal check.
- CommercialLarge buyers may prefer a cloud vendor's own catalogue or build a gateway in-house. Demand rests on public claims, not yet verified.
- TechnicalAdding a proxy hop adds latency and a new point of failure; enterprises will expect high uptime and streaming support.
- DataPrompts pass through the gateway. Enterprises may demand no-logging modes, regional hosting and security reviews before trial.
- TechnicalModels differ in output and tool-call formats, so fallbacks may change results. Per-workload testing is needed.
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.