Step 1
Choose a template
Start from a product idea, a generated prototype, or a known workload shape.
How It Works
Plan first, then add routing where it matters.
Spendplane helps you decide what stays local, what uses cost-effective cloud capacity, and when premium assist delivers value. The workflow below shows how teams make these decisions before production spend starts.
Template-first planningModel and GPU rate inputsLocal plus cloud routing profileControls when teams scale
How it works
From estimate to execution in four steps.
Start with a clear plan, then add managed routing and controls only where they earn their place.
Step 2
Price the stack
Estimate model, GPU, hosting, auth, storage, and job costs in one pass.
Step 3
Set the pace
Choose how much local compute, budget cloud, and premium assist should contribute.
Step 4
Run with Spendplane
Keep the plan free, or turn on Hybrid routing and controls when the project needs them.
Ingress
Requests enter Spendplane with tenant policy, budget, and runtime context attached.
Policy
The proxy checks routing rules, privacy mode, helper lane, and execution profile before egress.
Execution
Spendplane chooses the local or cloud lane, records the decision, and keeps usage within plan.
Execution flow
How routing decisions happen in real time
Every request is evaluated against policy, privacy settings, and runtime capacity before Spendplane picks the final execution lane.
How the Proxy Enforces Optimal Execution
1
Plan Creation
The planner analyzes your project requirements and creates a detailed execution profile with token budgets, model preferences, and routing policies based on analysis of thousands of past projects.
2
Proxy Interception
Every API request passes through our proxy. The proxy checks the execution profile, tracks token usage against budget, and enforces routing decisions in real-time.
3
Efficiency
The proxy automatically routes to cost-effective models when possible, switches to premium only when needed, and blocks when budget is exhausted. Your project stays on track and on budget.
Result: Projects can complete up to 20-40% faster while typically staying within 10% of estimated cost.
Supported pathsWorks after Replit, Loveable, Bolt, and v0Firebase and Vercel awareBudget and premium cloud readyLocal GPU awareOpenAI-compatible execution
Free planner first
Estimate project cost, delivery time, and stack requirements before deciding whether to run through Spendplane.
Works with existing builders and providers
Start from a prototype or a fresh project brief, then blend your current stack with efficient execution paths.
Built for full-project cost control
Model hosting, auth, database, queues, model usage, and GPU runtime instead of thinking about inference in isolation.
Device-aware by default
Estimate when local hardware can absorb background work and when cloud assist is worth paying for.
Use the planner for free, then add Hybrid routing and spend control only when you want Spendplane in the loop.
Get started
Start with Planner. Add Spendplane to the live path later.
Get the plan first. Turn on Hybrid routing or team controls when the project is ready.
What happens next
Pick a template or bring a project outline
Choose the right local and cloud mix
Turn on Hybrid routing or Enterprise controls when you need them