Your AI app doesn't have a seatbelt. Prune is one line away.
One URL change gives you security, spend control, verifiable receipts, and lower costs on every AI call.
The pain
Burning money on AI. Leaving keys in .env.
Waste hits the bill: repeats, overkill models, agents with no ceiling. Risk hits the stack: provider keys in env files, no firewall, nothing signed when something goes wrong.
from prune import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="gpt-5.5",
messages=[{"role": "user", "content": prompt}],
)Toggle what Prune applies
Watch the receipt update below ↓
Signed receipt
About 33% cheaper on this walkthrough. Toggle routing above and watch the math move.
Every live response carries the same shape. Verify a signed file.
Full receipt preview (copy for this walkthrough)
{ "model": "claude-opus-4-8", "choices": [{ "message": { "content": "{ ... }" } }], "usage": { "prompt_tokens": 684, "completion_tokens": 412 }, "prune_metadata": { "cache_hit": false, "cost_without_prune_usd": 0.460, "cost_with_prune_usd": 0.310, "unique_path_cost_saved_usd": 0.150, "cache_path_cost_saved_usd": 0, "tokens_saved": 3400, "optimizations_applied": [ "model_routing", "template_opt", "output_cap:512" ], "pii_shield": { "active": true, "fields_masked": ["email"] }, "receipt_signature": "kQ7vN2LxTz8mYw4pRc…hZ3w==", "receipt_signing_key_id": "prune-ed25519-2026-01" }}Stop paying full price for every AI call.
Asked it before? Reuse the answer. Easy ask? Use a cheaper model. Hard ask? Keep the strong one.
What a call looks like
Exact hit. Semantic hit. Routed cheaper. Or live when it must be.
Same shape as the Requests view in app.withprune.com.
Save. Shield. Prove.
Same SDKs and models. Cut waste on the way out, protect the key, stamp a receipt on the way back.
Cache hits and cheaper routing
Near-duplicate questions skip the model even when the wording changes. Simple calls route to cheaper models when safe. Exact repeats cache. You do not need a flood of identical traffic to see the bill move.
Vault the key. Block the junk.
Provider keys stay encrypted server-side, never in your app or repo. Personal data, secrets, and injection attempts get blocked before OpenAI, Claude, or Gemini, and responses are checked for leaks. Optional per-key spend caps stop runaway loops before the provider call.
A signed receipt on every response
Every call returns a signed receipt with cost, model, and what fired. Finance and auditors can verify the numbers instead of taking a dashboard's word for it.
Works with OpenAI, Claude, Gemini, OpenRouter, Bedrock
Early access. Not SOC2 yet. Keys encrypted at rest, isolated by account, checked before they leave your stack. How we handle trust.
What early builders said
Real notes from early access. Names used with permission.
We plugged Prune into FocusAI and it just worked. Our AI features feel faster on repeat use, costs are easier to manage, and we didn't have to rebuild everything. It feels like having a smart layer behind the scenes that quietly makes the product better.
AI productivity app
Simple pricing. You bring the keys.
BYOK. Platform fee only. Early access runs on Free. Paid tiers below are planned packaging when checkout opens.
Free
$0
15K req/mo or 14 days from first Optimize receipt
Starter
Planned$15/mo
250K requests/mo
Pro
Planned$49/mo
1M requests/mo
Team
Planned$149/mo
5M requests/mo
Planned packaging from docs. Enterprise custom.
Common questions
What is Prune?+
Prune is a managed AI gateway. Change one base URL so every call to OpenAI, Claude, Gemini, Bedrock, or OpenRouter goes through Prune - for unique-path cost control, Shield security, and a receipt on every response. Cache helps on repeats; it is not the product thesis.
How is Prune different from cache-focused AI gateways?+
Most AI gateways lead with cache hit rate. Prune leads with unique-path savings (routing, output caps, prompt cleanup) that apply when the prompt never repeats, plus Shield (vaulted keys, PII/secret checks, spend caps) and a machine-readable receipt. Cache is measured separately as upside.
Does Prune only help when prompts repeat?+
No. The non-negotiable thesis is that Prune must still win on unique traffic. Caching and semantic reuse are multipliers on top, shown separately in prune_metadata.
Is Prune secure?+
Provider API keys are stored in an encrypted vault (AES-256-GCM). Shield can block PII, secrets, and runaway spend before the provider call. Prune is in early access and is not SOC2-certified; evaluate architecture, not badge theater.
How long does it take to integrate?+
Typically under five minutes: create a Prune API key, vault your provider key, point your SDK at Prune, send one request, and read the receipt.
One base URL. Lower bill. Signed proof.
Early access is open. Point your SDK at Prune, send one request, and see cache, routing, and the receipt for yourself.
Get early access