You're not overpaying for your LLM.
You're overpaying for your retrieval.
100x fewer tokens.
Zero retry loops.
One API call.
✦ 200,000 queries free — enough to ship to production
Continue with Google →No credit card. No sales call.
Live in 10 minutes. MCP-compatible.
Works with 
MCP-compatible — add Shaped to Cursor or Claude in 5 minutes
The problem isn't your LLM. It's what you're feeding it.
Your stack today
User Query
↓
OpenAI Embeddings
API call #1
↓
Pinecone
200 candidates
↓
Cohere
API call #2
↓
LLM
receives context
↓
Response
4 services + Redis for memory. ~4,000 lines of glue code.
Vector databases retrieve documents.
Shaped retrieves decisions.
Shaped retrieves decisions.
With Shaped
User Query
↓
Shaped
embedding + retrieval + reranking in one call
↓
LLM
receives precise context
↓
Response
One API call. <50ms.
Your stack today:
OpenAI Embeddings + Pinecone + Cohere + LLM + ~4,000 lines of glue code
With Shaped:
One API call. <50ms.
Your current stack:
50,000 tokens · $0.50/query · 800ms
With Shaped:
2,100 tokens · $0.01/query · 38ms
What this looks like in practice
One query replaces your entire stack.
WITH SHAPED
Agent Request
"How do I cancel my subscription?"user_id: "user_892"
Python SDK
from shaped import RankQueryBuilder, TextSearch
results = (
RankQueryBuilder()
.from_entity('item')
.retrieve(TextSearch(
input_text_query='cancel subscription',
mode={
'type': 'vector',
'text_embedding_ref': 'content_embedding'
},
limit=50
))
.score(
value_model='click_through_rate',
input_user_id='user_892'
)
.limit(10)
.build()
)1.docs/billing/annual-plan-cancellation.md0.97
2.docs/billing/prorated-refund-policy.md0.94
3.docs/retention/high-usage-offers.md0.91
4.docs/account/data-export-before-cancel.md0.87
5.docs/billing/cancellation-confirmation.md0.83
WHAT THE AGENT RETURNS:
You’re on an Annual Pro plan (3 months in).
Here’s what applies to you specifically:
- Cancellation takes effect at renewal
(9 months away)
- You qualify for a prorated refund: $180
- Based on your usage, consider our pause
feature — keeps your data, stops billing
See docs/billing/annual-plan-cancellation.md WITHOUT SHAPED
Agent Request
"How do I cancel my subscription?"
Python + Pinecone + Cohere
import pinecone
import cohere
from openai import OpenAI
# 1. Embed user query
res = client.embeddings.create(
input=query,
model="text-embedding-3-small")
query_vec = res.data[0].embedding
# 2. Vector search (200 noisy results)
idx = pinecone.Index("my-index")
raw_docs = idx.query(
vector=query_vec,
top_k=200,
include_metadata=True)
# 3. Rerank with static model
co = cohere.Client('API_KEY')
reranked = co.rerank(
query=query,
documents=[d.metadata['text']
for d in raw_docs['matches']],
top_n=10)
# 4. Stuff context and hope
context = "\n".join(
[doc.document['text']
for doc in reranked.results])
prompt = f"Context: {'{context}'}\n\nAnswer: {'{query}'" 50,000 TOKENS OF CONTEXT 90% irrelevant. $0.50/query. Static forever.
WHAT THE LLM RETURNS:
To cancel your subscription, go to
Settings → Billing → Cancel Plan.
Cancellation policies vary by plan
type. Please review our terms of
service for details on refunds...
// Wrong plan type referenced.
// Refund amount not calculated.
// User’s usage history ignored.
// Agent retried 3x. Total: 2,340ms.10 results · 2,100 tokens · 38ms · Your agent gets exactly what it needs — nothing more.
Works alongside your existing stack. No migration required. Live in 10 minutes.
Outdoorsy replaced their entire retrieval stack in 30 days.
+11%
Search conversion lift
2,000 → 30
Lines of code
< 30 days
Time to production
“Shaped became the obvious choice.”
— Han Yuan, CTO, Outdoorsy
SidelineSwap +9% purchases · Overlap +13% engagement · AfterHour +8% CTR
.png){kind=logo}
Start retrieving decisions,
not documents.
200,000 free queries.
No card. Live in 10 minutes.
Free to start. Scales with usage.
No surprise bills.