Adaptensor Documentation¶
Welcome to Adaptensor β the Private Document AI Platform with AdaptHex compression.
What is Adaptensor?¶
Adaptensor is a private document AI platform that lets you:
- Upload documents (PDF, Word, TXT, HTML)
- Index them with TPU-accelerated embeddings
- Search semantically across all your content
- Query via API or Python SDK
All while keeping your data 100% private β no external API calls, no data sharing.
Why Adaptensor?¶
π Complete Privacy¶
Your documents never leave your cloud. Zero external API calls. Full data sovereignty for regulated industries.
β‘ AdaptHex Compression¶
Our proprietary hexadecimal quantization reduces vector storage by 4-8x while maintaining 99.6% accuracy.
π° Predictable Pricing¶
Simple pay-as-you-go: $0.10/1K queries, $0.50/GB storage, $1.00/10K chunks indexed.
π TPU Accelerated¶
Built on Google Cloud TPUs with our AdaptCoreβ’ middleware for 10x faster indexing than GPU alternatives.
Quick Example¶
from adaptensor import Adaptensor
# Initialize client
client = Adaptensor(api_key="sk_live_...")
# Upload a document
client.upload("contract.pdf")
# Search semantically
results = client.query("What are the payment terms?")
for result in results:
print(f"{result.score:.2f}: {result.text}")
Platform Architecture¶
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β Your Application β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β Adaptensor SDK / API β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β AdaptCoreβ’ β AdaptHexβ’ β AdaptLLMβ’ β
β TPU Middleware β 4-8x Compression β TPU Inference β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β Google Cloud TPUs β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Supported File Types¶
| Format | Extension | Max Size |
|---|---|---|
| 50 MB | ||
| Word | .docx | 50 MB |
| Text | .txt | 50 MB |
| HTML | .html | 50 MB |
| Markdown | .md | 50 MB |
Performance Benchmarks¶
| Metric | Adaptensor | Industry Average |
|---|---|---|
| Indexing Speed | 762 chunks in 170s | 762 chunks in 1700s+ |
| Query Latency | 18ms | 100-500ms |
| Storage Efficiency | 4-8x smaller | Baseline |
| Accuracy | 99.6% | Baseline |
Next Steps¶
- Create an account and get your API key
- Follow the Quick Start guide
- Upload your first document
- Start querying!
Need Help?¶
- π§ Email: support@adaptensor.com
- π¬ Discord: Join our community
- π Issues: GitHub