www.sciphi.ai Open in urlscan Pro
2606:4700:3036::ac43:beb5  Public Scan

Form analysis
0 forms found in the DOM

Text Content

Skip to content
Please leave the R2R framework a star 🤩

SciPhi
FeaturesUse CasesPricing

Search

Demo
Documentation
R2R DocsLogin

Build and deploy serverless RAG
Effortlessly build, deploy, and scale Retrieval-Augmented Generation systems
Focus on AI innovation while SciPhi helps you with the infrastructure
Start for FreeRead the Docs



Configure
Configure your RAG pipeline to your exact needs using the config.json.

Specify providers, adjust settings, and optimize performance with ease to build
powerful tailored RAG applications.
Customize
Customize your R2R RAG pipeline directly by modifying the underlying code.

Override default implementations, add custom logic, and seamlessly integrate
your own components to create specialized RAG solutions that perfectly align
with your unique requirements.
Deploy
Deploy production-ready RAG pipelines with just one click.

Leverage the power of the cloud to automatically scale your pipeline based on
demand. Go from experimentation to real-world RAG implementation in record time,
without the hassle of complex infrastructure setup.
Evaluate
Evaluate and monitor each pipeline performance with built-in capabilities.

Choose from multiple selected evaluation providers to measure quality with a
specified frequency. Identify areas for improvement and make data-driven
decisions to optimize your RAG solution.




Multi-User Q&AWeb RAGRAG Chatbot


FLEXIBLE DOCUMENT INGESTION

Default support for html, pdf, docx, and other document types


ANY VECTOR DATABASE

Scalable and configurable cloud vector database support. Hosted by SciPhi or
remote cloud (e.g. qdrant, pgvector,...).


MULTIPLE LLMS

Select any LLM provider for your embeddings and completions, from TogetherAI to
OpenAI.




PLUG-IN INTEGRATION

Seamlessly connect with third party data retrieval sources, including web
search.


CUSTOM ADAPTABILITY

Fully customize your RAG pipeline by integrating plug-ins with your proprietary
data to tackle complex scenarios.


DYNAMIC SCALABILITY

Effortlessly scale your RAG pipeline to handle increasing data volumes and
complex queries with cloud-native technologies.




CONTEXT-AWARE RESPONSES

Equip conversational agents with the ability to retrieve contextually relevant
information, enabling them to provide accurate and informative responses during
user interactions.


ENHANCED USER EXPERIENCE

Elevate user satisfaction by empowering chatbots and virtual assistants to
deliver high-quality, personalized support by leveraging external knowledge
sources.


INTELLIGENT ASSISTANTS

Transform conversational AI systems into intelligent assistants capable of
handling complex inquiries by augmenting their responses with retrieved
information.



class ChatbotRAGPipeline(RAGPipeline):
  # A basic RAG Chatbot pipeline, implemented in R2R


    def transform_query(self, query: str) -> str:
        message_payload = json.loads(query)
        return self._query_to_formatted_conversation(message_payload)


    def construct_prompt(self, inputs: dict[str, Union[str, list]]) -> str:
        # Construct a prompt for the chatbot
        ...  


    def run(self, query: str, generation_config: GenerationConfig) -> str:
        conversation = self.transform_query(query)
        search_query = self.generate_search_query(conversation)


        if search_query:
            search_results = self.search(search_query, filters={}, limit=5)
            prompt = self.construct_prompt({
                "conversation": conversation,
                "search_query": search_query,
                "search_results": search_results,
            })
        else:
            prompt = self.construct_prompt({"conversation": conversation})


        return self.generate_completion(prompt, generation_config)

|



NOT JUST YOUR STANDARD CLOUD OFFERING.

SciPhi combines cutting edge technology with scalable infrastructure
to enable accelerated development and deployment of state-of-the-art RAG
systems.

Read the docsTry the search demo


EASY CONFIGURATION


Choose from a multitude of vector database, LLM, and other providers with just a
json.


TOTAL CUSTOMIZATION


Design your pipeline - from custom embedding chunks to output prompts - or stick
to our defaults.


VERSION CONTROL


Automatic deployment and versioning provided via direct GitHub link.


CLOUD RUN


Deploy directly to the cloud and let SciPhi reliably manage your backend. Scale
up or down as needed.


FAST DEPLOYMENT


Deploy your first pipeline in minutes - not days with just one click.


SELF HOST


Use Docker to run SciPhi on your own infrastructure without hassle.


DOCUMENTATION


Comprehensive docs cover everything from setup to advanced usage, with detailed
guides.


OPEN SOURCE


Fully open source - powered by R2R, a comprehensive RAG framework from
experimentation to production.


COMMUNITY


SciPhi is supported by a large community of serious LLM application developers.


IT'S EASY TO BUILD A PROTOTYPE RAG PIPELINE —
IT'S HARD TO DEPLOY ONE THAT KEEPS UP WITH YOUR USERS

We spoke with literally hundreds of founders in the AI space and were surprised
to find most of them solving different aspects of the same problem from scratch.
Whether it was deployment or optimization, RAG was the most top of mind.

With SciPhi+R2R, building the best RAG system isn't so hard or confusing. Start
with the basic RAG pipeline and use the platform's observability and deployment
to iterate quickly when things start going wrong.

Owen Colegrove

Founder of SciPhi





DON'T JUST TAKE OUR WORD FOR IT


KEVIN TANG

Firebender (Ex-Two Sigma)

SciPhi cut our LLM costs, while also improving accuracy in responses. Support
has been phenomenal especially with expert guidance on improving/iterating our
RAG pipelines.


KEHINDE WILLIAMS

Shepherd (Ex-NVIDIA)

We use SciPhi to power help our students find relevant study resources and are
currently working with them to build out a multi-document RAG pipeline.




ANDREW WANG

GoldenBasis (Ex-Citadel)

Really enjoyed using SciPhi--I was able to set up a RAG to talk to dozens of
dense 100+ page PDF documents in just an hour.




WHY BUILD WITH SCIPHI + R2R?

R2R is supported by a thriving community of open source collaborators. With more
than 1,000 members in Discord and a direct line of access to the SciPhi team,
you can be sure your questions will not go unaswered for long.

pip install 'r2r[eval]'

|
Read documentation


SOLID FOUNDATIONS

R2R provides a strong, reliable foundation to build upon, with abstractions and
pipelines that have been proven.

The framework is designed to enable fast iteration and deployment.


PRICING FOR EVERY STAGE

Find the plan that works for you


FREE

Best for small projects.
Free

Max of 10 pipelines

Single developer

10,000 embeddings per pipeline

100,000 requests per month

Get Started


STARTUP

For startups and small teams.

$999

$499

Unlimited pipelines

Team workspace

Up to 1M embeddings

Up to 1M requests per month

Premium RAG Pipeline

Contact Us


ENTERPRISE

For larger organizations.

Custom

Everything in Startup, plus

Prioritized feature onboarding

On-prem deployment option

RAG pipeline consultation

Managed migration


Private beta access

Contact Us



SHIP YOUR FIRST APP IN MINUTES

Get Started Now
SciPhi

Effortlessly build, deploy, and scale Retrieval-Augmented Generation.
Focus on AI innovation while SciPhi helps you with the infrastructure

DocsR2R DocsContact