Now in public beta
Expert agents.Built in minutes.
Deploy AI agents that perform like domain specialists. In your cloud, or ours.
How it works
Collections
Your document collections and knowledge bases
legal_contracts
Master service agreements, NDAs, and procurement terms.
247 docs
nda_library
Ready92 docs
vendor_agreements
Ready137 docs
compliance_policies
Ready58 docs
Add documents, URLs, or raw text. Content is parsed, chunked, and indexed automatically.
Trusted by teams in production
Enterprise-grade AI, shaped by the real world.
Our teams explored methods for tackling some of the industry's toughest data challenges:
- Extraction reliability across highly complex and variable document layouts.
- AI retrieval that accurately navigates dense medical content.
- Rule-mapping that translates complex business content into clearer operational guidelines.
Opifiny strictly deploys AI capabilities within Opifiny-controlled infrastructure.
“With Knowledge², our customers go from scoping to a working proof-of-concept in minutes, not weeks. Their technical and business teams move in lockstep instead of stalling on infrastructure. The ability to deploy within regions that have strict data residency requirements removes the last blocker to adoption.”
“Knowledge² lets us swap the underlying LLM on any agent without touching the retrieval layer or application code. We benchmark models side by side on real customer data and ship the best fit, per workflow, in the same afternoon.”
For developers
From zero to deployed agent in minutes.
Everything you need is in the SDK. No pipelines, no glue code, no infra to manage.
Crawl a site, build an expert.
1# uv pip install knowledge22from knowledge2 import Knowledge23import httpx, xml.etree.ElementTree as ET45k2 = Knowledge2(api_key="k2_...")6project = k2.create_project("Site Expert")7corpus = k2.create_corpus(project["id"], "docs.example.com")89# Crawl sitemap and ingest every page10sitemap = httpx.get("https://docs.example.com/sitemap.xml")11urls = [loc.text for loc in ET.fromstring(sitemap.text).findall(".//{*}loc")]12k2.ingest_urls(corpus["id"], [{"url": u} for u in urls])13k2.sync_indexes(corpus["id"], wait=True)14# 1,200+ pages indexed1516resp = k2.search_generate(17 corpus["id"],18 "What are the rate limits for the batch API?",19 top_k=8,20 hybrid={"enabled": True},21)22print(resp["answer"])23# Cited answer grounded in docs.example.comBuilt for production. Built for trust.
What makes AI deployable in regulated environments.
Evidence attached
- Every output carries passage-level source references. If the evidence is not there, the agent says so.

Current by design
- Your retrieval stays current as content changes. No manual re-embedding or pipeline rebuilds.

LLM flexibility
- Swap LLM providers per request without changing application code. Anthropic, Google, OpenAI.

Pricing
Start free. Go Pro when you're ready.
Free
For developers evaluating the platform.
- 5,000 queries / month
- 500 documents
- 3 collections
- 2 agents
- Community support
Pro
$187 / mo, billed annually
For teams building production RAG applications.
- 100,000 queries / month included
- 50,000 documents
- 20 collections
- Unlimited agents
- Adaptive reranking
- Retrieval tuning and evaluation
- Pipelines and webhooks
- $0.80 / 1K queries overage
- Priority support
Enterprise
Dedicated infrastructure, forward deployed engineers, and enterprise controls.
- Custom query and document limits
- Private VPC or on-prem deployment
- Graph RAG and fine-tuning
- A2A (Agent-to-Agent) protocol
- SSO and audit logging
- Forward Deployed Engineer embedded with your team
- Custom SLA
- Premium support with dedicated CSM
Feature comparison
| Search and retrieval | Free | Pro | Enterprise |
| Semantic search | |||
| Keyword search (BM25) | |||
| Hybrid search (RRF fusion) | |||
| Metadata filtering | |||
| Adaptive reranking | |||
| Retrieval tuning and optimization | |||
| Graph RAG | |||
| Ingestion and processing | Free | Pro | Enterprise |
| PDF, Office, HTML, Markdown, CSV, images | |||
| Automatic chunking and embedding | |||
| Near-duplicate detection | |||
| Document analysis and evaluation | |||
| Fine-tuning | |||
| AI and agents | Free | Pro | Enterprise |
| RAG answer generation | |||
| Knowledge agents | 2 | Unlimited | Unlimited |
| Pipelines | |||
| Webhooks | |||
| A2A protocol | |||
| Platform and support | Free | Pro | Enterprise |
| Dashboard and ops UI | |||
| Python and TypeScript SDKs | |||
| REST API | |||
| SSO (SAML / OIDC) | Coming soon | ||
| Audit logging | |||
| VPC / on-prem deployment | |||
| Forward Deployed Engineer | |||
| SLA | Custom | ||
| Support | Community | Priority | Premium |
Your first agent is twenty minutes away
Upload a document, configure retrieval, ask a question. Evidence-backed answers in minutes.