tools
Contextual AI Platform
Crea e distribuisci agenti di IA che cercano tra documenti e dati aziendali, estraggono informazioni e producono risposte con citazioni delle fonti.
In inglese
Contextual AI Platform helps teams build agents and workflows for technical and other knowledge-intensive tasks. It connects to documents, databases, media, APIs, and MCP tools; extracts structured information; searches and reranks relevant content; and generates answers with citations and document bounding boxes.
Developers can use APIs, SDKs, or a visual builder, while organizations can deploy through Contextual SaaS, dedicated cloud environments, or private VPCs. Pricing is usage-based for on-demand access, while Enterprise plans use custom pricing. Structured data ingestion and continuous ingestion require contacting sales.
Features
- Build dynamic agents and static workflows with prompt-based or drag-and-drop tools
- Connect enterprise data through managed connectors, MCP, APIs, or file uploads
- Process PDFs, HTML with images, charts, tables, databases, spreadsheets, and CSVs
- Extract structured information from complex documents using agentic extraction
- Retrieve, rerank, filter, and generate grounded answers from enterprise data
- Provide citations and visual bounding boxes linked to source documents
- Evaluate groundedness, collect feedback, annotate errors, and optimize agents
- Use Python, TypeScript, and JavaScript integrations plus platform APIs
Use cases
- Answer technical customer questions using datasheets, call logs, and product data
- Diagnose device errors and anomalies in large technical log files
- Research intellectual property conflicts and regulatory compliance gaps
- Search scattered enterprise knowledge for internal teams
- Generate qualification reports and requirements traceability matrices
- Extract and analyze information from data-room documents
Pros
Cons
Latest updates
- Making Search Agents Faster and Smarter
Reranker improvements and RL-trained planners optimize search agents for speed and accuracy, reaching 60.7% accuracy.
- Introducing Agent Composer—AI for when it IS rocket science
Cut hours of specialized, technical work down to minutes with Agent Composer: Expert AI for complex engineering.
- An Agentic Alternative to GraphRAG
The new Metadata Search Tool is an agentic alternative to GraphRAG. It solves RAG reference traversal and unlocks powerful multi-hop reasoning.
Capabilities
- Choice of models — “Model selection for each workflow step” source
- Self-hosted — “Choose between multi-tenant SaaS, dedicated cloud instances, or private VPC deployment on your preferred cloud platform” source
- API — “Powerful APIs for the entire development lifecycle” source
- Official SDKs — “Integrate your existing SDKs” source
- Runs models for you — “The most grounded large language model in the world, engineered specifically to minimize hallucinations” source
- Builds agents and workflows — “Utilize an intuitive drag and drop GUI to build both dynamic agents and static workflows.” source
- Search over your data — “Gets relevant docs from knowledge base” source
- Traces and evaluates — “Built-in evaluation tools to assess responses for equivalence and groundedness” source
Get it
Security
Pricing
- Prices checked
- 2026-10-27
Enterprise
Price on request
- Custom pricing to meet performance and security needs
- Contact sales
- Unlimited workspaces
Parse
- $3 / 1,000 pages
- $40 / 1,000 pages
- Basic (text only)
- Standard (multimodal)
- Multi-stage document understanding pipeline for converting unstructured content into AI-ready formats
Rerank
- $0.05 per million tokens
- $0.02 per million tokens
- Rerank-v2
- Rerank-v2-mini
- Instruction-following reranker
Generate
- $3 per 1M tokens
- $15 per 1M tokens
- Input
- Output
- Model engineered to minimize hallucinations
LMUnit
- $3 per 1M tokens
- Input
- Evaluation-optimized model for preference, direct scoring, and natural language unit test evaluation