Platform, Agents & Technology
What is the Contextual AI platform?
The Contextual AI platform is a context layer for enterprise AI. You build agents that are grounded in your data and can retrieve information, reason over it, and take actions. Retrieval (search over your datastores, with citation) is a core capability of these agents, but they can also use tools (APIs, MCP, code execution), run multi-step workflows, and produce structured outputs. You build agents with Agent Composer (templates or custom workflows), connect data via datastores and connectors, and query via the UI or API.How does it differ from generic chatbots or LLM APIs?
Unlike generic LLMs, Contextual AI agents are wired to your private knowledge and can act on it. They retrieve relevant content from your datastores, cite sources, and—depending on the workflow—can call tools, run multi-step research, or execute tasks. That keeps responses accurate and aligned with your docs and policies while enabling agents to do more than single-turn Q&A. Agent Composer lets you control the full path: retrieval, reranking, tools, and generation.What are the core capabilities?
- Agents & Agent Composer — Build agents from templates (Basic Search, Agentic Search) or custom YAML/visual workflows. Agents combine retrieval (RAG) with multi-step reasoning, tool use (search, APIs, MCP), and cited or structured generation.
- Datastores & connectors — Ingest and sync content from local upload, GitHub, Confluence, Google Drive, SharePoint, Box, and more; incremental sync and published-content filtering where applicable.
- Parsing & chunking — State-of-the-art parsing for PDF, HTML, Markdown, and other formats; configurable chunking so agents can retrieve and reason over your content effectively.
- APIs & SDKs — Create agents, ingest documents, run queries, and manage evaluation/tuning via REST API and Python/Node SDKs.
- Security & governance — RBAC, audit trails, model armor, and compliance documentation (SOC 2, HIPAA, GDPR, etc.).
What does “context engineering” mean?
Context engineering is the practice of selecting, structuring, and delivering the right information to LLMs so that agent responses are accurate and grounded. On Contextual AI this includes: retrieval (search, rerank, filter over your datastores—a key part of every agent), orchestration (Agent Composer workflows, tools, conditional logic, multi-step actions), and prompt/config (system prompts, generation settings).Use Cases & Industries
Which industries are best suited for Contextual AI?
Contextual AI fits industries with large knowledge bases, complex documents (e.g. PDFs, specs, manuals, logs), and/or strong compliance needs. Typical fits include:- Engineering and manufacturing — Technical support agents, root cause analysis over device or process logs, and agents over manuals and specifications.
- Technical support — Documentation and support agents that answer from product docs, wikis, and runbooks with citations.
- R&D — Research agents over patents, trials, and internal knowledge; structured extraction and reporting.
- Financial services — Policy, research, and compliance-oriented agents over internal docs and regulations.
- Legal and professional services — Research and contract/document agents with citations and source control.
What are typical use cases?
- Documentation & support agents — Q&A over product docs, internal wikis, or policy; agents retrieve and cite sources so users can verify and share.
- Device log & root cause analysis — Multi-step agents that ingest logs, correlate with specs and process history, and produce investigation reports with ranked hypotheses (enterprise templates).
- Research and internal knowledge — Agents over Confluence, Drive, or SharePoint for onboarding, planning, and decision support; multi-step retrieval and synthesis.
- Task execution & planning — Agents that predict failures, recommend maintenance, or produce plans with structured outputs (enterprise).
- Structured extraction — Compliance evidence collection, incident and postmortem analysis, and other extraction workflows with defined output schemas (enterprise).
- Custom workflows — Agent Composer for domain-specific retrieval, tool use (APIs, MCP), and cited or structured generation.