AI Assistants
Assistants your team and your customers can trust: grounded in your documents and records, aware of permissions, able to take approved actions — in chat, in your tools, or on the phone.
Same engine — packaged for a faster start
What it is
An AI assistant is the front door to everything else the stack does: a conversation surface where your team or your customers ask in plain language and get answers grounded in your actual documents, tickets, and records — with citations. Beyond answering, the assistant can act: create the ticket, draft the reply, update the record — through governed tools, with approval gates where the action carries risk.
Why it matters
Generic chatbots answer from the model's training data; an enterprise assistant answers from yours. The difference is retrieval, permissions, and grounding: an assistant that cites its sources, respects who is allowed to see what, and declines what it doesn't know is one you can put in front of customers.
What we build
One assistant architecture, deployed where the conversation happens — an internal tool, an embedded website widget, or a full-duplex voice line.
- Retrieval over your documents, tickets, code, and records — RAG with citations
- Permission-aware answers: the assistant cannot leak what the asker cannot see
- Actions through governed tools (MCP), with human approval on risky steps
- Per-user memory that personalizes without drifting behavior for everyone
- Chat, embedded widget, or full-duplex voice on your phone line
- Bilingual EN/FR from day one
What it works with
Reads from the LLM-Ready Knowledge Base and the data catalog beneath it. Calls tools through the MCP Tool Registry. Runs on the Agent Runtime with the same traces, evals, and approval gates as every workflow. Customer-facing variants ship through Embedded Agents; the packaged expressions are the Concierge and Sidekick product lines.
When you need it
Signals: your team answers the same questions weekly from tribal memory; customers wait on answers that already exist in your documentation; internal data is quietly being pasted into consumer chatbots because there is no sanctioned alternative. An assistant with your data and your rules is usually the first AI system a company should own.
Related learning
A company knowledge base built so an AI system can cite real answers from it — sourced from documents, tickets, code, conversations, and structured records; chunked, embedded, permissioned, evaluated, and kept fresh on AWS.
An AI agent that lives inside your website, app, or voice line — with the persona, knowledge, and tools you approve, and a real outcome for every conversation that doesn't end in a clean answer.
AI agents that act on business events with the tools, memory, approvals, and traces you can audit. Workflows that ship to production, not just demos.
The pattern where an AI agent retrieves relevant context from your data before generating an answer — instead of relying only on what the model learned during training.
How an AI agent remembers the user it serves — what they said before, what they prefer, what context not to repeat — without that memory drifting the agent's behavior for everyone else.
AI agents you talk to — streaming speech-to-text, an agent turn, and streaming text-to-speech composed into a full-duplex loop where interruptions and latency are measured in milliseconds.