More Than a General-Purpose Chatbot
A general AI assistant can produce fluent HR guidance. Lisa is designed for a different job: helping a manager work through an HR situation inside a controlled system that can ask for context, apply organizational boundaries, preserve evidence, and involve people when the stakes rise.
Context before conclusions. Lisa can clarify the situation and use authorized organizational context rather than assuming that a generic answer fits.
Evidence that can be inspected. When approved retrieval tools are available, the system records which sources and tool results supported the response.
Human judgment remains central. Lisa supports decisions; it does not replace HR, Legal, or an authorized decision-maker.
What Is Built Today
The current production foundation focuses on governed conversations, traceability, and conservative handling of higher-risk situations. Availability can vary by organization, configuration, permissions, and the evidence approved for use.
- Manager-facing HR conversations with structured context gathering
- Organization-scoped access and configurable feature boundaries
- Risk classification and escalation records for human follow-up
- Governed model selection with server-controlled configuration
- Approved evidence retrieval when enabled for the organization
- Source receipts, tool traces, and per-turn provenance
- Evaluation workflows used to compare candidate changes before release
- Lisa is decision support, not legal advice or an autonomous employment decision-maker
- Higher-risk employment actions require appropriate HR or Legal review
- Evidence coverage varies; an unavailable or insufficient source should not be treated as proof
- Planned integrations are not represented as available until implemented and verified
- The current environment is approved for synthetic and demonstration data only; do not enter real employee data or PHI
- Any real-data pilot requires deployment-specific privacy, security, access, contractual, and provider review
A Governed Path From Question to Response
Lisa’s architecture separates the conversational experience from the controls that determine what context can be used, what evidence supports a claim, and when a person needs to take over.
Server-side controls are intended to reduce unsupported claims and unsafe overreach. They are safeguards, not guarantees: language, evidence, configuration, and real-world facts can still be incomplete or misunderstood. That is why the system preserves review paths instead of treating an AI response as final authority.
Governance Is the Differentiator
The model can change. The accountability framework remains.
Lisa is designed so that model choice, evidence access, organization context, risk handling, and review records are governed outside the model itself. That makes it possible to improve the underlying technology while retaining explicit controls and an inspectable operating history.
What Lisa Does Not Promise
Responsible HR technology should be clear about its boundaries.
Lisa Is Not
- A substitute for qualified HR or Legal review
- An autonomous hiring, compensation, discipline, or termination decision-maker
- A guarantee that every factual claim, risk signal, or escalation will be correct
- A claim that every planned data connection or workflow is currently available
Deployment Requires
- Organization-specific permissions and feature configuration
- Review of the data classifications and providers involved
- Verification of the evidence sources approved for the use case
- Named human owners for escalation and higher-risk decisions
Building Toward Deeper Organizational Context
The product direction is to make authorized company context, policies, evidence, and review workflows easier to use together—without relaxing the governance required for sensitive HR work.
Future integrations and capabilities remain roadmap items until they are implemented, tested, documented, and released. Model changes likewise require evaluation evidence and human approval.
Why Lisa: not because an AI system should make employment decisions, but because managers and HR teams need a more structured, evidence-aware, and accountable way to prepare for them.
This document will be updated as the product evolves.