How it works
Your business context.
Put to work.
Our UCE brings together the records, history, and rules each workflow needs. Lieutenant carries out the agreed steps; your team reviews exceptions and approves material actions.
Our Unified Context Engine
Data sources
- Email & Calendar
- Slack & Teams
- CRM Systems
- Documents & Drives
- Meetings & Transcripts
- AI Agents
- Grounded Search
- Exec Briefings
- Workflows
Illustrative example
An invoice arrives. More than an answer comes out.
A chatbot could explain the bill. A dashboard could show payables. In a configured Lieutenant workflow, our UCE supplies the records and rules so an agent can prepare the work and route the exception.
Trigger
A vendor bill arrives for one of your locations.
Context packet · 3 sources
- [1]Invoice and vendor record · Amount, location, and prior bills
- [2]Signed agreement · Payment terms and service scope
- [3]Approval policy · Location owner and review threshold
Prepared work and a held exception
- Matched the bill to the vendor and location; prepared a coded entry for review [1]
- Flagged a charge that differs from the signed agreement; attached the supporting records [1][2]
- Routed the exception to the location approver named in the policy; held the entry for a decision [3]
Illustrative example with sample data. Not a customer result or product screen. References help a reviewer check the work; they are not a guarantee of accuracy.
Why Lieutenant
Three things, working together.
Business memory, workflow execution, and human control in one configured process. Not just another place to ask a question or check a number.
- 01
Understands the business context
Our UCE connects records, agreements, and prior decisions to the task at hand.
- Linked records and entities
- Agreements and terms
- History and memory beyond one conversation
- 02
Carries out agreed work
Agents match records, prepare entries, and route exceptions in your connected systems, within agreed scope.
- Scoped execution in your systems
- Custom workflows on one foundation
- Shared context across the agents involved
- 03
Keeps your team in control
Review the supporting sources. Your permissions and approval rules determine what proceeds and what waits.
- Inspectable source evidence
- Role-based permissions
- Named reviewers and approvals
Chatbot, dashboard, or Lieutenant?
Answers explain. Dashboards show. Workflows act.
Chatbots and dashboards can also have connectors, memory, and automation. The comparison below describes their basic use, not a claim that these features exist only in Lieutenant.
Technical detailsHow our UCE works under the hoodThe six-stage pipeline, three-layer architecture, and eight core capabilities.
The pipeline
From raw records to grounded work, in six steps.
UCE is not one-time indexing or summarization. It repeatedly interprets organizational data from role-specific perspectives as sources refresh.
Step 01 of 06
Ingest
Connect the systems agreed in scope. The context layer captures new records on the refresh cadence set for each source.
Sample: inbox, shared drive folder, and accounting export, refreshed on an agreed schedule.
Architecture
Three layers. One context system.
UCE sits between your organization's records and the agents, workflows, and grounded answers that do the work.
Layer 1 · Sources in scope
- Shared files
- Meeting transcripts
- CRM
- Accounting / ERP
- Spreadsheets
Layer 2 · Unified Context Engine
- Continuous Ingestion
- Evidence Storage
- Hybrid Retrieval
- Permission Enforcement
- Temporal Reasoning
- Canonical Normalization
- Context Compilation
- Grounded Answering
Layer 3 · What uses the context
Agents
Do scoped work with persistent context.
Workflows
Context-triggered steps with review gates.
Grounded answers
Questions answered with references.
Briefings
Role-specific summaries on a schedule.
Core capabilities
What our context engine actually does.
Not just search. Not a chatbot over your files. An evolving context layer with memory, evidence, permissions, and judgment.
Continuous Ingestion
Ongoing sync from the connected sources in scope, such as email, files, transcripts, and business systems, on a defined refresh cadence.
Evidence Preservation
Facts are stored with source, timestamp, and provenance: the answer plus the chain of receipts.
Permission Enforcement
Context access is scoped to role and user, following the access you grant. People and agents see what they are permitted to see.
Canonical Normalization
Conflicting facts are detected and flagged for reconciliation, working toward one consistent view.
Hybrid Retrieval
Semantic search combined with keyword matching and structured filters.
Temporal Reasoning
Understands what happened when, surfacing relevant history without flooding every prompt with stale data.
Role-Aware Intelligence
The controller, the analyst, and the operator each get a different, appropriate slice of context.
Grounded Answering
Responses point to a specific source, quote, document, or event that a person can open and check.
Scoped on purpose.
Every engagement starts by agreeing what the engine can touch and who signs off.
- Source systems
- Only the systems named in scope are connected.
- Access
- Read and write access is granted by you and can be narrowed or revoked.
- Refresh cadence
- Each source refreshes on an agreed schedule, not assumed real time.
- Human permissions
- Named people approve material actions before anything leaves the building.
Questions
What people ask before a walkthrough.
How is this different from using a chatbot?
A chatbot is an interface for asking and answering. It can have memory, sources, and tools too. Lieutenant combines our UCE's business context with configured agents, system actions, and approval steps around a recurring job. The goal is not just a helpful response; it is carrying out the agreed work and routing the decisions your team must make.
Is Lieutenant another dashboard?
Reports help you see the business, but a chart alone does not reconcile an invoice or prepare an entry. Lieutenant does the scoped work behind the numbers and can provide the resulting reports. You can retain your existing dashboards and accounting system.
Is UCE a new foundation model?
No. Our UCE organizes and retrieves your permitted business context for the models, agents, and workflows doing the task. The distinction is how context, execution, and review work together, not a claim that we invented a new language model.
How does UCE fit with Lieutenant?
UCE supplies relevant organizational context, memory, and source evidence to agents and workflows. This page explains that architecture; the sources, capabilities, and deployment options for your engagement are agreed during scoping.
Which systems does it connect to?
Only the systems agreed in scope for your engagement, with the access you grant and a refresh cadence we set together. We do not assume every tool connects or updates in real time.
Do references mean the answer is always right?
No. References show where an output came from so a person can check it quickly. They make review faster; they are not a guarantee of correctness. Material actions stay with a human approver.
Who can see what?
Retrieval follows the roles and permissions you configure. Agents and people only receive context their access allows, and you can narrow scope at any time.
See our context engine on your own workflow.
We'll walk through which sources would be in scope, what our engine would compile, and where your people approve.