01
Goal
Operator submits intent in Command or Demo.
Product guide
How to use Asael.
A complete operating manual for the public site, dashboard readiness, command center, memory, workflows, connectors, governed tools, evaluations, observability, security, and production release gates.
System map
From request to release evidence
01
Goal
Operator submits intent in Command or Demo.
02
Identity
Session, tenant, actor, and role are resolved.
03
Context
Memory, RAG chunks, graph hints, and run history are assembled.
04
Plan
The goal becomes workflow nodes with risk and verification.
05
Tools
MCP, OpenAPI, and internal tools are classified before side effects.
06
Approval
Risky work pauses for human or policy approval.
07
Observe
Runtime events, SLOs, incidents, alerts, and diagnostics are recorded.
08
Release
Evaluation reports and release gates prove production readiness.
09
Learn
Useful results can become durable memory and future context.
Quick path
Start with the safest path to value.
New users should understand the product in demo mode, sign in with owner access, then follow readiness from the dashboard and command center.
01
Explore safely
Use the sample workspace to see how a goal becomes memory-backed, approval-aware, auditable agent work.
Open demo
02
Enter the private workspace
Use the configured owner account to continue into the operating workspace.
Sign in
03
Operate from the app
Use the dashboard and command center for live health, release readiness, workflows, approvals, and observability.
Open app
App flow
The workspace has one main loop.
Every serious task should pass through identity, context, planning, governed execution, observability, release evidence, and memory.
02
Identity
Session, tenant, actor, and role are resolved.
03
Context
Memory, RAG chunks, graph hints, and run history are assembled.
04
Plan
The goal becomes workflow nodes with risk and verification.
05
Tools
MCP, OpenAPI, and internal tools are classified before side effects.
06
Approval
Risky work pauses for human or policy approval.
07
Observe
Runtime events, SLOs, incidents, alerts, and diagnostics are recorded.
08
Release
Evaluation reports and release gates prove production readiness.
09
Learn
Useful results can become durable memory and future context.
Command center
Use Command when you want the system to act.
The command center is the working surface for goals, modes, context, tool calls, approvals, run history, and execution evidence.
Open command centerOrchestrate
Turn a goal into a governed plan, workflow, tool calls, evidence, and memory updates.
Research
Use RAG, memory, source chunks, and graph context to answer questions with provenance.
Execute
Run approved tools, dry runs, connector operations, and workflow ticks with audit records.
Learn
Capture successful outcomes as durable memory, reusable plans, and operational knowledge.
Feature guide
Every feature and what to use it for.
Use the app surfaces as a sequence: dashboard for posture, command for work, memory and connectors for inputs, tools and workflows for action, evaluations and observability for proof, security and settings for control.
Capture
Save a note or file to your second brain.
posturehealthrelease
Arsenal
Your specialists, capabilities, and learning loops.
posturehealthrelease
Command
Ask Asael a question or hand off a task.
posturehealthrelease
Today
Your work, decisions, and recent results.
posturehealthrelease
Missions
Durable outcomes, delegated work, and evidence.
posturehealthrelease
Inbox
Review actions that need your attention.
posturehealthrelease
Results
Review completed work with its evidence and verification.
posturehealthrelease
Workflows
Multi-step jobs that run in the background with retries and approvals.
Queue: DurableApprovals: PolicyRecovery: Auto
Memory
What your second brain knows and remembers.
Memory modes: 4Vector backend: HNSWProvenance: Chunk
Integrations
Connect external tools and APIs the agent is allowed to use.
Protocols: MCP + RESTSecrets: RedactedNetwork: Guarded
Tools
What the agent can do, with risk levels and an audit trail.
Risk levels: 0-3Dry run: FirstAudit: Full
Quality Checks
Automated tests that verify the agent behaves before you rely on it.
Release gates: 6Reports: SignedSmoke: CI
Monitoring
System health: events, errors, alerts, and incidents.
SLO: PolicyIncidents: LiveAlerts: Routed
Security
Access control, audit logs, and data isolation.
Tenant RLS: ForcedAuth: EnforcedAudits: Everywhere
Settings
Models, environment, and configuration.
Runtime: Vercel + FlyDatabase: NeonModels: Routed
Routes
Where everything lives.
The product is split into public education, private access, and authenticated operations.
/
Public product homepage
/platform
Platform overview and product architecture
/solutions
Use cases and deployment patterns
/security
Security, governance, and compliance posture
/pricing
Commercial packaging placeholder
/docs
This complete product guide
/changelog
Shipped slices and platform progress
/demo
Public sample workspace
/login
Secure session sign in
/app
Authenticated operations dashboard
/app/command
Goal entry and command center
/app/memory
Memory, RAG, graph, and provenance
/app/workflows
Durable workflows, queues, approvals, and recovery
/app/connectors
MCP and OpenAPI connector management
/app/tools
Governed tool execution and audit
/app/evaluations
Regression runs, signed reports, and release gates
/app/observability
Runtime events, SLOs, incidents, and alerts
/app/security
Tenant isolation, auth posture, and audit controls
/app/settings
Runtime, environment, tenant, and model posture
Use cases
What teams can get done.
Asael is useful whenever AI work needs memory, tools, approvals, repeatability, and production evidence.
Research and knowledge work
Ask complex questions over source-backed memory and preserve useful answers.
01Ingest documents
02Retrieve with RAG
03Summarize with provenance
04Save durable memory
Operations automation
Turn repeatable operational tasks into auditable workflows with approvals.
01Submit goal
02Plan workflow
03Run safe tools
04Recover or approve
Customer support agents
Retrieve policy, inspect history, draft responses, and escalate risky actions.
01Load ticket
02Retrieve policy
03Draft answer
04Request approval
Security and compliance
Track tenant isolation, auth failures, policy blocks, and signed release evidence.
01Inspect context
02Run eval
03Review audit
04Gate release
Connector-driven work
Register external APIs without exposing platform secrets or private networks.
01Import spec
02Classify risk
03Dry run operation
04Execute with audit
Release readiness
Use smoke tests, SLO snapshots, evaluation reports, and signing gates before deploys.
01Run smoke
02Check SLO
03Verify report
04Approve release
API map
The backend surface behind the UI.
These route handlers power the product. Anonymous users can access public surfaces; protected operational routes require session or trusted internal context.
Auth and identity4 routes
/api/auth/login
/api/auth/logout
/api/auth/session
/api/auth/control-plane
Agent, memory, RAG, and knowledge8 routes
/api/agent
/api/memory
/api/memory/graph
/api/knowledge
/api/ingest
/api/retrieval/plan
/api/runs
/api/capabilities
Workflows, triggers, approvals, and operations12 routes
/api/workflows
/api/workflows/[id]
/api/workflows/[id]/tick
/api/workflows/[id]/signal
/api/workflows/executions
/api/workflows/plan
/api/workflows/tick
/api/triggers
/api/triggers/[id]/dispatch
/api/approvals
/api/approvals/[id]
/api/operations
Connectors and governed tools9 routes
/api/connectors
/api/connectors/[id]
/api/connectors/[id]/discover
/api/openapi-connectors
/api/openapi-connectors/[id]
/api/openapi-connectors/[id]/import
/api/connection-catalog
/api/tools
/api/tools/execute
Evaluation, release, observability, and incidents14 routes
/api/evaluations
/api/evaluations/[id]
/api/evaluations/[id]/report
/api/evaluations/[id]/report/verify
/api/release/evidence
/api/observability
/api/observability/slo
/api/observability/slo/policies
/api/incidents
/api/incidents/[id]
/api/incidents/[id]/actions
/api/alerts
/api/diagnostics
/api/health
Security and tenant isolation3 routes
/api/security/context
/api/security/audits
/api/security/isolation-report
Production checklist
Use this before and after every serious run.
The app can connect to real systems, so the operating discipline matters as much as the model output.
1
Set production secrets in Vercel: OpenAI, database, auth bootstrap, cron, internal smoke, and signing values.
2
Confirm /api/health returns healthy before inviting operators.
3
Run Production Smoke after every deployment and inspect release evidence.
4
Use demo mode for prospects and dashboard readiness for new workspaces before connecting real systems.
5
Register one connector at a time, start with dry runs, and require approval for high-risk operations.
6
Review SLO warnings after smoke tests; cold-start latency warnings can be advisory when availability and errors are clean.
7
Keep access requests separate from real account creation until admin invitation, billing, and SSO policies are finalized.
8
Check observability, incidents, alert deliveries, and workflow queues after smoke runs or high-risk demos.