Inbox operations
Classify requests, extract details, prepare replies, and route work without replacing the inbox.
classifyextractdraftroute
Custom automation systems
Custom systems for operational work.
Relevo Lab builds automation, internal tools, AI workflows, and knowledge interfaces around the real processes teams run every day.
Inbox operations
Knowledge chatbots
Property management
AI assistant workflows
Internal infrastructure
Inbox operations
A request comes in. The system prepares the next step.
01
Read the incoming requestEmail, form, webhook, or document
input
02
Pull out the useful detailsDates, people, amounts, intent, owner
extract
03
Prepare a draft or updateReply, summary, record, report, next action
prepare
04
Wait for approval, then syncInbox, CRM, database, dashboard, team channel
approve
workflow.inbox_opsapproved
Receivedmailbox
Parsedschema
EnrichedAPI
Approvedhuman
Syncedsystem
What we build
Useful systems for operational work.
The work usually sits between existing tools: someone reads, copies, checks, routes, drafts, updates, and follows up. That is where automation belongs.
Internal tools
Build dashboards, admin panels, status views, audit logs, and manual override flows.
dashboardauditoverrideroles
Workflow bridges
Connect APIs, databases, spreadsheets, CRMs, M365, and custom business systems.
webhooknormalizesyncretry
AI-assisted workflows
Use language models for classification, extraction, drafting, validation, review support, and assistant briefings.
summarizevalidatebriefreview
Knowledge chatbots
Internal or guest-facing assistants over SOPs, company data, documents, and operational knowledge.
RAGSOPsourceescalate
Property operations
Lightweight systems for small hotels, resorts, independent operators, and Airbnb-style owners.
bookingsratesavailabilitychannels
How it works
Start small. Keep the system real.
The goal is not automation theater. It is a working system with clear inputs, approval points, logs, retries, and a path to operate it.
Map the workflow
Inputs, decisions, edge cases, tools, owners, and the manual steps that slow the work down.
Build the pilot
A small working system on real data. Useful first, wider later.
Add controls
Human approval, permissions, audit logs, retries, failure handling, and clear system state.
Operate and improve
Monitor, adjust, extend, and keep the system aligned with how the operation changes.
Selected systems
Built across several operating patterns.
No client names here. The public proof is the shape of the work: inboxes, knowledge assistants, property operations, internal tools, and workflow infrastructure.
system.inbox_ops
built
classification, extraction, drafts, routing, Outlook side panel
system.property_ops
built
bookings, rates, availability, roles, sync workflows, admin tools
system.assistant
built
daily briefing, calendar, mail, tasks, finance signals, review support
system.knowledge_bot
built
RAG, SOP answers, internal data, employee questions, escalation paths
system.workflow_stack
built
queues, retries, audit trail, monitoring, approvals, operator panel
Principles
Operational first. Clear by default.
01Existing tools first.
02Human approval where it matters.
03Small useful systems before big platforms.
04Clear logs over black boxes.
05Production behavior, not demos.
06Maintained after launch.
Contact
Send the process that should run better.
A few sentences are enough: what comes in, what needs to happen, which tools or data are involved, and where the current friction sits.
Input
Email, document, data, question, event, request
Friction
Manual checking, repeated answers, routing, updates, follow-up
Output
Answer, draft, dashboard, status update, ticket, synced record