Triage and labelling
Every incoming message is read, summarized, and sorted before you open the inbox.
- Urgent enquiries flagged for a human first
- Routine requests labelled and queued for drafting
- Newsletters and notifications filed out of the way
Inbox & administration
An AI inbox assistant that clears the backlog - with your approval
Triage, replies, follow-ups, scheduling, the weekly report - the everyday admin that quietly eats your team's hours. Olano puts a dedicated AI agent on that work: it reads and sorts everything, drafts the replies, and chases the silences, while anything outbound waits for your sign-off. Start with one workflow and be live in days, not months.
What it handles
One persistent agent working your mailbox like a careful assistant - with memory across sessions, grounded in the documents you upload, and running to a schedule you set.
Reply drafts, queued for approval
Replies are drafted from your own knowledge base - price lists, policies, SOPs - and wait in a queue. You edit, approve, or discard; nothing is sent until you say so.
Follow-up chasing
Threads waiting on someone else don't go stale. The agent tracks who owes a reply and prepares polite nudges at the intervals you set.
Calendar coordination
Scheduling requests are handled through your connected calendar: the agent proposes free slots, drafts the confirmation, and prepares reminders before the meeting.
Recurring admin on schedule
Scheduled tasks and proactive monitoring cover the repeatable jobs: a 7am triage digest, a Friday summary of open threads, an overnight sweep for anything left unanswered.
What waits for approval
Olano's human-in-the-loop model runs on trust levels 0-4 - Observer, Assistant, Collaborator, Autonomous, Developer - plus per-category approval gates for financial actions, external sends, file writes, secrets, and delegation. You decide how much autonomy each category gets, and you can change it at any time.
For an inbox assistant the defaults are simple. Reading, triage, and drafting run autonomously, because they change nothing outside the workspace. Sending is gated by default - every outbound email sits in the approval queue until you clear it. Many teams keep it that way; others later allow routine acknowledgements out at a higher trust level while quotes and commitments stay gated.
Everything on the record
Every action - each label applied, draft written, message sent, scheduled run completed - is recorded in an immutable audit trail. When a customer asks what was said and when, or you want to check what the agent did overnight, the full history is there to read.
Before and after
Before: Monday opens on a wall of unread mail, and the first hour goes to sorting it. Urgent enquiries sit behind newsletters, and the follow-up you meant to send on Thursday never went.
After: the agent has already triaged overnight. You open a prioritized queue, review a handful of drafts, approve the ones that are ready, and the day starts at the actual work - with every step logged.
Unread pileTriaged queueDrafts awaiting approvalSent + logged
Works with the inbox you already have
Gmail and standard email/IMAP are real channel adapters - the agent connects to your existing mailbox and works inside it. No migration, no new address, no forwarding rules to maintain.
Around the inbox, the agent reaches the rest of your stack through Olano's 75+ built-in integrations, each with its own specialist subagent: Google Calendar for scheduling, HubSpot for logging enquiries into the CRM, Notion for SOPs and meeting notes, Slack for internal handoffs. Teams on Microsoft 365 get a toolkit for mail and files tooling as well.
Because everything runs in your isolated workspace, connector tokens stay encrypted and each integration only gets the access you grant it.
FAQ
Does the inbox assistant send emails on its own?
Not unless you choose that. Drafting, triage, and research run autonomously because they change nothing outside your workspace. Anything outbound - an email send, a calendar invite, a CRM update - waits for human approval by default. Trust levels 0-4 let you widen autonomy gradually, per category, once the agent has earned it.
What stays private?
Each customer runs in an isolated workspace. Connector tokens and API keys are encrypted, access is role-based, and data is encrypted at rest and in transit. Custom AWS or GCP regions are available for data-residency needs, and every action the agent takes is recorded in an immutable audit trail.
How long does setup take?
A single inbox workflow is typically live within 1-2 business days of onboarding; bigger rollouts take 1-3 weeks. Every build is quoted and approved before we start.
How is an Olano engagement priced?
Every Olano engagement is scoped individually - pricing reflects the workflows involved, required integrations, operating volume, deployment architecture, and ongoing management requirements. Following an initial consultation, we provide a fixed proposal before any work begins. AI usage runs within hard spending controls agreed in advance, and access is never priced per seat.
Do we have to change our email address or provider?
No. Gmail and standard email/IMAP are real channel adapters, so the agent works inside the mailbox you already have. Teams on Microsoft 365 can use a mail and files toolkit for the same triage and drafting work.
Related
More on inbox automation with approvals - and the workflows teams usually add next.
AI inbox triage and follow-ups
The playbook: how to structure triage rules, draft queues, and follow-up cadences before you automate them.
Missed-lead recovery
When enquiries slip through the inbox, an agent that spots and revives them is often the fastest win.
WhatsApp AI customer service
The same approval-first model, applied to the channel your customers actually message you on.
Put an assistant on the inbox this week
Book a consultation. We map one inbox workflow, quote it before we build, and most single-workflow setups are live within 1-2 business days of onboarding.