Features · Agent platform
Built for the work between your tools.
Bezor agents read, decide and act across your inbox, CRM and ledger — inside permissions you set and with a person on every decision that matters.
01 — Capabilities
Six things every agent does by default.
Live run log
Streaming
Watch every step an agent takes, in order, with the reason it took it. Click any line to replay the run.
09:12:38 Ledger Read 14 new invoices from AP inbox
09:12:40 Ledger Matched INV-2207 to PO-881 · 3-way OK
09:12:44 Ledger Flagged $1,240 variance on INV-2209
09:12:45 Warden Variance above $1,000 · approval required
09:12:51 Ledger Sent to L. Ferrer with summary + sources
09:14:02 Human Approved by Lucía Ferrer · posted to NetSuite
Plain-language playbooks
Brief an agent the way you’d brief a new hire. Bezor turns it into steps, rules and tests.
Confidence routing
Low-confidence work goes to a named person with context, a draft and the sources used.
Owner dashboard
One screen for what your agents closed, escalated and learned this week.
1,284
tasks closed
37
escalated to people
12
playbook updates
Memory per account
Agents remember each customer, vendor and deal, so nobody explains things twice.
Human owners
Every agent reports to a person who approves changes and gets the weekly review.
Works where you work
Agents act inside Gmail, Zendesk, HubSpot and NetSuite. No new tab for your team to learn.
02 — Use cases
One platform, four departments.
Every ticket answered, triaged or handed off with context.
Scout reads Zendesk, Intercom and your shared inbox, resolves routine requests inside policy and passes the rest to the right person with a summary and a suggested reply.
Refunds under your limit approved automatically
VIP and churn-risk tickets flagged in seconds
Replies in your tone, signed by your team
9 min
median first reply
03 — Integrations
Plugs into the stack you have.
Gmail
Outlook
Zendesk
Support
Intercom
Support
Slack
Chat
Notion
Docs
HubSpot
CRM
Salesforce
CRM
Stripe
Payments
NetSuite
ERP
QuickBooks
Accounting
Webhooks + API
Custom
04 — Guardrails
Every permission is written down.
01
Read, draft, act
Three levels of access, set per tool and per action — never a blanket API key.
02
Thresholds
Dollar, risk and confidence limits that send work to a person before it ships.
03
Kill switch
Pause one agent or all of them from Slack, the dashboard or your phone.
04
Data boundaries
Region pinning, 30-day retention and no training on your data. Ever.
ledger.policy.yml
Approved · v12