IT playbook · AI Employee: Ivy

Okta Password / MFA Reset

Reset completed under 60s; audit_event written

The problem

Password reset is the single most common IT support ticket at every company. Every time it takes 30-60 minutes to route to a human, the requester is locked out of their job, the IT team loses time on a solved problem, and the org accumulates learned helplessness about how long simple things take. It should be seconds.

At a glance
Trigger
Chat (Slack/Teams/Portal)
Approvals
Step-up MFA on the requester
What it does
Writes to your systems
Systems
Okta · Identity verification
How it feels in production

An hour-by-hour walkthrough.

9:03am Wednesday. Priya can't log in. She DMs Ivy in Slack: "I can't reset my Okta password, it says my MFA isn't working." Ivy verifies Priya's identity via a step-up challenge to her registered device (or a live video verification if the device is what's broken), pulls up her Okta record, clears her MFA factors, sends a self-service password reset link to her recovery email, and posts the confirmation in the same thread. 9:04am. Priya gets the reset link, sets a new password, re-enrols her authenticator. She's back in. Every step of that flow lands in the audit log — who requested, when, who / what verified, what factors were cleared, what link was sent. Security gets a summary in their weekly digest of all password / MFA resets. Anomaly signals (multiple resets in a week, resets from unusual IP) escalate to a human immediately. If Priya's identity can't be verified via device or manager, the request escalates. The Assistant never resets a password on a request it can't attribute to the real employee.
How it works

Step by step.

  1. 01

    Identify requester

    Read the requester's identity from Slack / Teams SSO. Pull their Okta record — is the account active? Is it locked? Are there recent failed attempts?

    Slack · Teams · Okta
  2. 02

    Verify identity via step-up MFA

    Send a push challenge to the requester's registered device (or SMS to their registered phone). If the requester's registered device IS what's broken, escalate to manager-verification or live-video fallback per policy.

    Okta · Twilio · Duo
  3. 03

    Reset password or clear MFA factors

    On verified identity, clear the specific MFA factor that's failing (or all factors if the requester needs a full reset). Send a self-service reset link to the recovery email or Slack DM.

    Okta
  4. 04

    Force re-enrolment on next login

    Enrolment flags flipped on the account so first successful login prompts MFA setup fresh. Old factors invalidated.

    Okta
  5. 05

    Notify + audit

    Reset confirmation posted in the requester's Slack / Teams thread. Full event written to audit_events with actor, target, IP, device fingerprint of the verifier.

    Slack · Audit log
Systems and wiring

What you connect to make this run.

Okta

read+write

Service account with password-reset + factor-management scopes. Never uses admin-level 'set password' — always sends the user self-service link so nobody, including us, sees the plaintext.

Slack · Teams

trigger

Chat DM to Ivy triggers the flow. Requester identity read from Slack/Teams SSO. Never accepts requests via email or unauthenticated portal.

Twilio · Duo (fallback identity verification)

write

When the registered device IS what's broken, SMS or voice-call fallback to registered phone. Zero-fallback policy for high-privilege accounts — human verification required.

What changes

Before and after, honestly.

Time from request to reset
Before
30-60 minutes across ticket routing
After
Under 90 seconds in-chat
IT time per reset
Before
~10 minutes across ticket handling + verification + reset action
After
0 minutes for verified requests; escalations only
Password reset ticket volume
Before
~20% of all IT tickets
After
Near zero — resolved in-chat before a ticket exists
Fraudulent reset attempts caught
Before
Detected only in post-hoc audit
After
Anomaly signals surface within 1 hour
Frequently asked

Answers about this playbook.

How is this different from Okta's self-service reset?

Okta's self-service reset works when the user still has access to their MFA device. When they don't — which is the common failure mode — they hit an IT ticket. Ivy handles that recovery flow with identity verification, factor clearing, and audit — all in Slack, without a ticket.

What if the requester can't pass the identity verification?

Escalation to a human IT operator with the request context bundled. Ivy never resets a password on a request that failed identity verification. Manual paths (live video, manager verification) are documented in the runbook and available to the operator.

Can this handle SSO providers other than Okta?

Yes — same pattern for Microsoft Entra ID, Google Workspace, WorkOS, Auth0, JumpCloud, OneLogin. Each provider's API differs; the playbook shape is identical.

What about phishing risk — could an attacker trick Ivy into a reset?

Identity verification runs against a registered device or channel, not a claim made in-chat. Ivy never accepts 'this is Priya' as verification — always requires a factor the real Priya controls. Anomaly signals (unusual IP, multiple resets, resets right after suspicious login) escalate immediately.

Does this cover high-privilege accounts?

By default, high-privilege accounts (admin, IT, security) require human verification — the playbook escalates rather than auto-resets. Configurable per role in the Approval Policy.

See it run on your data.

Free plan, no credit card. Connect the systems this playbook needs and run it against a past event first.