6. End-to-end workflow example

~5 minute read. One realistic scenario, start to finish: a marketing team analyzing customer feedback with an AI agent. Total elapsed time, about 30 minutes.

Cast: Alice, marketing lead and workspace Admin. Bob, analyst, joining today as a Member.

Step 1 — Alice signs in and invites Bob

Alice goes to id.eworks.cloud, clicks Continue with Okta, approves the push on her phone, and lands in the acme-marketing workspace. She opens Settings → Members, enters bob@company.com, picks Member, and sends the invitation.

Elapsed: 4 minutes. Details: SSO setup, team management.

Step 2 — Bob joins

Bob clicks the link in his email, signs in with Okta, registers a passkey and a TOTP authenticator, and lands in the same workspace. He can chat and build agents; he cannot change billing or invite anyone.

Elapsed: 6 minutes.

Step 3 — They ask a question together

Both open e.chat, select Claude, and ground the conversation in the customer-feedback collection — Q3 survey responses and support tickets, already synced by e.gateway.

Alice: Summarize Q3 customer feedback. What are the top three themes and how did sentiment shift versus Q2?

Claude streams back a summary with citations. Bob follows up asking for the negative themes broken out by plan tier.

Two messages, cost $0.03. Details: first chat.

Step 4 — They turn it into a daily agent

The answer is good enough to want every morning. Bob opens e.agent → Create agent:

name: q3-feedback-digest
description: Daily themes and sentiment from customer feedback, emailed to marketing.
trigger:
  type: schedule
  cron: "0 9 * * *"        # daily at 09:00
memory:
  enabled: true
  retention_days: 30
steps:
  - id: search
    tool: knowledge.search
    with:
      collection: customer-feedback
      query: "feedback received in the last 24 hours"
      limit: 50

  - id: analyze
    model: claude
    prompt: |
      Extract the top themes and overall sentiment from the feedback below.
      Note anything new compared with yesterday's digest.
      {{ steps.search.results }}

  - id: email
    tool: email.send
    with:
      to: alice@company.com
      subject: "Customer feedback digest — {{ run.date }}"
      body: "{{ steps.analyze.output }}"

He runs it once manually, checks the output in the execution view, and saves.

Elapsed: 12 minutes. Details: first agent.

Step 5 — The next morning, in the audit trail

At 09:00 the agent fires. Alice opens e.audit and filters to the last 24 hours:

Two chat messages at $0.03, one agent execution at $0.05, and the outbound email recorded as an external action with recipient and message ID. Day-one spend: $0.08.

[Screenshot: Audit table filtered to the last 24 hours showing the five rows above]

Step 6 — Alice exports for compliance

For the Q3 review she filters 2026-07-01 to 2026-09-30, exports signed JSON, and drops it into the compliance folder. Because a customer asked what data the company holds on them, she also runs a DSAR from Compliance → Data subject request and gets a ZIP back in under a minute.

Elapsed: 5 minutes. Details: audit review.

Takeaway

Under 30 minutes from first sign-in to a scheduled agent emailing summaries — with every message, tool call, dollar, and export attributable to a named person, and nothing extra to configure to make that true. No servers were provisioned, and the feedback data never left the workspace's boundary.

Next: FAQ