Skip to content
Velum
← All solutions

Enterprise, every department

An agent for every employee, fluent in their job.

Generic assistants answer questions. A personal agent does the job around the job. Each employee gets an agent configured for their department, connected to the systems they already use and limited to the data they are already allowed to see. It starts the day with a brief, clears the inbox, prepares meetings, turns decisions into follow-ups and updates, and learns their preferences. Anything that sends, signs or spends waits for the person to approve, and every action is logged for IT and compliance.

5–7 h
Saved per employee, weekly
1 console
For every agent
0
Sends without approval

How it runs

Step through the pipeline.

Click a step or let it play. The workday record on the right fills in as the agents work.

Step 01 of 08

Provision by role

The agent is created from a department template, such as sales, finance, HR, legal or engineering, and inherits the employee's single sign-on identity and existing permissions. It can never see more than the person can.

Okta / Entra IDVelum Runtime
Workday record1 / 8 fields
Profile
Maya K. · Finance BP · template: Finance · SSO scopes inherited
Brief
—
Inbox
—
Meetings
—
Actions
—
Workflows
—
Memory
—
Governance
—
  1. 01 · Provision by role

    The agent is created from a department template, such as sales, finance, HR, legal or engineering, and inherits the employee's single sign-on identity and existing permissions. It can never see more than the person can.

    Tools: Okta / Entra ID, Velum Runtime. Output: Profile.

  2. 02 · Start the day with a brief

    Before the first meeting the agent summarises what changed overnight: priority emails, messages that need a reply, calendar conflicts and the metrics this person owns.

    Tools: Outlook / Gmail, Teams / Slack. Output: Brief.

  3. 03 · Triage the inbox

    Mail and messages are sorted into act, delegate, read and ignore, with drafts written for the routine replies in the employee's own tone. Nothing sends without a tap.

    Tools: Outlook / Gmail, Teams / Slack. Output: Inbox. A person approves before this step completes.

  4. 04 · Prepare every meeting

    For each meeting the agent pulls the attendees, the last conversation, the open actions and the relevant documents into a one-page brief, so nobody walks in cold.

    Tools: Calendar, SharePoint / Drive, CRM. Output: Meetings.

  5. 05 · Turn decisions into actions

    After each meeting the notes become tasks, tickets, CRM updates or follow-up emails in the right systems, assigned to the right people with due dates.

    Tools: Jira / Asana, CRM, ERP. Output: Actions.

  6. 06 · Run department workflows

    Each template carries the recurring work of that role: a sales agent updates the pipeline, a finance agent chases approvals, an HR agent answers policy questions, a legal agent checks clauses against the playbook.

    Tools: Department systems, Velum agent. Output: Workflows.

  7. 07 · Remember preferences

    The agent keeps a private memory of how the person works: who matters, preferred formats, recurring commitments. The employee can read and delete it at any time.

    Tools: Velum Runtime. Output: Memory.

  8. 08 · Govern centrally

    IT sees every agent, its permissions and its actions in one console, with policies per department, data-loss prevention and a full audit trail. Usage and hours saved are reported by team.

    Tools: Velum console, SIEM. Output: Governance.

Impact

What changes when it runs.

Illustrative targets from comparable engagements. Your baseline is measured in the audit and the targets are agreed before we build.

MeasureBeforeWith agentsChange
Time on email and messagesAbout 11 hours a weekAbout 5 hours−55%
Meeting prep20 to 30 minutes eachBrief ready on arrival−90%
Follow-ups actioned after meetingsAbout halfAll of them, tracked2×
Visibility of AI use for ITScattered toolsOne governed consoleComplete

Hours saved and ROI

What it is worth to your team.

Start from our defaults for this pipeline, then move the sliders to your numbers. The maths is simple and shown.

Hours reclaimed6 hper employee, every week

Independent 2026 surveys put the saving from AI assistance at roughly 6 hours a week per knowledge worker. Agents that also act inside department systems tend to sit at the upper end; the pilot measures your baseline first.

Worked example

Two hundred employees each spending 10 hours a week on email, meeting prep and follow-ups at a $65 loaded hourly cost, with just over half of it handled by their agent: about 1,100 hours a week back, roughly $3.4 million a year in capacity.

Gross capacity, not cash savings. Excludes implementation, software and supervision costs, which are quoted before we start.

Your numbers

200
10 h
$65
55%

An assumption to explore, not a promised result.

What that is worth

1,100 h

reclaimed every week

$3,718,000

value per year

4,763 h

every month

27.5 FTE

of capacity returned

Hours on repetitive work, per week

Today2,000 h
With agents900 h
Validate these numbers

Gross capacity estimate, not cash savings. Excludes implementation, software, and supervision costs. We validate assumptions in the audit.

Questions about this pipeline

01

How is this different from Copilot or ChatGPT Enterprise?

Those are excellent assistants you ask questions of. Personal agents are configured per role, connected to your department systems and allowed to act, with approvals and audit. We often build on top of the licences you already have.

02

Can an agent see data the employee cannot?

No. Each agent inherits the employee's identity and permissions through single sign-on. It is scoped to exactly what that person can already access, and IT can narrow it further per department.

03

Which departments do you start with?

Usually one high-volume team, often sales, finance or customer success, with 20 to 50 people. The templates are then extended to other departments once the pilot shows the hours.

04

Where does the data go?

Agents run in our managed cloud, your cloud account or on-premise. Model providers are configured with zero data retention, and memory stays in your tenancy.

05

Does this work for smaller companies too?

Yes. A 30-person company can give everyone an agent on the same runtime; there are simply fewer templates and a lighter governance setup.

Want this running for your team?

We start with one department of 20 to 50 people, measure the hours in a supervised pilot, then roll the templates out across the business.