Skip to content

The platform

The Mandate Engine

The accountability layer for your AI agents. It sets how each agent shares the work with your people, puts consequential decisions in front of the right people, and keeps a record of who owns every action.

Four things every governed agent needs

Mandates

Define what each agent is allowed to do, at what level of responsibility, and on whose authority.

Approvals

Consequential actions go to the accountable person for a decision before they happen.

Ownership record

Every consequential action is recorded with its triggering signal, the rule applied, and the person who owns it.

Governed Action Rate

One number for leadership: the share of observed, in-scope consequential actions that meet your governance requirements, read alongside governance coverage.

See how it works →

Decide how people and agents share the work

Every agent starts as a new member of the team and earns more responsibility over time.

  • Observe: the agent watches and reports to a person.
  • Recommend: the agent suggests and a person decides.
  • Draft: the agent prepares the work and a person approves it.
  • Act: the agent carries out work a person has already authorized.

The most important decisions always stay with a person.

Mandate · how the work is shared

ObserveRecommendDraftAct
Owner: named personEvery action owned

What counts as a consequential action

Not every agent action needs a person. The ones that do share four traits. They reach a customer, spend money, change a system of record, or speak for the company. Everything else can go ahead within its mandate.

See the glossary for definitions of every term.

Illustrative example

A budget shift that had an owner before it happened

A paid media agent sees cost per opportunity rising on one channel and drafts a plan to move $40,000 of monthly budget to another. Its mandate allows it to recommend budget changes, not make them. The recommendation goes to the Director of Paid Media, who approves half the shift for a two-week test.

Signal
Rising cost per opportunity on one channel
Mandate
Recommend only for budget changes
Decision
Approved $20,000 shift for two weeks
Owner
Director of Paid Media

One accountability layer. Many agents and systems.

Your teams are introducing AI agents across marketing tools, CRM platforms, service applications, and internal workflows. Each tool can be useful on its own. The harder question is who authorizes consequential work as several agents contribute to one outcome.

The Mandate Engine is being designed to bring a consistent approach to delegated authority and human accountability across that environment. People set mandates. Agents do the work within their assigned boundaries. Decisions that require judgment reach the right owner, and the resulting action retains a clear accountability record.

The goal is to let organizations adopt the AI tools that make sense for them while maintaining a shared standard for the work that matters.

Illustrative product concept

Awaiting owner decision
Mandate
Customer-facing product launch communications
Accountable owner
Director of Product Marketing
Participating agent
Campaign Content Agent
Responsibility level
Draft
Allowed work
Prepare email copy and recommendations
Decision boundary
Human approval before send
Proposed action
Send launch message to 40,000 customers
Decision context
Feature claim flagged for review

Accountability record

  1. Mandate
  2. Review
  3. Approval
  4. Action
  5. Owner
Product-design illustration. Intended workflow, not a live customer deployment or a representation of completed integrations.

Agent builders help teams create AI agents. Workflow tools coordinate tasks. Monitoring tools help teams see what happened. RealMandates focuses on the business authority behind consequential agent actions and the person accountable for the result. These functions can overlap with other tools; our focus is making the ownership question operational.

What The Mandate Engine is not

Not an agent builder.

Build agents wherever your teams prefer.

Not a language model.

It works with the AI you already use.

Not an after-the-fact audit.

Accountability is settled before the action, not in a quarterly review.

Help shape The Mandate Engine

We are inviting qualified design partners to help shape The Mandate Engine. The platform is under development, and the first access opportunities will depend on product readiness and program fit.

What you get

  • Early access opportunities as development progresses
  • Direct influence on the roadmap
  • Working sessions with Mark Tack
  • Preferred commercial consideration at launch

What we ask

  • An executive sponsor
  • One AI agent use case to govern
  • A short feedback session each month
See the design partner program

Frequently asked questions

Consequential actions taken by AI agents, meaning actions that reach customers, spend money, change records, or speak for the company.

It is designed to govern agents wherever they were built, not to replace them. Design partners help decide the first integrations.

Governed Action Rate (GAR) is the percentage of observed consequential AI agent actions within a defined measurement scope that meet the organization's authorization, approval, and named-owner requirements. RealMandates proposes 100% as the operating goal for in-scope actions. The rate must be read alongside governance coverage, because actions the organization cannot observe cannot be assumed to be governed.

No. Routine actions go ahead within the agent's mandate. Human decisions are reserved for consequential actions, and the mandate decides which are which.

The Mandate Engine is in development. We are recruiting design partners to help validate the first use cases, integrations, and product experience. Request a conversation to learn about participation.

Every action owned, from day one.