Skip to content
Promenaut

Promenaut joins forces with HSBC as strategic investor and design partner.Read More

The automation journey

Where is your organisation on the journey to a digital workforce?

Three questions place you on six waves of automation, from manual work to agent teams. Your answers stay in this page.

Three questions

1. In your most automated process today, who does most of the work?
2. When that work needs to double, what do you add?
3. Afterwards, can you see what the AI did, what it cost and who approved it?

Answer all three to see where you are. Your answers stay in this page; nothing is sent.

The six waves · Choose one to see it in the journey

Human-led

people drive, the machine helps

The shift · Agent-led

the work moves to the machine

The human and AI splits are illustrative.

01The journey

From manual work to a digital workforce

Automation has moved in waves: from spreadsheets, to scripted bots, to assistants you prompt, to copilots at your side, to single agents that act. The first four are human-led. The shift comes when the work itself moves to the machine. The next step is a governed digital workforce: teams of agents that take whole stages of work end to end, with a person deciding where the workflow says.

0ManualSpreadsheets and email
1RPAScripted bots
2AssistantChat assistants
3CopilotAssistants inside your tools
4AgenticSingle agents
5 · DestinationAgentic TeamsA digital workforce
What it is
People do the steps by hand: spreadsheets, documents, email, and copy-paste between systems.
Scripts that mimic clicks and keystrokes to repeat a fixed, rules-based process.
A general model you chat with: it drafts, summarises and answers on request.
An assistant embedded in the tools you already use, suggesting as you work.
A single autonomous agent that plans, uses tools and completes a multi-step task.
Teams of governed agents that take whole stages of work end to end, each team handing its output to the next like a workforce.
Who does the work
Human 100%AI 0%

People, end to end.

Human 75%AI 25%

A bot runs the steps; people build and babysit it.

Human 60%AI 40%

You ask each time; the model drafts the answer.

Human 50%AI 50%

You still drive; it suggests alongside you.

Human 30%AI 70%

The agent does the task; a person reviews it.

Human 15%AI 85%

Agent teams do the work; people set intent and approve.

In parallel
1 · one person
1 · one bot
1 · one chat
1 · one session
1 · one agent, one lane
Many · a team, in parallel
Where it breaks down
Slow, error-prone and hard to scale: each extra unit of output needs another pair of hands.
Brittle. One screen or rule change and it breaks; it does only what it was scripted to do.
Waits to be asked, forgets between sessions, and can’t touch your systems or finish a job itself.
Help per keystroke, not work end to end. It speeds a task up; it doesn’t own the outcome.
One agent in one lane: no division of labour, and no budget, audit or approval to trust it at scale.
The destination: whole stages of work done by governed agent teams, within a budget, replayable afterwards, and held for a person where the workflow says.

The human and AI splits are illustrative. On a phone, scroll the table sideways; the row names stay in place.

02Getting there

From wherever you are to agent teams, in four stages.

Take one process you already run. We take it the whole way with you: mapped, redesigned, running and governed on the platform.

01 · Map

How the work runs today

Your own documents, spreadsheets and diagrams are read into a map of the process, its rules and its data, each fact citing where it came from.

02 · Architect

Redesigned for agent teams

The work becomes stages that teams of agents take end to end, with the points where your people decide set in the workflow.

Your people choose where they decide

03 · Coordinate

Running, team to team

Agent teams do the stages, each handing its output to the next, while platform teams check the work as it runs.

04 · Control

Governed from the first run

Runs are held to a budget, wait for a person where the workflow says, and can be replayed step by step afterwards.

Your people approve, and see what it cost

Take one of your processes to wave 5 →

Running today for leading industry partners, including a global tier 1 bank.

Wave 5, running today

One piece of work at the destination: an investment case, examined through separate lenses at the same time and weighed into one go, conditional or no-go verdict, with a board-ready report.

Business Case Evaluation · one runagent teams · a person decides
Investment caseas a document or a draft
Separate lenses, at the same time
StrategyFinancialsReturnAssumptionsOptionsRiskReadiness
One verdictwith conditions and evidence
GoConditionalNo-go
Board-ready reporta document and a deck

Your people decide on funding. The verdict informs it.

Next step

A maturity conversation on your own material

Bring one process you already run. In an hour with Promenaut people, we place it on the journey, look at what an agent team would take on, and where people would still decide.

Book the conversationSee the platform