Data stays in your tenant
Agents run inside your own Microsoft 365 environment. Your content is not used to train the underlying models and it does not leave the Microsoft 365 service boundary.
Home / The Agent Journey
Capability map
The Agent Journey.
The Agent Journey is FiveForward's five-stage map of Copilot agent capability: Use an agent, Build your first agent, Team agents, Copilot Studio and A team of agents. It takes a team from never having used an agent to running agents across the business, all inside your own Microsoft 365 tenant. At every stage there are two doors: learn to do it yourself or have us build it for you.
Security first
Are Copilot agents secure?
It is the first question most leaders ask: where does our data go? The short answer is that it stays where it already lives. Copilot agents are grounded securely within your Microsoft environment, at every stage of this journey.
Agents inherit your permissions
An agent can only read what the person asking is already allowed to see. Your existing SharePoint and Microsoft 365 permissions carry straight over.
People see what they can already see
If someone cannot open a document today, an agent will not show it to them tomorrow. There is no new route to restricted content.
No third-party AI tools
Every stage of this journey uses Microsoft 365, Copilot and Copilot Studio. No extra vendors and no data leaving your Microsoft environment.
Governance still matters. Loose permissions become visible quickly once an agent starts answering questions. That is a reason to tidy them, not a reason to avoid agents. The Copilot agents guide covers governance in more detail.
A Copilot agent is Copilot given one job. You give it a name, instructions written in plain English and a set of knowledge sources, such as a SharePoint site. It then answers questions about that job from that content.
The useful ones have named jobs. An agent that answers HR policy questions. An agent that drafts replies in the company tone of voice. An agent that summarises incoming post. Not an assistant for everything, but a colleague's worth of help with one thing.
Because the agent is grounded in your own company data, its answers come from your policies, your templates and your files rather than the open internet. For the full picture of where agents fit, what makes a good use case and how to govern them, read the plain-English Copilot agents guide.
The five stages below are not our sales process. They are stages of your team's capability, and you can stop at any of them. Plenty of organisations get lasting value from stage two or three alone.
At every stage there are two doors. Learn this: we train your people to do it themselves. Have it built: we build it for you, document it and hand it over. Most organisations mix the two, learning the early stages and commissioning the later ones.
The five stages
From first agent to a team of agents.
Each stage is small and specific on its own. Together they compound into something bigger: agents embedded across the business, each doing one named job.
01
Use an agent
Build skill needed: none
Time: minutes
Use an agent is the stage where staff ask an existing agent questions and get answers grounded in company content, with nothing to build or install.
Someone shares an agent with you, or you pick one from the sidebar in Copilot Chat. You ask it questions and it answers from your company's own content rather than the open internet. That difference has a name: grounding. A grounded agent points to the document its answer came from, so you can check it.
There is nothing to build and nothing to install. If your organisation runs on Microsoft 365, this is available now.
What this looks like
- Ask an HR agent what the home working policy actually says.
- Ask a handbook agent how to book annual leave.
- Check which version of a process document is current.
02
Build your first agent
Build skill needed: none, plain English only
Time: about 15 minutes
Build your first agent is the stage where anyone describes a personal agent in plain English and has it working in about 15 minutes.
In Copilot Chat, the agent builder lets you describe an agent in plain English. Tell it the job, point it at a SharePoint site and it writes the instructions for you. This kind of agent is called a declarative agent: instructions plus knowledge, no code.
From a standing start to a working agent takes about 15 minutes, and anyone who can write a clear email can do it. It is how we close our training sessions, because watching your own agent answer from your own documents changes what people think agents are.
What this looks like
- A personal agent that answers from your team's project site.
- An agent that knows one client folder inside out.
- A rough first pass at the HR policy agent before it is shared more widely.
03
Team agents
Build skill needed: careful writing, no code
Time: days, mostly agreeing sources and testing
Team agents is the stage where one agent does one named job for a whole team, with agreed sources and a named owner.
The step up from a personal experiment is an agent a whole team relies on. That takes proper instructions, agreed knowledge sources and a named owner. The agent does one job for one team and does it consistently.
This is where content quality starts to matter. An agent grounded on a tidy, current SharePoint site gives answers people trust. An agent grounded on an old file dump does not.
What this looks like
- Answers HR policy questions for the whole company.
- Drafts replies in the company tone of voice.
- Summarises incoming post for the office team.
04
Copilot Studio
Build skill needed: maker level, trainable
Time: weeks, including proper testing
Copilot Studio is the stage where agents gain connectors, workflows and actions, so they carry out steps rather than just answer.
Copilot Studio adds connectors, workflows and actions. An action is a step the agent can carry out rather than just describe, so agents stop only answering questions and start doing the next step.
Raising a ticket, updating a record, moving a file, triggering a Power Automate flow. Each action runs with agreed review points, so a person still approves anything that matters.
What this looks like
- Takes an IT problem report and raises the ticket.
- Takes a holiday request and starts the approval flow.
- Files a signed contract and updates the client record.
05
A team of agents
Build skill needed: a capability your team owns
Time: grown over months, one agent at a time
A team of agents is the stage where each department runs agents with named jobs, owners and a review rhythm, embedded across the business.
Each agent owns one job. HR has the policy agent, finance has the reporting agent, operations has the intake agent. They work alongside your staff like a set of well-briefed assistants, each with a named owner and a review rhythm.
Everything runs inside your own Microsoft 365 tenant. Every agent inherits the permissions and governance you already have. There is nothing new to secure and no new place your data lives.
What this looks like
- A dozen agents, each with one job, one owner and one set of knowledge sources.
- New agent ideas come from your own team, because by now they know what a good one looks like.
- A review rhythm keeps agents current as policies and processes change.
The whole firm
Three layers, five stages.
Seen from the leadership side, the five stages are also a Copilot adoption journey for the whole firm. It happens in three layers, and each layer maps onto the stages above.
Stages 1 and 2
Everyone uses Copilot
All staff use an agent and build their first one, working confidently with Copilot in Outlook, Word, Excel and Teams. Every training session ends with a real agent built and running in your tenant.
Stages 3 and 4
Some go further
A smaller group starts building for colleagues: team agents first, then Copilot Studio, workflows and automation. Some of the firm, not every member of staff.
Stage 5
The firm embeds it
The business begins embedding agents and automation into its actual processes, one named job at a time. A journey the firm starts and progresses through, not a one-off project.
That is what Copilot for the whole firm looks like in practice: everyone confident in the everyday apps, some building, and the firm steadily embedding what works.
Two tools, one journey
Copilot agent builder vs Copilot Studio.
You do not have to choose one. Most journeys start in the agent builder and move to Copilot Studio when an agent needs to act, not just answer.
Stages 1 to 3
Copilot Chat agent builder
- Built inside Copilot Chat in plain English.
- Declarative agents: instructions plus knowledge sources.
- Answers questions and drafts text. It does not take actions.
- Grounded on SharePoint sites and Microsoft 365 content.
- The right tool for a first agent and for most team agents.
Stages 4 and 5
Copilot Studio
- A full build environment for agents, licensed separately.
- Adds connectors, workflows and actions.
- Agents can raise tickets, update records and trigger Power Automate flows.
- Needs maker-level skills, which teams learn or bring in.
- The right tool once an agent needs to do things, not just answer.
Where to go next
The map is here. These are the transport.
Plain English guide
Microsoft Copilot agents
The companion guide: what agents are, where they fit, good and poor use cases and how to govern them.
Have it built
Copilot Studio Agents
The build service behind the second door: scoping, grounding, testing, rollout and handover.
Learn this
Microsoft 365 Agents Training
A plain-English introduction to agents, use cases and responsible adoption for leaders and teams.
Learn this
Copilot Studio Training
For the people who will own agents with connectors, workflows and actions.
How we deliver
The FiveForward Framework
The journey is your capability map. The framework is how we deliver each step: Discover, Design, Enable, Automate and Embed.
Fixed prices
Pricing
Fixed prices for training, agent builds and ongoing support. No hourly or day rates.
What is a Copilot agent?
A Copilot agent is Copilot given one defined job. It has a name, plain-English instructions and named knowledge sources, such as a SharePoint site, and it answers questions about that job from that content. Agents assist people and run with agreed human review points. They do not replace staff.
Are Copilot agents secure?
Agents run inside your own Microsoft 365 tenant and inherit your existing permissions, so people only ever see content they can already open. Your data is not used to train the underlying models and it does not leave the Microsoft 365 service boundary. No third-party AI tools are involved.
What is the difference between the Copilot agent builder and Copilot Studio?
The agent builder in Copilot Chat creates declarative agents in plain English: instructions plus knowledge, good for answering questions from your own content. Copilot Studio adds connectors, workflows and actions, so an agent can raise tickets, update records and trigger Power Automate flows.
Do we need to be technical to build an agent?
No. A first declarative agent is described in plain English and grounded on a SharePoint site in about 15 minutes. Copilot Studio work needs more skill, which teams either learn through training or hand to a build partner.
Do Copilot agents replace staff?
No. Agents take on defined jobs such as answering policy questions or preparing drafts, and they run with agreed human review points. The aim is fewer repeated interruptions and less manual admin, not fewer people.
Where should we start with Copilot agents?
Use an agent, then build one. The first two stages of the journey need nothing bought and nothing built beyond the Copilot access most Microsoft 365 organisations already have. When a team wants a shared agent doing a named job, choose a door: learn to build it yourselves or have it built for you.
Next step
Work out which stage you are at.
A short call is usually enough to place your team on the journey and pick the right door for the next stage: learn it or have it built.