Almost every agency team already uses AI daily. But hardly any have a real plan.
Successful agencies roll out AI in 6 phases: from generic AI use to the agentic agency.
This framework shows you where you stand right now and what the next phase needs. 🚀
Why does AI in agencies need a plan?
Because AI use and AI readiness are two different things.
98% of agency staff use AI every day, but only 16% of agencies actually feel ready for it.
The 6 phases turn scattered, one-off prompting into a plannable path.
AI is now a fixed part of agency life: around half of all creative work will soon be done by AI (agency teams say so themselves), and a growing share of clients actively demand the use of AI on new projects.
Yet the rollout stalls in most agencies - not because people prompt too little, but because knowledge, tools and agents aren't shared, structured or billable. And so they're never really built into how the agency is run.
The way out isn't another chatbot model, it's structure.
Successful agencies take the path in 6 phases- and the good news: you don't have to leap, you just have to place yourself once and take the next step.
Phase 1: Generic AI use
Status: Individual users rely on a general AI chat tool for everyday tasks: writing copy, researching, analysing, crunching data, brainstorming, writing code. Everyone on their own, everyone figuring it out for themselves. The value is immediate, but none of it is shared or repeatable.
How you reach phase 2: As soon as people start sharing their best prompts, tool tips and results with each other instead of keeping them to themselves. That needs a place where results can be compared and shared.
Phase 2: Word-of-mouth best practices
Status: Your team talks about AI: word-of-mouth tool recommendations, inspiring use cases, lots of experimenting, casually swapping experiences. Real progress, but the knowledge lives in people's heads and chat threads rather than in a structure.
👉 This is where most agencies are.
How you reach phase 3: When you bring the scattered knowledge together centrally: agency, client, brand and project context in one place the AI can access. Water-cooler chat becomes a shared knowledge structure.
Phase 3: Agency knowledge
Status: Multiple AI tools work with your agency's central knowledge. The context is shared and available:
- Agency context: processes & guides
- Client context: rates, people
- Brand context: design, voice
- Project context: briefing, timing
This makes results relevant instead of generic.
How you reach phase 4: When the AI doesn't just know this context but acts on it inside your tools, connected to your (industry) software via MCP and integrations.
Phase 4: Agency tools
Status: Assistants now carry out actions in your systems. The AI no longer just talks, it acts.
For example:
- reading the CRM
- calculating a quote in the ERP
- creating & planning a project in the PM tool
- creating tasks
- updating the team in Teams
How you reach phase 5: When individual actions turn into reusable agents: knowledge, tools and process bundled together, built in the agency and shared across the team. That needs model-independent AI infrastructure, clear access rights and context awareness.
Phase 5: Specialised agents
Status: The agency builds its own agents for specific jobs and shares them across the team. Each one bundles knowledge, tools and process:
- Research agent
- Brand-safety agent
- Briefing agent
- Calculation agent
How you reach phase 6: When these agents stop working in isolation and start playing together in complex workflows across multiple disciplines, orchestrated by the team.
Phase 6: The agentic agency
Status: People orchestrate, agents execute. Multiple agents work together on complex workflows across disciplines. Users make the decisions, agents act accordingly, and the team can build and run workflows on its own.
Behind it sits a simple principle: your people don't reinvent AI over and over.
The agency provides the knowledge structure (= context), tools and agents — so everyone can focus more productively on the actual value they create.
The 6 phases at a glance
- 1: Generic AI use: individuals sharpen their skills. → move on once results get shared
- 2: Word-of-mouth best practices: experiences shared informally. → move on once knowledge becomes central (this is where most agencies are)
- 3: Agency knowledge: AI works with central context. → move on once AI acts in your tools
- 4: Agency tools: AI acts in your systems. → move on once actions become agents
- 5: Specialised agents: your own, shared agents. → move on once agents play together
- 6: Agentic agency: people orchestrate, agents deliver.
So each phase requires its own tools and structures – from AI chat to workflow orchestration.
All the tools required for these six phases are part of awork. This means you can build your AI readiness using a single platform rather than a patchwork of individual tools.
👉 The takeaway: your agency doesn't become AI-ready through more prompts, but through a team-shared structure.
