Planning an AI strategy for your agency is difficult when the technology changes every quarter. Last year’s best tools are this year’s legacy systems.
But underneath the noise, there are clear trajectories. The agencies that will thrive in 2028 are not the ones chasing every new release. They are the ones building adaptable systems and investing in the capabilities that compound over time.
Here is a practical roadmap based on what we are seeing across the agencies we work with.
What is mature now (exploit these)
These AI capabilities are reliable, cost-effective, and ready for production use in agency workflows. If you are not using them already, you are behind.
Content generation and editing. Large language models produce usable first drafts for most written content: blog posts, social copy, email sequences, proposals, reports. The quality gap has narrowed to the point where editing time is minimal. Build it into every content workflow.
Research and analysis. AI-powered research (competitor analysis, market trends, audience insights, data synthesis) is faster and more thorough than manual methods. If your strategists are still doing desk research manually, they are spending time on the wrong work.
Code assistance. AI pair programming (Copilot, Cursor, Claude Code) accelerates development by 30 to 50% on routine tasks. For web development agencies, this is the baseline expectation for productivity.
Reporting and data visualisation. AI can pull data from multiple sources, identify trends, generate insights, and draft report narratives. The human role is curation and strategic interpretation, not data wrangling.
Administrative automation. Meeting transcription, summary generation, follow-up emails, status updates, timesheet processing. Fully automatable with current tools.
What is maturing (invest in these now)
These capabilities are moving from experimental to production-ready. Agencies that build expertise here now will have a significant advantage by 2027.
AI agents and multi-step workflows. Autonomous AI agents that can execute multi-step tasks (research a prospect, draft an email, schedule a follow-up, update the CRM) are becoming reliable enough for production use. The technology is not perfect, but it is good enough for supervised workflows where a human checks the output before it ships.
Build your agent literacy now. Learn how to design agent workflows, set appropriate guardrails, and build human-in-the-loop checkpoints. Within 18 months, agencies will be selling agent-powered services to clients.
AI-powered design tools. Beyond image generation, AI is becoming useful for layout generation, design system management, responsive adaptation, and asset production. Figma’s AI features are improving rapidly. For design agencies, the investment is in learning how to integrate these without sacrificing quality.
Voice and audio AI. Text-to-speech, voice cloning (with consent), podcast editing, and audio content production are reaching commercial quality. Agencies producing audio content for clients should be experimenting here.
Personalisation at scale. AI can generate genuinely personalised content variations (email sequences, landing pages, ad copy) tailored to specific audience segments. This is moving beyond simple variable substitution into contextually aware personalisation.
What is emerging (watch these, do not bet on them yet)
These are exciting but not reliable enough for client-facing work. Experiment internally, but do not sell them.
AI video generation. Tools like Sora, Runway, and Kling produce impressive short clips. They are not reliable enough for commercial work. Quality is inconsistent and brand control is limited. By 2028, this will likely be production-ready. For now, it is a demo, not a deliverable.
Real-time AI collaboration. The vision of AI as a live collaborator in meetings and brainstorming sessions is tantalising but early. Genuine real-time creative collaboration is 18 to 24 months away from being useful.
Autonomous client-facing AI. Chatbots and AI assistants that interact directly with clients on behalf of your agency. The risk-reward ratio is still poor. One bad interaction damages the relationship. Keep AI behind the scenes.
Building adaptable systems
The biggest mistake agencies make is building their AI strategy around specific tools. Tools change. Capabilities endure.
Invest in capabilities, not products. Train your team on prompt engineering principles, not on “how to use ChatGPT.” Teach them workflow design, not “how to use Zapier.” The specific tools will evolve; the skills transfer.
Build modular workflows. Design your AI-augmented processes so that any individual tool can be swapped out without rebuilding the entire workflow. Use APIs and middleware rather than depending on a single platform. When (not if) a better tool emerges, you should be able to switch in days, not months.
Document everything. Your AI workflows, prompt libraries, and process documentation are intellectual property. They are also insurance. When a tool changes its API, raises its prices, or gets acquired, your documentation lets you rebuild on a different platform quickly.
Team development priorities
The skills your team needs in 2028 are different from the skills they need today. Start the transition now.
Priority 1: AI literacy for everyone. Every team member should be comfortable using AI tools in their daily work. This is not optional and not role-specific. Designers, developers, strategists, project managers, account managers. Everyone.
Priority 2: Workflow design for senior staff. Your senior people should be able to design AI-augmented workflows, not just use AI tools. This means understanding what AI can and cannot do, how to build effective prompts, and how to design human-in-the-loop processes.
Priority 3: AI strategy for leadership. Agency leaders need to understand how AI changes agency economics, pricing, positioning, team structure, and client expectations. This is not a technology conversation. It is a business strategy conversation.
What clients will expect by 2028
Plan for these expectations now.
Speed. Turnaround times that seem fast today will be standard by 2028. Clients will expect research in days, not weeks. Content in hours, not days. Reports in real-time, not monthly.
Personalisation. Generic deliverables will feel lazy. Clients will expect data-driven personalisation in campaigns, content, and strategy. “One size fits all” will be a reason to switch agencies.
Transparency. Clients will ask how you use AI. They will want to understand your methodology, your quality controls, and your ethical framework. The agencies that treat AI as a black box will lose trust.
Proactive insight. Clients will expect their agency to surface opportunities and threats before being asked. AI-powered monitoring and analysis makes this feasible. Agencies that only respond to briefs will be replaced by agencies that anticipate needs.
The 90-day starting point
Do not try to build a two-year roadmap in a week. Start with 90 days.
- Audit your current AI usage. Where are you using AI? Where are the gaps? Where are you still doing things manually that AI could handle?
- Pick two maturing capabilities from the list above and run pilot projects.
- Start a team training programme. One hour per week, structured, with specific skill targets.
- Document your workflows. Start building the institutional knowledge that will compound over time.
The agencies that win in 2028 will not be the ones that adopted the flashiest tools. They will be the ones that built systems, skills, and habits that adapted as the technology evolved.
Start building. Start now.
This is part of The Pitch Stack, a series on winning new business with AI. Subscribe to the newsletter to get new articles weekly.
