At a recent Second Wind JAM Session on AI and automation, members shared where the tools are making a difference inside their agencies, where they are running into friction with clients, and how they are thinking about staying ahead. Here's what came up.
Moving AI Beyond the Creative Department
For most agencies, AI adoption started in creative and copywriting. Several members are now pushing it further into account service, operations, and new business development.
- One agency uses Microsoft Copilot to record client meetings and generate draft project briefs from the transcript. The output requires human review, but it has improved efficiency. The choice of Copilot was intentional: working in regulated industries with competitively sensitive information made data security a priority, and the Microsoft 365 integration keeps data from being used to train outside models.
- Another agency used Claude to convert complex monthly reconciliation spreadsheets into clean, readable client reports. Clients noticed the difference right away, and work that had been painful to produce became something people could actually engage with.
- One agency built per-client knowledge bases inside a ChatGPT workspace, uploading brand foundations, personas, and source materials for each client. Anyone with shared access can query within that client's specific context rather than starting from a generic prompt, producing more consistent output and keeping the team from importing information that shouldn't be there.
- On the new business side, one agency with deep anniversary campaign experience used Perplexity, Claude, and a third tool to research which companies in their agricultural niche were approaching milestone years, 25th, 50th, 75th, and 100th anniversaries, and built a prioritized prospect list overnight. The result was targeted outreach built around a service the agency already does well, timed to land when prospects are most likely to be thinking about exactly that kind of work. Any agency with a defined specialty can run the same play.
When Clients Think They Don't Need You
The most sustained discussion was about clients who have decided AI makes agencies optional. A few members shared how those situations have played out.
- One agency nearly lost a long-term client after an acquisition brought in new leadership convinced they could build a website with AI for free. Rather than push back, the agency offered to do the website at no charge in exchange for a monthly services engagement. The long-term relationship was worth more than the project fee, and they are close to signing.
- Another member described a client who objected to the price of a PowerPoint template, arguing AI could produce one in 15 minutes. The question the room landed on: if AI can handle certain deliverables cheaply, is the right move to defend the price or redirect the relationship toward higher-value work and let the lower-end stuff go?
- Several members are seeing clients use AI to generate content or design assets, then return to the agency when they cannot edit the output, integrate it with other files, or make it functional. Framing the agency's value around results rather than process tends to land better in those moments, as does grounding the conversation in audience, intent, and what the work actually needs to accomplish.
Staying Ahead of the Tools
Keeping pace with AI requires more than curiosity. Several members have made deliberate structural investments to stay ahead of it.
- One agency put together a five-person internal workgroup focused on AI, charged with identifying where it adds value and where it produces genuinely better work. The group treats prompt expertise as a real skill, recognizing that what goes in determines what comes out.
- Another agency fed roughly 18 months of accumulated scoping spreadsheets into a private Claude instance and built a working pricing model that calculates budgets, hours by role, and profit margins for any service combination. Proposals that used to take significant time can now be generated and adjusted quickly.
- Several agencies have formal policies, or are developing them, covering what can and cannot go into AI tools: no financials, no personally identifiable information, no HIPAA-relevant client data. For client work, staff use enterprise or private instances rather than public-facing tools.
- One agency's internal framework sums it up simply: human, AI, human. A person defines the input, AI does the work, and a person reviews the output before it goes anywhere near a client. That structure has also helped ease staff anxiety about displacement by making the human role explicit on both ends.
The Longer View
When Adobe's Creative Suite became widely available, clients gained access to the same design tools agencies had been using. Some tried doing their own creative work. Most came back, because access to the tools was never the same thing as knowing how to use them. AI is a bigger version of that story, with higher stakes and faster timelines. The agencies investing now in understanding how to use these tools, not just that they exist, are building the kind of expertise that kept clients coming back then.
