This episode of the Jukebox Podcast features Matt Schwartz, founder of Inspry, an Atlanta WordPress and WooCommerce agency, and creator of CheckView, a WordPress testing product. The conversation focuses less on AI creating sites or content, and more on how agencies can use AI to improve processes, workflows, and client value behind the scenes.
Matt’s background: he started building sites as a kid, worked with Drupal in university, and launched Inspry in 2011. Over the years he discovered that running an agency is as much about designing repeatable processes as it is about building sites. The arrival of AI, he says, shifts attention from execution to the higher-level value an agency delivers.
Why agencies should care now
Execution is becoming a commodity. Simple brochure sites and routine builds are increasingly automated or accelerated by tools, and clients are aware of that. The competitive edge will come from clearer workflows, more useful touchpoints, and services that actually move the client needle. AI can widen the gap between agencies that use it thoughtfully and those that ignore it, but Matt warns that adoption should be intentional, not reactive.
Disclosing AI use to clients
Matt recommends transparency when AI affects the delivered product. If AI is creating content or influencing product behavior, clients should know. For internal processes and automation, disclosure is not always necessary unless the client is directly involved in redesigning their internal workflows with AI.
An AI vision document
Rather than experimenting blindly, Matt suggests creating an AI vision document: a written plan that maps an agency’s processes, highlights repetitive or time-consuming tasks, and sets guardrails for where AI should and should not be used. This helps teams know where to apply AI, where human judgment is mandatory, and how to reduce risk to reputation and client outcomes. The vision document can be informed by data, for example by analyzing time-tracking to spot repetitive tasks AI could speed up.
AI as a core service offering
Agencies that develop AI expertise can productize that skillset and sell it to clients. This might mean building lightweight custom web apps, automations, or process improvements that previously would have been unaffordable. Use cases include automating internal workflows with tools like n8n, creating custom automations that save clients time or money, and offering ongoing managed automation as a recurring service. Matt cautions against yolo building without due diligence; security and maintainability matter.
Data analysis and marketing
AI excels at pattern recognition across large datasets. Agencies can use that ability to deliver insights previously out of scope for small budgets: segmenting customers, analyzing support tickets, or optimizing geotargeted marketing. Offering this kind of data-backed guidance unlocks new value that clients can immediately appreciate.
AI inside agency operations
Low-hanging fruit includes meeting summaries, automated drafts of proposals and statements of work, and synthesizing discovery notes with client site data. AI can assemble richer, more tailored proposals by combining meeting transcripts, the client website, and past proposals. Teams should still review and refine AI outputs, but AI reduces the time required to create higher-quality drafts.
AI for support workflows
AI can improve support by aggregating context from helpdesk records, project management, server logs, and the site itself to suggest likely causes and initial resolutions. Matt stresses this should augment, not replace, humans. Support interactions are high-friction moments; misstepping with pure bot-first approaches can frustrate users. Well-designed AI can give support staff the context they need to resolve issues faster and handle more tickets without making customers feel ignored.
AI-assisted debugging and WordPress management
Connecting AI agents to WordPress via REST APIs, hosting APIs, or specialized tools can speed troubleshooting and reduce time spent on one-off technical problems. These agents can analyze logs, pull in related tickets, and propose fixes or code changes. Matt still insists on human review based on risk: agents can get you most of the way to a solution, but humans should validate edits, check the live site, and take responsibility for final changes.
WordPress as infrastructure for AI
Rather than absorbing everything, the WordPress ecosystem is becoming a foundation that AI tools can talk to. As tools and page builders evolve to expose structured interfaces, agents can operate against known schemas and block formats. That makes it easier to manage content and even perform edits via AI, but again, Matt recommends human oversight and incremental adoption.
Ethics, trust, and the human-in-the-loop
Matt emphasizes guardrails and intention. AI can be powerful, but it has limits and risks: hallucinations, security gaps, and unpredictable outputs. Agencies should set policies for where AI is allowed, require human review for higher-risk tasks, and avoid overpromising. An AI vision document helps enforce these practices and keeps teams aligned.
Part one wrap-up
This episode covers eight of sixteen points Matt compiled about AI in agencies: the big shift from execution to value, why now, disclosure, the AI vision document, AI as a service offering, data analysis, AI inside operations, support workflows, and AI-assisted debugging. It establishes a recurring theme: use AI to extend what an agency can offer, not to replace the human judgment that clients pay for.
This is part one of a two-part conversation. In part two, they will continue through the remaining topics and examples for agencies adopting AI. If you run a WordPress agency or want to understand how AI might help your processes, listening to both parts will give practical ideas for careful, valuable adoption.