---
title: How Digital Agencies Use AI to Transform Marketing & Design
description: Half of agencies now run agentic AI. How AI-native agencies use agents, AI search and platform automation, and how to tell real capability from hype.
image: https://blog.madebyxds.com/hubfs/iStock-1480239160.jpg
---

[xds.](https://madebyxds.com/)

[Start a Project](https://madebyxds.com/contact-xds/)

About us

 A founder-driven consultancy built around experience, clarity, and collaboration.

- [More on XDS→](https://madebyxds.com/about-xds/)
- [Insights→](https://blog.madebyxds.com/)
- [Careers→](https://madebyxds.com/xds-careers/)
- [Contact us→](https://madebyxds.com/contact-xds/)

Our services

 Strategy, design, marketing, and technology working together to move healthcare forward.

- [Services overview→](https://madebyxds.com/services/)
- [Brand strategy→](https://madebyxds.com/services/brand-strategy/)
- [Product & experience design→](https://madebyxds.com/services/product-experience-design/)
- [Digital marketing→](https://madebyxds.com/services/digital-marketing/)
- [AI, innovation & technology→](https://madebyxds.com/services/ai-innovation-technology/)

Our work

 See how we turn complex healthcare challenges into meaningful experiences.

- [Featured case studies→](https://madebyxds.com/the-work/)
- [Shockwave→](https://madebyxds.com/the-work/shockwave-medical/)
- [Upneeq→](https://madebyxds.com/the-work/upneeq/)
- [Arcus→](https://madebyxds.com/the-work/arcus-biosciences-dx/)
- [AngioSafe→](https://madebyxds.com/the-work/angiosafe-cardiovascular/)
- [Aidoc→](https://madebyxds.com/the-work/aidoc-seo-strategy/)
- [Seer Bio→](https://madebyxds.com/the-work/seer-bio/)

[Start a Project](https://madebyxds.com/contact-xds/)

 Insights from XDS

# How Digital Agencies Are Using AI to Transform Marketing and Design

Last Updated: 10/5/2026

TL;DR · How agencies actually use AI, late 2026

- **The shift is from tools to agents.** Nine in 10 US agencies use generative AI and half now run agentic AI for marketing execution ([Forrester / 4A's](https://www.hausadvisors.com/blog/ai-adoption-marketing-agencies-statistics)). The question is no longer which chatbot you use, but which workflows run without someone pushing every button.
- **AI search is now the top threat and the top opportunity.** 64% of agencies name Google's AI Overviews as their biggest concern ([AgencyAnalytics 2026](https://agencyanalytics.com/agency-benchmarks-2026)). Impressions rise while clicks fall.
- **Paid media automation is no longer optional.** Google is moving Search campaigns onto AI Max and retiring Dynamic Search Ads, and Meta is targeting fully automated ad creation by year end.
- **Regulators are using AI too.** FDA has said it uses AI tools to surveil drug promotion, including AI-generated health content and chatbots. Healthcare brands need AI workflows that leave an audit trail.
- **Cost discipline is the new differentiator.** Agencies that route work to the right model and reuse client context deliver faster without passing token waste to clients.
- **The right question is still** "which workflows have you rebuilt around AI, and what changed?", now with a follow-up: "what runs on its own, and who checks it?"

Which of your workflows should AI run?

Free 30-minute AI Workflow Teardown. Bring one process, leave with a plan.

[Book a teardown](https://madebyxds.com/contact-xds/)

[How do digital agencies use AI in late 2026?](https://blog.madebyxds.com/how-digital-agencies-use-ai-to-transform-marketing-design-xds#agency-ai-defined)  
[The six AI use cases delivering ROI now](https://blog.madebyxds.com/how-digital-agencies-use-ai-to-transform-marketing-design-xds#five-use-cases)  
[Agency AI adoption benchmarks, 2026](https://blog.madebyxds.com/how-digital-agencies-use-ai-to-transform-marketing-design-xds#adoption-benchmarks)  
[How XDS uses AI today](https://blog.madebyxds.com/how-digital-agencies-use-ai-to-transform-marketing-design-xds#xds-approach)  
[Where AI still won't replace agency work](https://blog.madebyxds.com/how-digital-agencies-use-ai-to-transform-marketing-design-xds#what-ai-cannot-do)  
[How to evaluate an agency's AI capabilities](https://blog.madebyxds.com/how-digital-agencies-use-ai-to-transform-marketing-design-xds#evaluating-an-agency)  
[Frequently asked questions](https://blog.madebyxds.com/how-digital-agencies-use-ai-to-transform-marketing-design-xds#agency-ai-faq)

 

## How do digital agencies use AI in late 2026?

Agencies now use AI in three layers: assistants that draft and analyze, agents that run multi-step workflows across tools, and platform AI inside Google, Meta and search engines that agencies steer rather than operate. The agencies getting results rebuilt specific workflows across all three, and they can show what changed.

In 2025, "using AI" mostly meant a person prompting a chatbot. In 2026 it increasingly means an agent with access to the project tracker, the analytics account and the shared drive, doing a recurring job end to end while a person reviews. Connectors built on the Model Context Protocol (MCP) made that practical: one assistant can now read Asana, Slack, Google Analytics and an ad account in the same task.

Adoption is wide but still shallow. BCG's 2026 survey of 300 CMOs found 42% use generative AI only to assist people with discrete tasks ([BCG](https://www.bcg.com/publications/2026/making-the-agentic-marketing-transformation-a-reality)). That gap between "has AI" and "rebuilt around AI" is where agency value now sits.

For healthcare, pharma and medical device clients, every layer needs governance: what an agent can touch, what it can publish, and who signs off. See [AI-generated pharma content and FDA compliance](https://blog.madebyxds.com/ai-generated-pharma-content-fda-compliance-2026) and [fair balance in AI-generated pharma content](https://blog.madebyxds.com/fair-balance-ai-generated-pharma-content-2026) for the compliance view.

 

## The six AI use cases delivering ROI now

### 1. Agentic workflow automation

Internal operations are the most common agentic use case: 38% of agencies already automate internal workflows with agents, and 79% save five or more hours a week with AI, led by reporting and performance summaries ([AgencyAnalytics 2026](https://agencyanalytics.com/agency-benchmarks-2026)). Think status reports, monitoring cycles, timesheet reconstruction, QA sweeps and invoicing prep.

The value is not the hours saved on one task. It is that recurring work becomes a documented, repeatable runbook instead of tribal knowledge.

### 2. Brand-tuned content generation with guardrails

The first draft is a commodity. The value is in the context layer: client-specific projects or skills that carry approved terminology, voice, claims and, for regulated brands, ISI and fair-balance rules. Generated content respects those constraints before a human editor sees it. It still goes through MLR.

### 3. AI search visibility (GEO / AEO)

AI Overviews, AI Mode and assistants like ChatGPT and Claude now answer many questions before anyone clicks. 51% of agencies are already optimizing for AI-driven search beyond Google, and nearly half struggle to track users who discover brands through AI tools ([AgencyAnalytics 2026](https://agencyanalytics.com/agency-benchmarks-2026)). The work: structured data, entity consistency, quotable answers, third-party citations, and measuring AI referrals separately. See our [GEO guide for healthcare brands](https://blog.madebyxds.com/generative-engine-optimization-geo-healthcare-brands).

How does your brand show up in ChatGPT and AI Mode? [Get an AI visibility check →](https://madebyxds.com/contact-xds/)

### 4. Paid media: steering platform AI, not bidding

Google is replacing Dynamic Search Ads with AI Max and auto-upgrading campaigns that use automatically created assets or campaign-level broad match ([Google](https://blog.google/products/ads-commerce/dsa-upgrade-to-ai-max-2026/)). Google reports an average 7% lift in conversions at similar CPA or ROAS when advertisers use the full AI Max feature set ([JumpFly](https://www.jumpfly.com/blog/ai-in-online-advertising-5-key-trends-from-april-2026/)). Meta has said it aims for fully automated ad creation by the end of 2026.

The agency job is now inputs and guardrails: conversion data quality, brand and URL exclusions, approved asset libraries, and catching AI-generated copy that a regulated brand cannot run. See [pharma PPC and FDA-compliant Google Ads](https://blog.madebyxds.com/pharma-ppc-fda-compliant-google-ads-2026).

Campaigns auto-upgraded this fall? [Get an AI Max risk review →](https://madebyxds.com/contact-xds/)

### 5. Generative design and vibe-coded prototypes

Concept to clickable prototype now takes hours. Tools like [Lovable](https://lovable.dev/), [Replit](https://replit.com/) and Figma's AI features let teams put a working flow in front of clients instead of a static comp, then move to production code with AI coding assistants. The feedback loop changes because stakeholders react to behavior, not pictures.

### 6. Measurement and predictive analytics

The top client question for agencies in 2026 is how to connect marketing to revenue (55%, [AgencyAnalytics](https://agencyanalytics.com/agency-benchmarks-2026)). AI helps with forecasting and anomaly detection, but only on clean, consented data. For healthcare, that starts with a [HIPAA-compliant GA4 setup](https://blog.madebyxds.com/hipaa-compliant-ga4-setup-healthcare-analytics) and the [attribution and measurement guide](https://blog.madebyxds.com/healthcare-marketing-attribution-measurement-guide).

*Chatbots, a standalone use case in earlier versions of this post, now sit inside agentic workflows and regulated-brand compliance. See the [pharma chatbot compliance guide](https://blog.madebyxds.com/pharma-chatbot-compliance-patient-hcp-bots).*

 

## Agency AI adoption benchmarks, 2026

Adoption is near universal; monetization and depth are not. Most agencies use AI, about half run agents, and few have turned it into new revenue.

| Finding | Figure | Source |
| --- | --- | --- |
| US agencies using generative AI / agentic AI for execution | ~90% / ~50% | [Forrester / 4A's, June 2026](https://www.hausadvisors.com/blog/ai-adoption-marketing-agencies-statistics) |
| Agencies that treat AI as a cost of business with limited direct monetization | 61% | [Forrester / 4A's, June 2026](https://www.hausadvisors.com/blog/ai-adoption-marketing-agencies-statistics) |
| Agencies saving 5+ hours a week with AI | 79% | [AgencyAnalytics 2026](https://agencyanalytics.com/agency-benchmarks-2026) |
| Agencies citing Google AI Overviews as top industry concern | 64% | [AgencyAnalytics 2026](https://agencyanalytics.com/agency-benchmarks-2026) |
| CMOs who say GenAI is already reshaping how consumers discover brands | 90% | [BCG, 300 CMOs, 2026](https://www.bcg.com/publications/2026/making-the-agentic-marketing-transformation-a-reality) |
| Organizations scaling an agentic AI system in at least one function | 23% | [McKinsey State of AI, via BusySeed](https://busyseed.com/what-is-agentic-ai-in-marketing-in-2026-moving-beyond-basic-content-generation) |

What this means for buyers: an agency claiming AI capability is the baseline. The differentiator is depth, governance and whether the savings show up in your timeline and budget.

**By agency size (XDS observation)**

| Agency size | Where AI pays off | Typical change |
| --- | --- | --- |
| **Boutique (2-15)** | Agentic ops, content, prototyping, reporting | Senior team covers more accounts without adding junior headcount |
| **Mid-size (16-75)** | Per-client context libraries, paid media guardrails, AI search | Margin on retainers, faster turnaround |
| **Enterprise (75+)** | Governed agent platforms, MLR pre-review integrations | New AI advisory and enablement revenue lines |

 

## How XDS uses AI today

At XDS, AI is how the studio runs: a firm-wide assistant, per-client context, written runbooks that agents execute, and a cost discipline that keeps it efficient. Here is what that looks like in practice.

### A firm-wide assistant connected to our real tools

Every XDS team member can ask an internal Claude assistant about project status, deadlines, files or past decisions. It connects through MCP to Asana, Slack, Microsoft 365 and Dropbox, so answers come back with links to the source. Quick questions route to a fast model; synthesis across many sources routes to a deeper one.

### Per-client context, built once

Each account has its own AI workspace holding stakeholders, terminology, brand voice and compliance rules. For a medical device client, we built a content generator with the FDA-required safety information rules written into its instructions, so drafts arrive already structured for review. No one re-explains the account at the start of every session, and nothing client-specific leaks across accounts.

### Runbooks that agents execute

Recurring work lives in written skills that an agent follows step by step, with a person reviewing the output:

- Bi-weekly social and ad engagement monitoring across LinkedIn, X, Facebook, Instagram and paid channels, run through a browser agent.
- Weekly paid media sweeps across Google, Meta, LinkedIn and programmatic platforms, with pacing checks and change logs.
- Consent and tag diagnostics when a cookie banner or tracking pixel stops firing.
- Timesheet reconstruction and monthly invoicing prep from Harvest, calendars and meeting notes.

### AI search, tested on ourselves first

We track our own sites the way we track clients'. Over the past year, impressions rose 34% while organic clicks fell about 8.5%, a pattern consistent with AI Mode generating impressions that rarely click. AI referral traffic, mostly ChatGPT, grew 34% but recovered only about 5% of the lost sessions. Branded queries held; informational queries collapsed. That is why our [GEO work](https://blog.madebyxds.com/generative-engine-optimization-geo-healthcare-brands) measures citations and branded demand, not just rankings.

### Paid media with human guardrails

We let Performance Max, AI Max and Advantage+ do what they do well: audience expansion, bidding and asset rotation. People own conversion definitions, exclusions, approved copy and creative direction. For regulated brands, any AI-generated ad variant is treated as new copy that needs approval. See our [Aerin Medical paid media case](https://blog.madebyxds.com/optimizing-paid-media-for-healthcare-lessons-from-aerin-medical) for a healthcare example.

### Attribution and consent before creative

Our rule for regulated clients still stands: fix measurement before adding AI to creative. Consent mode, a working consent platform, server-side tagging where it fits, and clean UTMs come first. A predictive model trained on broken or non-consented data is a liability.

### TokenOps: AI cost discipline

We run AI usage like any other cost line, in four stages: see which accounts and workflows use the most, reuse context instead of rebuilding it, route routine tasks to lighter models and save frontier models for strategy, then pass the efficiency to clients as speed. It is part of onboarding for every new account lead.

Want this operating model on your account?

Per-brand AI context, agent runbooks and AI search measurement. Start with one workflow.

[Start a project](https://madebyxds.com/contact-xds/)

Run an agency yourself?

We're a white-label partner for AI development, SEO, paid media and analytics.

[Partner with us](https://madebyxds.com/contact-xds/)

 

## Where AI still won't replace agency work

Agents take over execution, so judgment and accountability are worth more, not less.

- **Strategic positioning.** AI summarizes a category; it cannot argue for a defensible position for one brand.
- **Regulatory judgment.** FDA does not give AI-generated promotion a separate lane, and it has said it uses AI to surveil drug ads, including AI-generated health content and chatbots ([Latham & Watkins](https://www.lw.com/en/insights/fda-begins-crackdown-on-direct-to-consumer-pharmaceutical-advertising)). AI can speed MLR review; it cannot own it. See the [OPDP submission and review guide](https://blog.madebyxds.com/opdp-submission-process-pharma-fda-review-guide).
- **Deciding what an agent may do.** Permissions, review gates and incident response for agents are human design decisions. They are now part of the deliverable.
- **Creative territory for launches.** AI generates strong variants. Choosing the territory a launch should own still takes research and a strategist.
- **Senior client relationships and ethical calls.** Trust and reputational decisions remain human accountability.

 

## How to evaluate an agency's AI capabilities before hiring

Every agency claims AI. These seven questions separate rebuilt operations from slideware. The longer version is in [5 questions to ask before you buy an agency's AI pitch](https://blog.madebyxds.com/5-questions-to-ask-before-you-buy-an-agencys-ai-pitch).

1. **Which workflows have you rebuilt around AI, and what changed?** Expect three to five named workflows with before and after numbers.
2. **What runs on its own, and who reviews it?** Ask to see one agentic workflow and its review step.
3. **How do you hold our brand voice, terminology and compliance rules in your AI stack?** Per-client projects or skills are a real answer. "We tell people not to paste client data" is not.
4. **What can your agents access, and how is our data separated from other clients'?** Ask about connector permissions and logging.
5. **How do you measure our visibility in AI search?** If the answer is rankings only, they are measuring last year's web.
6. **What is your AI governance policy, and how does AI-assisted content reach MLR?** Regulated brands should require it in writing.
7. **How do you manage AI cost, and does the efficiency reach our invoice?** An agency without cost discipline either eats the waste or bills it to you.

Score your current agency

The seven questions, what strong answers sound like, and the red flags. One page.

[Request the scorecard](https://madebyxds.com/contact-xds/)

The healthcare version of this checklist is in [how to choose a healthcare marketing agency](https://blog.madebyxds.com/how-to-choose-healthcare-marketing-agency).

 

## Frequently asked questions about agencies and AI

### What is agentic AI in a marketing agency?

Agentic AI completes multi-step tasks across tools instead of answering one prompt. An agent can pull a week of campaign data, check pacing against plan, draft the status note and flag anomalies for a person to review. The agency's job shifts to designing the workflow, setting permissions and reviewing output.

### Which agencies combine AI with regulated healthcare experience?

Look for an agency that can show three things together: AI workflows with documented review gates, experience with MLR and FDA promotional rules, and brand, UX and engineering under one roof so AI is designed into the experience rather than added later. The Experience Design Studio (XDS) is built for that mix, working with pharma, biotech, medical device and digital health brands. [Book a teardown](https://madebyxds.com/contact-xds/) to see how we'd apply it to your workflow.

### Should I hire a specialist AI agency or a full-service agency that uses AI?

For most brands, a full-service agency that has rebuilt workflows around AI is the better choice. Specialist AI agencies often deliver excellent point solutions but lack the strategic and creative discipline to run a full brand program. The exception is a well-defined AI-specific problem, such as deploying a compliant conversational agent behind a regulated brand. For regulated industries, always require the agency to demonstrate governance and MLR-safe workflow experience.

### What AI tools do agencies actually use in 2026?

Typically: Claude and ChatGPT for reasoning, drafting and agents; MCP connectors into project, chat, file and analytics tools; AI coding assistants plus Lovable or Replit for prototypes; Google AI Max, Performance Max and Meta Advantage+ for paid; Figma, Adobe Firefly and Canva for design; and meeting intelligence such as Otter.ai. The stack matters less than which workflows it runs.

### Is AI-generated content safe for pharma and medical device marketing?

Only when it goes through the same MLR review as any other promotional content. FDA has no AI-specific promotional rule; existing requirements apply no matter who wrote the first draft. Treat AI output as a draft carrying the same liability, and keep a record of how it was produced. Details: [fair balance in AI-generated pharma content](https://blog.madebyxds.com/fair-balance-ai-generated-pharma-content-2026).

### How should we respond to Google AI Max and Meta's automation push?

Don't fight it; feed it well. Clean conversion data, strict brand and URL exclusions, approved asset libraries and a human check on AI-generated copy matter more than manual bidding. Regulated brands should confirm which auto-generated text features are on before campaigns are upgraded.

### Is SEO dead now that AI answers the question?

No, but the goal changed. Informational clicks are falling while branded demand and AI citations matter more. Optimize to be the cited source: clear answers, structured data, consistent entity information and third-party proof. Measure AI referrals and branded search alongside rankings.

### How does AI change agency pricing?

Execution-heavy work compresses in price; strategy, AI infrastructure and governance become premium. Only 31% of agencies plan to monetize agentic AI within 24 months ([Forrester / 4A's](https://www.hausadvisors.com/blog/ai-adoption-marketing-agencies-statistics)). Ask your agency how efficiency gains show up in your scope or timeline.

### Can AI replace an in-house marketing team?

Not for most brands. It replaces execution volume, not judgment, stakeholder navigation or accountability. Expect leaner in-house teams working with AI-native partners.

### What is the biggest AI risk for regulated healthcare brands working with agencies?

Off-label or unbalanced content moving through AI-assisted workflows without the same MLR scrutiny as traditional content. If an agency's workflow bypasses or outpaces review, the brand carries the enforcement risk. Require your agency to show the review path for AI-assisted content before signing.

 

## See what AI should be doing for your brand

The Experience Design Studio is an AI-native [digital agency](https://madebyxds.com/) for healthcare, pharma, medical device and biotech brands. We build AI into strategy, creative, engineering and marketing, with MLR-safe review paths and measurement you can trust.

Start with a free 30-minute AI Workflow Teardown

1. Pick one recurring workflow.
2. We map what an agent takes over and what stays human.
3. You leave with a one-page plan, either way.

[Book your teardown](https://madebyxds.com/contact-xds/)

Prefer to start smaller? [Request the Agency AI Scorecard](https://madebyxds.com/contact-xds/) or [see our healthcare work](https://madebyxds.com/the-work/).

Related reading: [practical AI in regulated healthcare](https://blog.madebyxds.com/practical-ai-in-regulated-healthcare-a-marketers-guide-xds-blog), [AI-generated pharma content and FDA compliance](https://blog.madebyxds.com/ai-generated-pharma-content-fda-compliance-2026), [AI sales enablement for healthcare, pharma and medtech](https://blog.madebyxds.com/ai-sales-enablement-healthcare-pharma-medtech), [AI in healthcare marketing](https://blog.madebyxds.com/ai-in-healthcare-marketing), and [how to choose a healthcare marketing agency](https://blog.madebyxds.com/how-to-choose-healthcare-marketing-agency).

## Related insights

[All insights →](https://blog.madebyxds.com/)

### [Clinical Evidence Messaging for Medical Devices (2026)](https://blog.madebyxds.com/clinical-evidence-messaging-for-medical-devices-2026)

### [Patient vs. HCP Traffic on Trial Sites: What the Data Shows](https://blog.madebyxds.com/trial-site-traffic-patients-or-hcps-no-public-split-found)

### [Ensuring Cookie Consent Compliance for Life-Science Websites](https://blog.madebyxds.com/cookie-consent-for-life-sciences-websites-configuring-gtm-and-ga4-without-breaking-analytics-or-inviting-a-lawsuit)

 We’ll Navigate the Chaos

## Let’s work together.

[![XDS](https://blog.madebyxds.com/hubfs/raw_assets/public/Custom/blog/images/xds-logo.png)](https://madebyxds.com/)

 The **Experience Design Studio** (XDS) is a strategic consulting agency for modern healthcare experiences.

- [About us](https://madebyxds.com/about-xds/)
- [Services](https://madebyxds.com/services/)
- [Our work](https://madebyxds.com/the-work/)
- [Insights](https://blog.madebyxds.com/)
- [Contact](https://madebyxds.com/contact-xds/)
- [Careers](https://madebyxds.com/xds-careers/)
- [Linkedin](https://www.linkedin.com/company/experiencedesignstudio/)
- [Dribbble](https://dribbble.com/madebyxds)
- [Instagram](https://instagram.com/madebyxds)

© 2026 The Experience Design Studio, Inc. All rights reserved. | [Privacy Policy](https://madebyxds.com/privacy-policy/)

```json
{
  "@context" : "https://schema.org",
  "@type" : "BlogPosting",
  "author" : {
    "@type" : "Person",
    "name" : "Shai Reichert",
    "url" : "https://blog.madebyxds.com/author/shai-reichert"
  },
  "dateModified" : "2026-10-05T16:22:07.072Z",
  "datePublished" : "2025-03-10T19:35:09.000Z",
  "headline" : "How Digital Agencies Use AI to Transform Marketing & Design",
  "mainEntityOfPage" : {
    "@id" : "https://blog.madebyxds.com/how-digital-agencies-use-ai-to-transform-marketing-design-xds",
    "@type" : "WebPage"
  },
  "publisher" : {
    "@type" : "Organization",
    "logo" : {
      "@type" : "ImageObject",
      "url" : "https://blog.madebyxds.com/hubfs/logo.svg"
    },
    "name" : "The Experience Design Studio"
  }
}
```