Gartner expects over 40% of agentic AI projects to be canceled by the end of 2027. Adobe’s agentic coworker for marketing is about to land on that exact battlefield. Before you sign an enterprise rollout, you need a pilot that can fail cheaply. Here is how to build one.
What Adobe actually announced
At Summit, Adobe announced CX Enterprise Coworker, an agentic layer that works across Adobe Experience Platform and apps such as Real-Time CDP, Customer Journey Analytics and Journey Optimizer. Adobe describes it as autonomous with human oversight, built on open standards including Model Context Protocol (MCP) and Agent2Agent (A2A).
The pitch: agents monitor signals, recommend next-best actions and execute across channels based on goals you define. Adobe says it will also operate alongside AI platforms from AWS, Anthropic, Google Cloud, Microsoft and OpenAI.
One detail matters for your calendar. Adobe says it will be generally available in the coming months. So you are evaluating a promise, not a shipped product. That changes how you scope the work.
An announcement is not a capability. Treat every agentic feature as unproven until it has run on your data, under your governance, against your baseline.
Why a pilot, and why now?
Gartner’s cancellation forecast points to three causes: escalating costs, unclear business value and inadequate risk controls. None of those are model problems. They are operating problems, and a pilot is the cheapest place to find them.
There is a second reason. Gartner has flagged “agent washing,” where vendors rebrand existing assistants, RPA and chatbots as agentic. Gartner estimates that only about 130 of the thousands of agentic AI vendors are real. Adobe is a serious vendor. Still, your procurement team should ask which behaviors are truly autonomous and which are scheduled automation with a new label.
We made a similar argument about the creator side of this stack in our piece on Adobe’s creator gap. The logic carries over here: define failure before you start.
The pilot framework: five decisions before day one
1. Pick one narrow workflow
Do not pilot “personalization.” Pilot one loop. A good candidate: a single lifecycle journey in Journey Optimizer where the agent recommends the next action and a human approves it. Narrow scope makes attribution cleaner and cost visible.
Avoid anything touching paid creator payouts or legal claims in round one. Those carry compliance exposure that a pilot should not absorb.
2. Set the baseline first
Without a control group, you will credit the agent for seasonality. Hold out a comparable segment running the current manual or rules-based process. Measure the same things in both arms:
- Outcome: conversion, retention or revenue per contact, whichever your journey is built to move.
- Efficiency: hours of practitioner time per campaign cycle, from brief to launch.
- Quality: share of agent recommendations your team approves without edits.
- Cost: licensing, implementation and review labor, all in one number.
That last line item is where projects die quietly. Review labor is real spend. If your senior strategists spend more time checking the agent than they saved, you have bought a very expensive intern.
3. Write the guardrails as rules, not intentions
Adobe says humans stay in the loop. Fine. Decide exactly where. Which actions can the agent execute alone? Which need approval? Which are off limits (suppression lists, offer eligibility, regulated claims)? Put it in a one-page permissions matrix and have legal sign it.
Identity is the sleeper issue. An agent acting on customer profiles needs its own accountable identity, and we cover that in agent and customer identity. If agents share context across tasks, set retention rules too, as outlined in our guide to shared agent memory.
4. Audit the data the agent will touch
Adobe says the coworker draws on Real-Time CDP, Customer Journey Analytics, Marketo Engage, Target, CRM platforms and external signals from social and news. Every source you connect widens the blast radius. Connect the minimum set for your pilot workflow and nothing more.
Ask a blunt question: if the agent recommends something odd, can we trace which input drove it? If the answer is no, you cannot debug it, and you cannot defend it to your CFO.
5. Write kill criteria in advance
This is the step teams skip. Agree on thresholds before launch, in writing, with your finance partner. For example:
- Stop if approved-without-edit recommendations stay below a threshold you set for two consecutive review cycles.
- Stop if the treatment arm fails to beat the holdout on your primary outcome metric.
- Stop if any guardrail breach occurs that legal classifies as material.
- Stop if total cost per incremental outcome exceeds the manual process.
Numbers are yours to set. The discipline is the point. For a deeper treatment of this approach, see our note on pilots that can fail.
Questions to put to Adobe before you start
Vendor demos show best-case runs. Your pilot brief should ask for the rest.
- Which capabilities are generally available today, and which are roadmap? Get the answer in writing.
- What logging exists for each agent action, and can we export it?
- How are third-party model calls (via MCP or A2A) priced, and who pays?
- What happens to our data if we exit?
- Can we run the pilot in a sandbox with a subset of profiles?
The open-standards angle is attractive because it suggests less lock-in. Test that claim directly. Ask a technical lead to connect one non-Adobe agent during the pilot and see how much friction appears.
Who should sit at the table?
A pilot owned only by marketing ops will miss risk. A pilot owned only by IT will miss value. Build a small group: a marketing lead who owns the outcome metric, a data owner, a legal or privacy reviewer, and a finance partner who owns the kill criteria. Give one person final call.
If AI-generated insights feed budget decisions, hold them to the same standard. Our five checks for AI insights work well as a review template for the agent’s recommendations.
What about creators and influencer programs?
Adobe’s coworker is a customer experience tool. It is not an influencer platform. Even so, the signals it consumes (social, news, loyalty behavior) overlap with what influencer teams track. If you want to link creator-driven demand to journey outcomes, treat that as a phase-two test, not a pilot objective.
Matching and vetting creators with agents is a separate problem with its own failure modes. We cover it in agentic creator matching.
From pilot to rollout: the go/no-go gate
At the end of the pilot, hold one meeting with one question: did the treatment arm beat the baseline, net of all costs, without a material guardrail breach? If yes, expand to a second workflow, not to the whole enterprise. If no, you have spent a small budget to avoid a large one. That is a win too.
Document what you learned either way. The record is your best defense in the next renewal negotiation.
FAQs
What is Adobe CX Enterprise Coworker?
It is Adobe’s agentic AI offering for customer experience orchestration, announced at Adobe Summit. It sits across Adobe Experience Platform and related apps, and is designed to monitor signals, recommend actions and execute across channels with human oversight. Adobe says general availability is expected in the coming months.
How long should a pilot of an agentic marketing tool run?
Long enough to cover at least several full review cycles of your chosen workflow, so you can see variance and not a single lucky result. Set the duration up front with your finance partner, and do not extend it just because early results look ambiguous.
What should a pilot measure?
Four things: the business outcome versus a holdout group, practitioner time saved, the rate of agent recommendations approved without edits, and total cost including review labor. Measure all of them in both the agent arm and the baseline arm.
What is agent washing, and why does it matter for procurement?
Agent washing is Gartner’s term for vendors rebranding existing assistants, RPA or chatbots as agentic AI without real autonomous capability. It matters because you could pay agentic prices for automation you already own. Ask vendors to demonstrate autonomous behavior on your data.
Does Adobe’s open architecture remove lock-in risk?
Not by itself. Adobe says the product is built on MCP and A2A, which supports interoperability in principle. Verify it by connecting a non-Adobe agent during the pilot and confirming you can export logs and data.
Should influencer marketing teams join the pilot?
Observe it, but keep creator payouts and claims out of the first workflow. Once the core loop proves value, test whether creator-driven signals improve journey performance in a second phase.
Next step: this week, draft a one-page pilot charter with one workflow, one holdout group, one permissions matrix and written kill criteria, then send it to Adobe as the condition for any trial.
Top Influencer Marketing Agencies
The leading agencies shaping influencer marketing in 2026
Agencies ranked by campaign performance, client diversity, platform expertise, proven ROI, industry recognition, and client satisfaction. Assessed through verified case studies, reviews, and industry consultations.
Moburst
-
2
The Shelf
Boutique Beauty & Lifestyle Influencer AgencyA data-driven boutique agency specializing exclusively in beauty, wellness, and lifestyle influencer campaigns on Instagram and TikTok. Best for brands already focused on the beauty/personal care space that need curated, aesthetic-driven content.Clients: Pepsi, The Honest Company, Hims, Elf Cosmetics, Pure LeafVisit The Shelf → -
3
Audiencly
Niche Gaming & Esports Influencer AgencyA specialized agency focused exclusively on gaming and esports creators on YouTube, Twitch, and TikTok. Ideal if your campaign is 100% gaming-focused — from game launches to hardware and esports events.Clients: Epic Games, NordVPN, Ubisoft, Wargaming, Tencent GamesVisit Audiencly → -
4
Viral Nation
Global Influencer Marketing & Talent AgencyA dual talent management and marketing agency with proprietary brand safety tools and a global creator network spanning nano-influencers to celebrities across all major platforms.Clients: Meta, Activision Blizzard, Energizer, Aston Martin, WalmartVisit Viral Nation → -
5
The Influencer Marketing Factory
TikTok, Instagram & YouTube CampaignsA full-service agency with strong TikTok expertise, offering end-to-end campaign management from influencer discovery through performance reporting with a focus on platform-native content.Clients: Google, Snapchat, Universal Music, Bumble, YelpVisit TIMF → -
6
NeoReach
Enterprise Analytics & Influencer CampaignsAn enterprise-focused agency combining managed campaigns with a powerful self-service data platform for influencer search, audience analytics, and attribution modeling.Clients: Amazon, Airbnb, Netflix, Honda, The New York TimesVisit NeoReach → -
7
Ubiquitous
Creator-First Marketing PlatformA tech-driven platform combining self-service tools with managed campaign options, emphasizing speed and scalability for brands managing multiple influencer relationships.Clients: Lyft, Disney, Target, American Eagle, NetflixVisit Ubiquitous → -
8
Obviously
Scalable Enterprise Influencer CampaignsA tech-enabled agency built for high-volume campaigns, coordinating hundreds of creators simultaneously with end-to-end logistics, content rights management, and product seeding.Clients: Google, Ulta Beauty, Converse, AmazonVisit Obviously →