MiaProva’s New Adobe Intelligence Services: A First Look at Chat-Powered Target Optimization
Adobe Target teams are sitting on an incredible amount of signal – activities, offers, audiences, A4T data, and years of experimentation history. The hard part isn’t having data; it’s knowing what to do next.
In a new early-peek demo, we walked through the Adobe Intelligence Services we’re building inside MiaProva to make that “what next?” question dramatically easier to answer. These services connect Adobe Target + Adobe Analytics + Gemini into a conversational experience that feels like chatting with an expert who already knows your program.
This post recaps what we showed in the video and how we see teams using it.
Turning on AI Across MiaProva
Once an organization enables our Gemini integration, it doesn’t just power one feature—it unlocks AI across the MiaProva platform:
- Insights & reporting that can read test results, explain them in plain language, and suggest next moves
- Idea generation that’s grounded in your actual activities, audiences, and KPIs
- Recommendations that don’t rely on generic benchmarks, but on the reality of how you’re using Adobe Target
Crucially, organizations can control their agents:
- Customize prompts so the agent speaks in your brand’s language (B2B SaaS vs. retail vs. nonprofit, etc.)
- Tune models for different use cases (e.g., more creative ideation vs. more analytical deep dives)
Instead of a one-size-fits-all assistant, each tenant gets an agent that’s tuned to their business and experimentation culture.
A MCP-Style Experience for Adobe Target
At the heart of the demo is a Model Context Protocol (MCP)–style experience for Adobe Target. Think of it as:
“What if a knowledgeable practitioner could sit on top of your Target APIs and your analytics, 24/7, and answer questions on demand?”
You start a new conversation, and the agent resets context for a clean session. A simple opener like:
“What things can you do?”
returns a guided overview of capabilities so users can explore without needing to memorize commands or API docs.
Behind the scenes, the agent is wired into Adobe Target APIs and, where available, Adobe Analytics / A4T so it can do much more than just describe concepts. It can act and analyze, not just chat.
From “Holiday Strategy” to Concrete Target Plays
In the demo, we asked:
“With the holiday season fast approaching, how do you recommend our organization makes use of Adobe Target?”
Instead of spitting out generic “best practices,” the agent:
- Pulls in context about your existing activities and usage
- Suggests specific ways to lean into Target for peak season, such as:
- Prioritizing certain activity types
- Testing specific parts of the journey
- Using offers and audiences more strategically
- Proposes next steps you can actually execute (not just high-level ideas)
It’s the difference between reading a blog post about holiday optimization and having a system that looks at your actual account and says, “Here’s what you’re currently doing—and here’s where you’re leaving money on the table.”
Query Your Adobe Target Account Like a Colleague
The agent isn’t limited to strategy. It can also help with everyday operational questions, like:
“List all approved or live activities.”
The tool:
- Pulls live, scheduled, and ended activities
- Returns IDs, names, states, and other useful details
- Asks clarifying questions when needed, based on the agent configuration
This is more than a thin layer over the APIs. We’ve augmented the experience with additional context so teams can quickly move from:
- Discover – “What’s currently running?”
- Inspect – “What’s the performance of this specific activity?”
- Decide – “Should we iterate, expand, or retire it?”
All without leaving the conversation.
Deep Dives Powered by Adobe Analytics & A4T
When you ask:
“How is the performance of this activity?”
the agent taps into Adobe Analytics / A4T (Analytics for Target) where configured. In the demo, we show it:
- Pulling in A4T event data (e.g., Event 7 as a key metric)
- Surfacing performance in a structured, interpretable way
- Providing narrative commentary on what’s working and what needs attention
We’re actively expanding this integration so account-level reporting and activity-level deep dives are not just available in dashboards, but conversationally available in MiaProva.
That means you don’t have to go hunting through multiple interfaces; you can ask, “What happened?” and get an answer that’s grounded in the right data.
Automated Account Audits & Strategic Recommendations
One of the most powerful flows in the demo is a full-account audit:
“Please examine my Adobe Target account and provide recommendations on ways I can use it more effectively, and provide a high-level audit of our company’s use of it.”
The agent then:
- Looks across:
- Activity mix (A/B tests, auto-allocate, AP, etc.)
- Offer usage
- Audiences and targeting strategies
- Monitoring and governance patterns
- Surfaces findings about how you’re currently using Target
- Suggests strategic improvements, such as:
- Where to expand testing
- Where to tighten governance or approval flows
- How to better leverage A4T and other integrations
In the demo, it correctly calls out a strong emphasis on monitoring activities—because that’s true of the underlying account. This is the kind of insight that traditionally takes a seasoned practitioner several hours in the UI and documentation; now it can be generated in a few seconds and used as a starting point for planning.
Built-In Training: “Help Me Set Up My First A/B Test”
We know many teams are constantly onboarding new practitioners. So we also showed how the agent can act as a training assistant:
“Help me set up my first A/B activity in Adobe Target.”
The agent responds with:
- A step-by-step walkthrough of creating an A/B test
- Explanations of key concepts (activities, experiences, audiences, success metrics, etc.)
- Practical guidance that reflects how organizations actually use Target in the real world
This turns MiaProva into more than a reporting and management layer. It becomes a coach that helps new team members ramp up faster while still aligning with your program’s standards and definitions.
Demystifying Profile Scripts for Non-Developers
Profile scripts in Adobe Target are incredibly powerful—but they can feel intimidating, particularly for those without a technical background.
In the video, we ask the agent to help with profile scripts. While Adobe Target doesn’t offer an API to create them directly, the agent can still help by:
- Asking clarifying questions about what you’re trying to track or derive
- Drafting profile script code tailored to that use case
- Providing instructions for how to implement and maintain the script in Target
So instead of staring at a blank profile script editor, teams can start from a working, explanation-rich template and adapt from there.
What’s Coming Next
What you see in the demo is just the beginning of our Adobe Intelligence Services roadmap. In active development right now:
- Deeper Adobe Analytics and CJA support inside the same conversational interface
- Expanded training and enablement modules, so teams can learn by doing, guided by the agent
- Additional intelligence components that connect:
- Governance
- Alerting
- Journey-level KPIs
- Cross-tool insights (Target, Analytics, GA4, etc.)
Our goal is simple:
Make MiaProva the orchestration layer where your tests, tools, and AI agents all work together to continuously improve customer experience and business outcomes.
Tell Us What You’d Like This Agent to Do Next
This is an early peek, and we’re building it with practitioners, not in a vacuum.
If you’re using Adobe Target, Adobe Analytics, or CJA, we’d love to hear:
- What questions you wish you could ask your accounts today
- What audits, reports, or workflows you’d love to automate
- What training scenarios would make your team’s life easier
You can watch the full demo here:
“An early peek at our Adobe Intelligence Services we are working on”
https://www.youtube.com/watch?v=gPrYI9gR-E4
And if you’d like to see how these services could work with your own Adobe stack, reach out to the MiaProva team—we’re actively shaping this with customer input.
