Posted by: Jeroen Wijdeven
11 November 2025
Why agentic AI now?
Agentic AI moves beyond single-task assistants. It coordinates a set of specialised agents that plan, adapt and act toward an outcome, which is why it fits marketing and experimentation where work spans multiple systems. IBM describes agentic AI as systems that can decide and act with minimal supervision, orchestrating multi-step tasks across tools.
Optimizely Opal brings this paradigm into Optimizely One. It provides agents and workflow orchestration inside CMS, CMP, Personalisation, Experimentation and Data Platform so teams can research, create, test and publish inside one environment. (Source: Optimizely Support) (Source: Optimizely Support)
What is Optimizely Opal?
Opal is Optimizely’s agent orchestration layer. You can deploy default agents from the Agent Directory, build specialised agents for your workflows or compose multi-step workflow agents to connect tasks across products. (Source: Optimizely Support) (Source: Optimizely Support) (Source: Optimizely Support)
Opal is natively embedded across Optimizely One and draws on your brand assets, prior campaign data and experiments to improve output quality. (Source: Optimizely)
Benchmarks you can plan against
Optimizely’s 2025 Opal AI Benchmark Report analysed 47,000 Opal interactions across nearly 900 adopters and found: +78.7% more experiments run, +24.1% more personalisation campaigns, +9.3% higher win rates and -53.7% faster campaign completion among Opal users. (Source: Optimizely Report) (Source: Optimizely Press)
AI agents versus agentic AI
| Type | Description | Example |
|---|---|---|
| Simple Assistants | Execute single prompts. | “Summarise this article.” |
| Specialised Agents | Use domain context and tools to deliver a repeatable task. | “Optimise metadata for SEO.” |
| Workflow Agents | Orchestrate multi-step processes across systems. | “Build, review and publish a campaign with approvals.” |
Agentic AI means orchestration, not just automation. It is the difference between a copy helper and an outcome-driven digital workforce. (Source: IBM Think)
Benefits for CMS and experimentation
- Faster content throughput - Opal agents create briefs, drafts and translations in CMP, then hand off to CMS for publishing, cutting re-entry and delays. (Source: Optimizely Support)
- Higher experiment velocity - Variation development and experiment review agents reduce build time and help teams reach significance sooner. (Source: Optimizely Support)
- Measurable lift - Real-world adopters reported +78.7% experiment volume and -53.7% campaign time using Opal workflows. (Source: Optimizely Report)
How Opal credits work
Opal uses credits whenever a feature makes an AI model call. Credits are pooled across eligible Optimizely products, so activity in CMS, CMP or Experimentation draws from one allocation. From 1 October 2025 to 30 September 2026, each Opal instance receives a complimentary 200 credits monthly. If usage regularly exceeds that level, Optimizely may propose a tailored plan. (Source: Optimizely Support)
Cost and compliance considerations
- Budget by workflow - estimate credits for high-frequency actions like content generation, translation and experiment analysis, then apply guardrails to avoid runaway usage. Optimizely exposes usage in Admin with notifications. (Source: Optimizely Support)
- Data protection - Gemini for Google Cloud states prompts and responses are not used to train models, with enterprise controls for data isolation and compliance. (Source: Google Cloud) (Source: Google Cloud)
- Platform compliance - Optimizely lists certifications including ISO 27001, ISO 27017, ISO 27018 and SOC 2 Type 2 on its Trust Center. (Source: Optimizely Trust Center)
Security, governance and trust
Enterprise AI must balance speed with oversight. Opal runs within Optimizely’s governed environment with service status transparency and scheduled maintenance windows. (Source: Optimizely Status) (Source: Optimizely Support)
Optimizely documents security practices and a 99.9% SLA at the application level for DXP, while Google provides enterprise data controls for Gemini. Use role-based access, audit logs and approval workflows to keep AI activity traceable. (Source: Optimizely Docs) (Source: Google Cloud)
Growcreate operates under ISO 27001 and Cyber Essentials controls across projects, with ISO 27001 certification announced publicly. Cyber Essentials is a UK government-backed scheme for baseline cyber hygiene. (Source: Growcreate) (Source: GOV.UK)
What Opal can do today
| Category | Example Agents | Primary Benefit |
|---|---|---|
| Content Creation | Keyword Researcher, Campaign Brief Generator, Press Release Agent | Faster briefs and on-brand drafts |
| Experimentation | Variation Development Agent, Experiment Review Agent | Shorter cycles and better test quality |
| Personalisation | Personalisation Advisor, Market Researcher | Clearer segments and targeted messaging |
| Compliance & QA | Web Accessibility Evaluator, Brand Tone Checker | Consistency and governance |
| Automation | Workflow Builder, Scheduler, Integration Manager | No-code cross-platform orchestration |
Browse the latest agents in the directory. (Source: Optimizely Support)
Opal is available across CMS, CMP, Feature Experimentation and selected Data Platform features. (Source: Optimizely Support) (Source: Optimizely Support)
How Opal compares to other orchestration platforms
| Platform | Focus | Orchestration & Interop | Data & Compliance Notes |
|---|---|---|---|
| Optimizely Opal | Agent orchestration for marketing across CMS, CMP, Experimentation and ODP | Embedded in Optimizely One, workflow agents, Agent-to-Agent interoperability with Gemini Enterprise | Optimizely Trust Center certifications, Gemini enterprise data protection |
| AWS Bedrock Agents | Build and run agents on AWS | Custom orchestration and multi-agent collaboration GA with monitoring and templates | Enterprise controls via AWS; architecture varies by deployment |
| Microsoft Copilot Studio | Build custom agents across M365 and Teams | New “computer use” enables UI automation for web and desktop apps | Runs under Microsoft 365 governance and admin controls |
Citations: Opal overview and Gemini Enterprise availability (Source: Optimizely) (Source: Optimizely Press); AWS custom orchestration, multi-agent collaboration (Source: AWS) (Source: AWS); Copilot Studio computer use (Source: The Verge).
Note: Industry analysts warn of “agent washing” and early-stage failure rates, so governance and pilot design matter. (Source: Reuters)
Growcreate’s Support. Enhance. Evolve. framework for Opal
| Phase | Focus | Outcomes |
|---|---|---|
| Support | Configure, secure and stabilise Opal across CMS, CMP and Experimentation | 99.99% uptime, zero critical errors, 6 h → 30 min content cycle time |
| Enhance | Optimise workflows and analytics for measurable ROI | +43% faster publishing, -40% MTTR, +22% campaign velocity |
| Evolve | Build adaptive, agentic frameworks that scale | 3× faster experimentation, +18% conversion, 100% governance compliance |
Explore our Optimizely services
Integrations you can use from day one
Opal sits inside your Optimizely estate and can connect with Azure, Salesforce and Dynamics to keep workflows in sync. (Source: Optimizely + Microsoft) (Source: Optimizely + Salesforce) (Source: Optimizely Support - Salesforce CMS 12) (Source: Optimizely Support - Dynamics)
How we implement Opal
- Discovery and Readiness Audit - review data, security and workflow maturity.
- Opal Configuration - set up agents, credits and roles.
- Workflow Design - map content and experiment lifecycles into orchestrated processes.
- Pilot Launch - run controlled experiments with reporting.
- Enablement and Training - upskill teams on prompts, metrics and governance.
Every rollout includes checks aligned to ISO 27001, Cyber Essentials and GDPR so teams can innovate safely. (Source: GOV.UK)
Who leads Opal at Growcreate?
Jeroen Wijdeven – Optimizely MVP
As an Optimizely MVP and veteran Azure architect, Jeroen leads Growcreate's engineering and innovation practice as CTO.
"Agentic AI changes how teams deliver value. Our job is to make it secure, measurable, and simple to use."
His team combines expertise in Optimizely and Azure DevOps to build robust, compliant environments for Opal deployments.
Renée van Liempt – Opal Developer Accreditation
Renée was one of the first developers in Europe to achieve the Optimizely Opal Developer Accreditation, having gained extensive experience in .NET and Azure.
"Opal gives developers and marketers real power to scale," Renée explains. "It connects creativity with automation—letting teams design intelligent systems rather than just write code."
FAQs
Opal is the agent orchestration platform inside Optimizely One that automates content, marketing and experimentation workflows. (Source: Optimizely)
Yes. The 2025 Opal AI Benchmark Report found campaign completion time decreased by 53.7% among Opal users. (Source: Optimizely Report)
Credits are consumed when Opal features call AI models and are pooled across Optimizely products. From 1 October 2025 to 30 September 2026, instances receive 200 complimentary credits monthly. Track usage in Admin and set policies to control spend. (Source: Optimizely Support)
Optimizely runs on governed infrastructure with documented certifications. Google states Gemini for Google Cloud does not use your prompts or responses to train its models and provides enterprise controls for data isolation and compliance. (Source: Optimizely Trust Center) (Source: Google Cloud)
Optimizely reports rapid adoption across hundreds of customers and strong daily usage growth since mid-2025. (Source: PR Newswire)
