Meet Opal, Optimizely's AI Assistant

Opal is Optimizely's AI assistant, embedded directly into the products teams already use rather than bolted on as a separate tool. It sits inside the CMS, Content Marketing Platform, Experimentation and Personalisation, so assistance is available in the flow of work instead of in a separate window.

Because Opal understands the context it is invoked from — the page being edited, the campaign being planned, the experiment being analysed — its suggestions are grounded in your actual content and data rather than generic output. For Circle Health, that means an assistant that works against the real site structure, content types and brand guidelines.

Content creation. Draft, rewrite, summarise and translate copy against brand and tone-of-voice guidelines, with on-brand output as the default rather than something to be edited in afterwards.

Specialised agents. Purpose-built agents handle repeatable jobs end to end — building a page from a design or document, generating metadata, or producing variant copy — and can be configured to match your own workflows.

Experimentation support. Generate hypotheses, produce variant copy at volume, and interpret results, including what a given outcome means for the next test in the roadmap.

Personalisation. Identify meaningful audience segments and tailor messaging to each, so the right content reaches the right visitor at the right point in their journey.

CMS workflows. Create and populate content items, structure pages from existing blocks, generate SEO metadata and alt text, and reduce the manual overhead of routine content operations.

Scaling clinical and treatment content. Circle Health maintains a large library of treatment, hospital and specialist pages. Opal can draft and structure these from clinical source material, keeping terminology and tone consistent across hundreds of pages — with clinical review remaining a required step before publication.

Standardising consultant profiles. Agents can normalise specialist pages so every profile follows the same structure, length and tone, however the source information arrives.

Optimising the booking pathway. Opal can rapidly generate alternative headline, body and call-to-action copy for enquiry and booking journeys, giving experimentation programmes more variants to test and a faster route to measurable conversion gains.

Improving content operations. Metadata, alt text and media tagging can be generated at scale, improving both SEO performance and accessibility compliance — a material consideration for a healthcare provider.

Supporting personalisation. Content variations can be produced for different audiences, such as self-pay versus insured patients, or visitors arriving from condition-specific searches.

Opal accelerates production; it does not replace clinical sign-off. All patient-facing medical content should continue to pass through Circle Health's established review process before publication. Treat Opal output as a well-informed first draft that shortens the path to review, not as approved copy.

In practice this means using Opal to handle structure, consistency and volume, while subject-matter experts retain ownership of clinical accuracy, regulatory compliance and final approval.

1. Identify the repetitive work. Look for tasks that are high-volume and low-variation — metadata, profile pages, variant copy. These give the fastest, most visible return.

2. Start inside the product. Open Opal in the CMS and use it on a live, low-risk task to build familiarity before committing to larger workflows.

3. Use agents for repeatable jobs. Where a task recurs in a consistent shape, configure a specialised agent so the workflow is standardised rather than reinvented each time.

4. Pair it with experimentation. The clearest proof of value comes from measurable outcomes, so route Opal-generated variants through the testing programme and track the results.

5. Agree governance up front. Define which content requires clinical review and at which stage, so the team can move quickly without creating compliance risk.