Explore. Build. Grow
Explore. Build. Grow

Our flexible three-stage framework to craft digital experiences.

Learn more
Expertise
Outcomes
View our portfolio
View our portfolio

Explore our range of digital solutions.

Learn more
Industries
Our Approach to AI
Our Approach to AI

Keeping ourselves at the forefront of emerging technology is part of our DNA. But we recognise that ‘technology for technology’s sake’ is not the end goal – there is a balance to be struck.

Learn more
DXP
Headless
Cloud Platforms
Marketing Optimisation
Contact Us
Contact Us
Explore. Build. Grow
Explore. Build. Grow

Our flexible three-stage framework to craft digital experiences.

Learn more
Expertise
Outcomes
View our portfolio
View our portfolio

Explore our range of digital solutions.

Learn more
Industries
Our Approach to AI
Our Approach to AI

Keeping ourselves at the forefront of emerging technology is part of our DNA. But we recognise that ‘technology for technology’s sake’ is not the end goal – there is a balance to be struck.

Learn more
DXP
Headless
Cloud Platforms
Marketing Optimisation
Contact Us
Picture of a human and robot hand trying to touch
Home / Insights / ＿

The best Optimizely Opal agents for marketers in 2026

We went through Optimizely's Agent Directory and ranked the four agents marketers should be installing right away.

Mark MacSmith
Mark MacSmith

18 August 2026

6 minute read

My colleague, Andy Thompson, recently wrote the honest version of the Optimizely Agent Platform pitch: what's good, what's fine print, and whether you're ready for it. Consider this the practical follow-up: which agents should you actually install?

Optimizely lists over 70 pre-built agents, and new ones keep arriving. That's a lot of clicking for someone who just wants to know where to start.

We picked four worth your first week with Opal and counted them down.

Campaign Brief Generation agent

The Campaign Brief Generation agent allows you to select a campaign type ('webinar', 'event', 'content campaign'), optionally an audience, an objective, and the assets you need. You get a standardised brief in seven sections, from objectives through to measurement.

The good bit: Deliverables are listed individually, so each can become its own task if you are also using Optimizely's Content Marketing Platform (CMP). It's the only agent here whose output slots straight into a workflow rather than needing someone to reformat it first.

The catch: This agent doesn’t use any tools. It's working from what you tell it, so it won't know your last product launch underperformed on registration unless you mention it.

Verdict: An amazing template engine, and a genuinely handy tool. Its structure is the value.

Competitive Insights agent

The Competitive Insights agent can run a competitive insights report with just one input: a competitor name or URL. You get a 30-day competitive intelligence report as a canvas: executive summary, recent headlines, marketing activity, analyst commentary, community signals, and next best actions grouped by priority, plus high-impact events from the past 60 days.

The good bit: It's one of a few agents running on Pro, which is Optimizely's maximum reasoning tier, and it shows in the output. Its research tool is date-aware, so it knows what 'the past 30 days' actually means.

The catch: Only one competitor per run, and a 30-day window. For share-of-voice across a competitive set, you may want to upgrade to the Conductor Competitive AI Share of Voice instead, which needs an active Conductor subscription and connector. 

Verdict: Great to use monthly, per key competitor. Just don't set it looping across 20 competitors and walk away. Credit budgets have a way of vanishing when nobody's watching.

GEO Recommendations agent

The GEO Recommendation agent only requires your brand's page URL, and then runs a generative engine optimisation audit. It returns a canvas covering crawler access, Core Web Vitals, schema markup, E-E-A-T signals and citation readiness, with a prioritised action plan.

The good bit: It produces one of the most complete single artefacts here, and returns hard measurements rather than a judgement. GEO is the question every client is asking right now, so reviewing the ranked opportunities list is something you could work on in partnership with your agency or digital team immediately.

The catch: The agent only delivers a report on one page per run, so a serious GEO program will still need expert guidance and oversight.

Verdict: The agent most likely to produce something you'd happily forward to your agency and/or digital team. What keeps it off top spot is that it hands you a great task list rather than doing the tasks itself.

Experiment Planning agent

With just two inputs: a URL and a test idea, the Experiment Planning agent is ready to go. You get a structured test plan: hypothesis, variant descriptions, targeting, a metrics table, statistical sizing guidance, risks and assumptions.

The good bit: Firstly, it actually looks at the page. Its tools include screenshotting alongside HTML browsing, so it reasons about what's actually on the URL rather than what your test idea implies.

Secondly, the statistical sizing assumptions are a seriously helpful first draft. Sample size maths is where good experimentation programs come unstuck. Putting the required sample size in the plan saves a test from burning time, traffic and a fortnight of everyone's patience on a result that was never going to hold up.

The catch: It plans nicely; it just doesn’t cross-check on its own. The agent doesn't natively look at your live tests or history; however, you can easily bridge this gap by linking it with its neighbouring agents, Experiment Backlog Prioritisation and Experiment Conflict Checker, inside an Opal workflow agent.

Verdict: Optimizely has spent close to two decades as an experimentation platform, and that history shows up in how well this agent reasons about a test. The quality of its output is second to none.

What the top three Opal agents have in common

Experimentation, GEO and competitive analysis are public disciplines. How to frame a hypothesis, how to size a sample, how schema markup and E-E-A-T signals work, how you structure a competitive read. All of it is codified and sitting in the training data. So when you hand Experiment Planning a URL and a rough idea, the expertise is already in the room. The agent isn't inventing a method. It's applying one the model knows.

These top agents feel disproportionately clever for how little you actually give them, because you're renting expert knowledge that was always there. That's where agents can truly deliver value.

Questions we get asked

How many agents are in the Optimizely Opal Agent Directory?

Optimizely's documentation currently lists 76 pre-built agents (at the time of writing) across five categories: Brainstorm and plan, Create, Experiment and analyse, Optimise, and Productivity. New agents keep arriving, so treat any figure as a floor.

Which Optimizely Opal agent should I install first?

If you run experiments at all, start with Experiment Planning. Then GEO Recommendations, which produces review-ready output on day one with no configuration.

Is there a GEO agent in Optimizely Opal?

Several. GEO Recommendations and GEO Auditor are both standalone page-level auditors with overlapping scope. GEO Schema Optimisation writes JSON-LD back into the CMS and needs Optimizely CMS (SaaS). E-E-A-T Checker handles the trust-signal slice on its own. Three further AEO agents come from Conductor and need an active Conductor subscription plus a configured connector.

What do the Opal inference levels mean for cost?

Agents run at a set level: Balanced, Complex or Pro, with Pro documented as maximum reasoning depth. Higher tiers reason harder and consume more credits per run.

Which Opal agents work with no setup?

Experiment Planning, GEO Recommendations and Competitive Insights produce useful output the moment you install them, because the expertise they draw on is public and already in the model. 

Who can install agents from the Agent Directory?

Opal administrators, agent builders, and Opal users with the 'Add, edit, and install specialised agents' attribute on a custom role. Your organisation also needs to be generative AI-enabled, and the agent must meet any hosting or configuration requirements on its details page.

Do pre-built agents update automatically?

No. The Agent Directory shows a badge when a new version is available, but you have to open the agent and click 'Update Agent' yourself. Worth putting on someone's monthly checklist.

Mark MacSmith
Mark MacSmith

18 August 2026

6 minute read

Enjoyed this article? Share it...

Want to tap into the expertise of a multi-award winning Optimizely Solution Partner?

Get in touch

Keep Reading

Want more? Here are some other blog posts you might be interested in.

Follow
Contact Us
1300 554 891
hi@luminary.com

Our locations

Stay informed
Our best articles, observations and news delivered to your inbox
Subscribe
Flags

We acknowledge the Traditional Custodians of the land on which we work, live and engage. We pay respect to Elders past and present.

Luminary is a registered trademark of Luminary Digital Pty Ltd

Terms of Services
Privacy Policy