Optimizely is steering marketing AI procurement toward task-specific models and agent governance, a shift that will shape RFPs for 2027 platform renewals. On September 1, 2026, the company introduced a family of post-trained marketing models, following its August 31 launch of role-based AI "Virtual Teammates." The announcements arrive as an Optimizely-commissioned survey of 100 UK marketing professionals found 63% lack a cookieless personalization plan and 54% have no defined first-party data strategy.
The RFP is changing: from model vendor to task fit
Optimizely's model announcement describes a portfolio approach that matches different workloads to different models, promising "frontier-level quality at a fraction of the cost." The practical takeaway for marketing operations teams is that evaluation criteria are shifting. The question is no longer which single large language model provider to standardize on. It becomes which specific marketing tasks are being automated, what acceptance testing will prove quality for each task, and how latency and cost map to the actual workflows driving budget.
For procurement, this reframes pricing conversations. A model tuned for a narrow workload lets teams ask for unit pricing tied to the task: cost per asset version generated, cost per audience build, or cost per experiment insight. That is a cleaner negotiation than enterprise-wide token budgeting when the heaviest usage often sits inside one marketing ops team and a handful of agencies.
Virtual teammates pull identity and permissions into martech evaluation
The Virtual Teammates announcement positions these agents as digital coworkers that "retain context" and work across marketing tools "within defined identity and permissions." That language matters because it moves AI into the same governance orbit as any automation that can move data and change production systems. The biggest risk is not output quality. It is what the agent is allowed to touch.
For enterprise teams, this means identity and access management suddenly belongs on the martech procurement checklist. Buyers need to ask where retained context lives, how long it is stored, and whether it can be exported for audit or eDiscovery in regulated environments. Marketing may own the budget, but IT security and identity teams now need a seat at the table. The operational promise of cross-tool agents is real, but so are the requirements: role-based access control, audit trails, and clear separation between sandbox and production workspaces.
The personalization demo is easy; the data plumbing is the work
At Shoptalk Spring 2026, Optimizely ran a booth activation where attendees chatted with an AI bot about scent preferences and received a personalized perfume mixed on-site by EveryHuman, an algorithmic perfumery. The front-end experience was compelling. But the same Optimizely survey reported by MarketingTech found 83% of marketers said their current personalization is based on assumptions rather than data-driven insights, and 74% worried their personalization technology would become obsolete.
Those numbers point to a familiar gap. A QR-driven chat can capture useful first-party data, but the operational lift is in consent, storage, identity linkage, and downstream activation. If a team cannot map where preference data lands, how it ties to an identity graph, and which channels can legally activate it, individualized personalization remains a stage trick that does not scale. Fragmented stacks compound the problem: 70% of surveyed marketers said they combine multiple technologies to achieve their current level of personalization, which tends to create slow approvals and brittle integrations.
Questions for the 2027 martech RFP
Optimizely's own research, published June 30, 2026, surveyed more than 2,000 marketing leaders across seven markets and found AI is increasing production volume but not delivering the time savings and creative breathing room marketers expected. The bottleneck is often review cycles, compliance steps, and channel execution - not content generation speed.
For teams planning platform renewals, several questions belong in the next RFP. For each proposed AI model or teammate, what specific marketing tasks is it designed to do, and what acceptance test will prove quality for that task rather than a generic demo? Where does retained context live, how long is it stored, and what controls exist for deletion, export, and audit by role and geography? What are the exact integration points for first-party and zero-party data capture, and how does that data map to identity resolution when third-party cookies are unavailable? If multiple tools are still required, who owns end-to-end uptime and troubleshooting when personalization fails in production?
Why this matters for marketing teams
The tooling is getting more specialized, and the agent layer is becoming more powerful. The limiting factor is increasingly the organization's ability to govern data and permissions across a fragmented stack. For teams with multiple brands, geographies, or regulated claims review, the cost of an AI mistake is rarely a bad paragraph. It is the wrong offer, the wrong audience, or a compliance escalation that burns weeks. The fastest path to individualized personalization will be a short, high-governance use case that proves data capture, consent, and handoff end to end - not a platform-wide AI rollout. For marketing managers building these capabilities, AI for Marketing Managers offers practical grounding in campaign optimization and analytics that aligns with the operational shift underway.
Your membership also unlocks:
