Product overview
The Agentic Marketing OS
In the agentic marketing era, AI will take on much of the day-to-day execution - while marketers focus on higher-order strategy and outcomes. This shift requires more than adding AI to existing workflows. It needs a new operating model for how marketing is run. Nexie was architected around that model from the ground up.
User interaction
Guardrails and learnings
Data
Layer by layer
What each layer actually does.
The same four layers, explained from the bottom of the stack upward.
Data
Data
Everything starts with what your store already produces. Core data covers customer profiles, your full product catalog, the raw event stream and the metrics that summarise it: orders, browsing, email opens, clicks and unsubscribes.
Integrations feed that data through the tools you already run, so there is no migration, no warehouse project and no new source of truth to maintain.
- Profiles with lifetime value, frequency and recency
- Catalog with variants, margin and stock position
- Raw events from browsing, clicks and opt-outs
- Metrics that turn raw events into signal
- Shopify, Klaviyo, loyalty and reviews feed the data continuously
Agents
Agents
Four agents work on that data. The Opportunities Agent finds the revenue moments worth acting on. The Personalization Agent decides who each moment belongs to and how the message should change per micro-segment.
The Email Generation Agent builds the send itself, and the Analytics and Learning Agent reads the result and turns it into a conclusion the others can reuse.
- Opportunities: what is worth sending, and when
- Personalization: the right version for each buyer
- Email generation: banner, copy, layout and products
- Analytics and learning: what worked, per micro-segment
Guardrails and learnings
Guardrails and learnings
Above the agents sits the control layer. Guardrails are set by your marketers: tone of voice, language, discount policy and the brand palette. Every agent composes inside them.
Learnings move the other way. Agents write them from real performance as segment learnings, catalog learnings, brand understanding and domain knowledge, but a learning only becomes a rule once a marketer approves it, so the account compounds without ever drifting off brand.
- Guardrails: tone of voice, language, discount policy, brand palette
- Learnings: segment, catalog, brand understanding and domain knowledge
- Learnings proposed by agents, approved by a human
- Approved learnings apply automatically to the next send
User interaction
User interaction
On top sits the layer your team actually touches. Marketers set the goals, review the drafts the agents propose, and sign off on policy before anything ships. The agents do the work; your team stays in charge of direction and approval.
Nothing leaves Nexie without a human review on the goals, the design and the policy that govern it, so the system compounds quickly without ever running away from you.
- Marketing Goals: what the agents optimise toward
- Design Approvals: drafts reviewed before they ship
- Policy reviews: guardrails kept current as the brand grows
Get started
See the whole stack
running on your store
Connect Shopify and Klaviyo, and Nexie comes back with your segments, the campaigns worth running next, and drafts waiting for your approval.