Tool profileAuxia
Agentic personalization with holdout groups built into the decisioning engine
Auxia builds global holdout groups into its decisioning engine rather than leaving measurement to the customer, which is unusual in this category. Pricing is decision-based and unpublished, and the product is shaped around enterprise deployment.
Visit AuxiaIs Auxia worth it?
- What it is
- Agentic personalization with holdout groups built into the decisioning engine
- Worth a look if
- You have the traffic volume for per-customer decisioning and want lift measured against an embedded global holdout by default.
- Skip it if
- Your site lacks the traffic density for model-based decisioning, or you want to trial a tool without a sales process.
- Cost to start
- Not published. Usage-based for Agent Studio, decision-based for Decisioning. No. None published.
Auxia at a glance
| Job category | Automation and Lifecycle |
|---|---|
| What it replaces | Segment-rule personalization with lift measured against a prior period |
| Pricing model | Paid only |
| Starting price | Not published. Usage-based for Agent Studio, decision-based for Decisioning. |
| Free tier | None published. |
| Getting started | Sales contact required |
| Built for | Enterprises with the volume for per-customer decisioning to beat segment rules |
| Company | Not published |
| Last verified | August 2026 |
Auxia pricing and plans
Auxia does not publish pricing. Not published. Usage-based for Agent Studio, decision-based for Decisioning. None published.
- No pricing is published. Agent Studio is described as usage-based and Decisioning as decision-based, with no rates attached.
Billing is described as usage-based and decision-based on volume, so cost tracks the number of customer decisions made. Model it against your active customer base rather than seats.
Auxia features: what it does
- Embedded global holdout groups mean lift is measured against a concurrent control by default rather than by customer discipline
- Per-customer real-time decisioning with an ML feature store and model library, rather than segment rules dressed as AI
- Multi-model experimentation allows competing approaches to be compared rather than assuming one wins
- Orchestrates across existing tools including Braze and Salesforce Marketing Cloud instead of demanding replacement
Auxia limitations
- No pricing published, and decision-based billing means cost scales with customer volume in ways that need modelling before commitment
- The agentic framing promises more autonomy than the described approval-routing workflow actually delivers
- Enterprise-shaped, with a demo-led process that rules out quick evaluation
- No public statement of the statistical method used to determine when a holdout result is significant
Who Auxia is for
| Your situation | What that means here |
|---|---|
| Enterprise with high customer volume | Per-customer decisioning needs density to beat segment rules, and that is the intended deployment. |
| Team burned by indefensible lift figures | Embedded global holdouts mean lift is measured against a concurrent control by default rather than by customer discipline. |
| Braze or Salesforce Marketing Cloud user | Positioned as orchestration on top of an existing platform rather than a replacement for it. |
| Smaller site | Model-based decisioning without traffic density produces confident output from thin evidence. |
| Buyer wanting to self-serve trial | Demo-led and enterprise-shaped, which rules out quick evaluation. |
| Team expecting agents to replace judgement | The agentic framing promises more autonomy than the described approval-routing workflow delivers. |
Auxia use cases for marketers
- Lifecycle orchestration on top of Braze or Salesforce Marketing Cloud, without replacing either.
- Personalization measurement, where the holdout group is part of the decisioning engine rather than something the team has to remember to set up.
Auxia FAQ
Why do embedded global holdouts matter?
Because the common alternative compares personalized visitors against a previous period, which confounds the effect with seasonality and every other change. A permanent holdout gives a concurrent control.
Is agentic here different from automation?
Less than the language implies. What is described is analysis, brief generation, content retrieval and approval routing through Slack, with humans approving the decisions.
Auxia sources and references
- Auxia describes an agentic marketing platform with an Agent Studio that analyses campaigns, builds briefs, pulls CMS content, routes approvals via Slack and orchestrates across tools including Braze and Salesforce Marketing Cloud, and a Decisioning Engine making real-time per-customer decisions with an ML feature store, model library, multi-model experimentation and embedded global holdout groups. Pricing is usage-based and decision-based and is not published. Named customers include Mercari, Atlassian, Comcast, The Guardian, Assurant, Konami, MUFG and NTT Docomo.auxia.io
- Auxia publishes product documentation at docs.auxia.io, including configuration guides covering how the platform is set up and a section on common configuration questions.docs.auxia.io
Compiled by Alston Antony from the sources above:how we compile these entries and ouraffiliate disclosure.