Introducing aiRA by Capillary
On-demand insights and analytics agent for eCommerce, growth, marketing, CRM, loyalty, and data and BI teams.
Ask questions about your customer and marketing data in plain language. Get real-time insights, the reasoning behind them, and a recommended move built and waiting on your approval.
Customer brands live with aiRA
aiRA works where 100+ enterprise brands already trust Capillary
Solutions
Every industry asks its own questions.
A grocer's week looks nothing like a carrier's or a hotel's. aiRA knows each industry down to its categories, and the questions that come with them.
- Find the members who installed the app and have never paid with it.
- What is a fuel-only customer worth against one who also shops inside?
- Find the online customers who have never been matched to a store visit.
- Which customers have dropped a department they used to buy every quarter?
- Which first orders are likely to be returned in full before we ship them?
- Is the base moving out of formalwear into casual and active, and how fast?
- Which delivery slots are filling with baskets too small to pay for the van?
- Of the shoppers we won back last quarter, what did they actually respond to?
- Reconcile front-of-store sales against the loyalty ledger for last week.
- Which repeat purchases move online and which stay in store, by category?
- Find the buyers who furnished one room and never came back for the next.
- Which past buyers are most likely to be starting a second room this year?
- Where does repeat purchase stall first: price, range or delivery?
- What counts as an active customer under our current programme rules?
- Which online-first buyers later bought in store, and what brought them in?
- Separate the enthusiasts from the gift buyers in last season's customer file.
- How did last month's limited-time offer read by daypart, region and location?
- Which menu items actually travelled together at lunch, and at what margin?
- Which regulars have broken a weekly habit this month, not just missed a week?
- Separate the habitual customers from the occasional ones, and keep it for next time.
- Which members are one stay or one segment short of status while the qualification window is still open?
- Of last year's shoulder-season bookings, which came from campaigns and which were coming anyway?
- Which dormant members are actually churning, and which are simply between trips?
- What did the members we re-engaged last year book first when they came back?
- Did last quarter's upgrade campaign move ARPU, or just move spend between plans?
- Split that by prepaid and postpaid, and show me where the lift actually came from.
- Which lines look like churn under our own lapsed rule, not a generic inactivity window?
- Which usage drops in the last 30 days are real risk, and which are seasonal?
How it fits
aiRA sits between your
data and your channels.
You already own the warehouse. You already own the channels. What is missing is the judgement in between, and today that runs on analyst tickets and Monday meetings.
System of record
Reads where your data already sits
The intelligence layer
System of action
Executes anywhere, in any channel
-
Zero copy
aiRA reads your data in place. No migration, no rebuild.
-
Nothing to rip out
It plugs into what you already own and pushes to what you already run.
-
A person approves
Every move waits on your approval before it goes live.
Ask, reason, approve
Every answer arrives with its reasoning and the move it earns.
Most of the shortlist stops at the answer. The same question carries through to a built move, and the move waits for a person.
Which customers have dropped a department they used to buy every quarter?
Who lapsed, the quarter the habit broke, and what they buy now instead.
Submit “Cross-department win-back” for approval
Customer-facingThis messages real customers. Submitting sends it for sign-off, and nothing goes live until it is approved.
Every answer shows its working
The reasoning trace comes with the answer, not as a separate audit you have to ask for.
The answer arrives as a built move
Audience resolved, budget and window set, before anyone briefs another team.
A person decides, every time
Nothing reaches a customer until someone approves it, on an audit trail.
Customer story
NASCAR gets fan insights in minutes, not weeks.
NASCAR's marketing and CRM teams use aiRA, which already understands their fans, their data and the rhythm of a race season. Analysis that used to take weeks now takes minutes, so the team spends less time explaining and more time acting.
Against the category
Where aiRA sits against the category.
Pricing
No per-seat pricing. Pay based on usage.
A credit is consumed when aiRA runs an analysis or builds an action. Nobody pays for a login they never use, and adding the tenth marketer costs nothing.
Credits, charged on work done
A credit is consumed when aiRA runs an analysis or builds an action. No per-seat license.
2 to 4 weeks to a pilot
On your real customer data, with your business context loaded. No engineering headcount from your side.
An immutable credit ledger
Every credit aiRA spends is written to a ledger that cannot be edited, so usage is auditable line by line.
Security
Your data stays where it is, and nothing writes without you.
Security, privacy and governance are the first questions on any evaluation. These are the answers, and the full brief arrives before the session rather than after the third call.
Your data stays in your region
aiRA reads it where it already sits, through Anthropic Claude on AWS Bedrock inside your own region. Zero retention, stateless, and Capillary never trains on your data.
Masked before the model sees it
PII is masked ahead of every tool call, and queries run sandboxed and read-only with no outbound network.
Nothing writes without a person
Every customer-impacting change stops at an approval gate, on an audit trail.
SOC 2 Type II · ISO 27001 · PCI DSS · GDPR, CCPA and PDPA. HIPAA-capable under a BAA.
FAQ
Frequently asked, plainly answered.
The questions that come up before every evaluation: what aiRA is, where your data goes, what it costs, and what a pilot actually runs on.
Yours is more specific? Bring it to a working session.
aiRA is an on-demand insights and analytics agent. You ask in plain language, it reads your warehouse, comes back with reasoning, and hands back insights or the segment of customers you wanted or the complete marketing journey with hyper-personalised messaging, waiting for your approval.
No. aiRA reads it where it already sits, in Databricks, Snowflake, BigQuery or Redshift. No migration, no rebuild, and every query runs sandboxed and read-only.
No. Every customer-impacting move stops at an approval gate, on an audit trail. aiRA proposes, a person decides, and that holds for every playbook it runs.
Through Anthropic Claude on AWS Bedrock inside your own region, with PII masked ahead of every call. Zero retention, stateless, and Capillary never trains on your data. SOC 2 Type II, ISO 27001, PCI DSS, GDPR, CCPA and PDPA.
Days, not months. Industry playbooks ship pre-built and your business context layers on rather than being modelled from scratch. A pilot on your real customer data takes two to four weeks.
Credits, charged on work done. A credit is consumed when aiRA runs an analysis or builds an action, every credit is written to an immutable ledger, and there is no per-seat license.
Retail, restaurants and QSR, travel and hospitality, telecom, fuel and convenience, and airlines, on top of the industries Capillary already runs loyalty for. Each one arrives with the vocabulary and the playbooks of that industry rather than a blank prompt box, and retail goes a level deeper again into its categories. If yours is not on the list, the setup does not change: aiRA reads your warehouse in week one and learns your own business context from there.
On your real customer data, with your business context loaded. No engineering headcount from your side.
The question your team is still waiting on.
Thirty minutes, one real question, answered live. We reply within one business day with a time and the security brief.
Book a working session
The working session
You have seen how it works. Bring us the question it has not answered yet.
Thirty minutes, one real question, no deck and no discovery calls stacked in front of it. You leave with the answer, the reasoning behind it and the recommended next move, not a product demo.
We reply within one business day.
With a time and the security brief. No demo gauntlet.
Security brief comes first.
Shared before the session, not after the third call.
We will tell you where aiRA is the wrong fit.
It happens. Bring your actual shortlist and we will go through it.