Does Braintree Remember Between Runs? The Vault Holds. The Run Reasoning Does Not.
Does Braintree Remember Between Runs? The Vault Holds. The Run Reasoning Does Not. Ask whether Braintree remembers between runs and the answer splits cleanly in two. Braintree remembers the payment state of a customer relationship in detail, because that is its job as a payment platform. It does not remember what an automation or an AI agent was thinking while it worked on that customer. Those are different kinds of memory, and mixing them up is where scheduled billing workflows quietly start r
Does Braintree Remember Between Runs? The Vault Holds. The Run Reasoning Does Not.
Ask whether Braintree remembers between runs and the answer splits cleanly in two. Braintree remembers the payment state of a customer relationship in detail, because that is its job as a payment platform. It does not remember what an automation or an AI agent was thinking while it worked on that customer. Those are different kinds of memory, and mixing them up is where scheduled billing workflows quietly start repeating themselves.
What Braintree actually keeps
At the center of Braintree sits the Vault. The Vault securely stores customer information, including payment method information, so customers can be kept on file and charged again without re-entering their details. When payment information is stored, it is encrypted by the gateway and associated with a unique payment method token. Your own system holds the token, never the raw card number.
The token lifecycle matters for any recurring workflow. A single-use payment method expires three hours after it is created, or after it is used once. Vaulting consumes that single-use method, stores the underlying payment information, and creates a multi-use payment method instead. Multi-use payment methods never expire, and the stored ID can be used for unlimited future charges. That one design decision is why a subscription can renew months later without the customer ever returning to a checkout page.
Customers group the rest together. A customer record can hold personal information about anyone who holds a vaulted payment method or starts a transaction, and it has a one-to-many relationship with both payment methods and transactions. Every charge produces a transaction record that tracks the attempted movement of money, changes status as funds move toward settlement, and explains itself when a charge fails. A transaction can later be reversed before settlement or refunded afterwards, and that history stays attached to the record.
Subscriptions sit on top of the Vault. A subscription charges a customer automatically on a cycle, built from a plan, a price, a billing date, and a number of billing cycles, with add-ons, discounts, and trial periods where they apply. There is no limit on how many subscriptions can be associated with one payment method. For a scheduled workflow, this layer is dependable: a job never has to ask a customer for a card again, or guess whether a subscription is active, past due, expired, or cancelled. It queries the record and gets an answer that was written down when the event happened.
Where the record stops
The record is precise, and it is also narrower than it looks. Two documented behaviors show the edges clearly.
First, adding a new payment method to a customer's Vault record does not automatically update any of that customer's subscriptions. The subscription keeps funding from its existing payment method token until that token is deliberately updated, either in the Control Panel or through the API. The Vault can therefore know a newer card while a renewal still runs against the older one, simply because nobody told the subscription to switch.
Second, changing a plan does not rewrite the subscriptions already attached to it. When a plan price changes, only future subscriptions reflect the new price. Existing subscriptions keep their price until each one is edited directly. The platform stores each layer faithfully, but it does not reconcile the layers for you, and it does not explain to the next run why any of it was set up that way.
That gap is where automations lose the thread. Consider a scheduled dunning workflow that checks past due subscriptions every morning. Braintree tells it which subscriptions are past due, which transactions failed, and what status each one carries now. It does not tell it what the workflow already tried yesterday, which customers were already contacted, what a support conversation concluded, or why one account was promised a grace period while another was not. The run can read the payment record perfectly and still send a second card-update email to a customer who updated their card, or retry a charge that a human already agreed to pause.
The same gap appears when several tools share the work. A billing agent in one platform, a support agent in another, and a reporting job somewhere else can all read Braintree and still hold three different understandings of the same customer, because each one keeps its own notes, or keeps none at all. The payment record is shared. The reasoning never is.
What a shared memory layer adds
The fix is not to replace the Vault. The Vault does its job well. The fix is to give the runs themselves a memory that sits alongside the payment record, in a place every tool can reach.
That is the role Vilix AI is built for. Vilix AI is a cloud-hosted memory layer with zero infrastructure to run: there is no database to provision, no vector store to maintain, and no server to keep alive between runs. An automation connects over MCP, the standard protocol AI tools already use, and the same memory is available everywhere that connection exists, in n8n, Make, Zapier-driven workflows, coding tools, and chat assistants alike.
What gets stored is also different in kind. Vilix AI keeps full conversation history, not just extracted facts, so a run can recover not only what was decided but the discussion that produced the decision. When the morning dunning run starts, it can check what yesterday's run already attempted, which customers were contacted, and what was promised, before it sends anything. When a different tool picks the customer up later, it reads the same history instead of starting a second, conflicting one.
The practical rhythm is simple. A run begins by retrieving the relevant memory for the customer or task in front of it. It does its work against Braintree, reading the Vault, the subscription, and the transaction record as usual. Before it ends, it writes back what happened: what it checked, what it sent, what it changed, and what should happen next. The next run, in any tool, inherits all of it.
Access stays under your control throughout. Memory can be exported or deleted at any time, so the history your automations accumulate remains yours to move or remove. The free plan is free forever, which makes it practical to give one small workflow a memory first and expand from there, and the Pro trial runs for 7 days with no card required.
The short answer
Does Braintree remember between runs? The Vault remembers customers, payment methods, subscriptions, and transactions with real durability, and multi-use tokens are built to survive between charges indefinitely. What Braintree does not remember is your workflow's own work: the attempts, the contacts, the promises, and the reasoning. Give those runs one shared memory of their own, and the next run stops starting over. It starts where the last one left off, which is what remembering was supposed to mean in the first place.