Free forever, no credit card.Get Started for Free →
← All posts
November 21, 2026 · 5 min read

Does Recurly Remember Between Runs? The Subscription Record Holds. The Run Reasoning Does Not.

Does Recurly Remember Between Runs? The Subscription Record Holds. The Run Reasoning Does Not. Ask whether Recurly remembers between runs and the honest answer splits in two. Recurly remembers the billing state almost perfectly. It does not remember what an automation or an AI agent was thinking while it worked on that state. Those are two different kinds of memory, and confusing them is where scheduled billing workflows quietly go wrong. What Recurly actually keeps Recurly is a subscription

Does Recurly Remember Between Runs? The Subscription Record Holds. The Run Reasoning Does Not.

Ask whether Recurly remembers between runs and the honest answer splits in two. Recurly remembers the billing state almost perfectly. It does not remember what an automation or an AI agent was thinking while it worked on that state. Those are two different kinds of memory, and confusing them is where scheduled billing workflows quietly go wrong.

What Recurly actually keeps

Recurly is a subscription management and recurring billing platform, and its data model is built to persist. At the center sits the Account, which holds the customer: account code, contact details, billing information, tax information, and account notes. Attached to the account are Subscriptions, each recording the plan, the state (active, paused, canceled, expired), currency, activation date, trial details, add-ons, and cancellation or expiration information. Recurly even keeps a subscription history export with every version created after any change, so the evolution of a subscription is on the record.

Around that core, the operational trail is just as durable. Invoices capture what was billed and whether it is pending, paid, or past due. Transactions record each charge attempt with its amount, status, payment method, and source. Dunning campaigns remember the recovery sequence: which retry and which email come next when a payment fails. Coupons, redemptions, credit payments, gift cards, and usage records all persist too, and all of it is reachable through the admin console, data exports, and the API.

So if a scheduled job asks a narrow question — what plan is this account on, did the last invoice clear, how many dunning attempts have run — Recurly answers from stored fact. Nothing about that layer forgets.

What the next run never sees

Now consider what a billing automation actually does between those stored facts. A nightly revenue-recovery workflow reads failed invoices, weighs which accounts deserve a personal outreach instead of the standard dunning email, checks whether an account was already granted an exception last month, drafts a message, and decides to hold one account because a support conversation is still open. A weekly churn-risk job compares usage against plan limits and flags accounts drifting toward downgrade.

When that run ends, where does its reasoning live? Not in Recurly. The account record shows the subscription state the run left behind. It does not show why the run skipped an account, which explanation it already sent, what promise was made in the last outreach, or which hypothesis it tested and discarded. Account notes exist, but they are a manual field a human types into, not a working memory an agent writes and reads on its own.

The next scheduled run therefore starts the way every run starts: by re-reading the current state and reconstructing the story from scratch. Sometimes it reconstructs wrong. It sends a second exception offer to an account that already used one. It re-opens a question the previous run already answered. It treats a deliberately paused recovery sequence as an oversight and restarts it. The billing data was never lost. The context around the billing data was never kept.

Why state is not memory

This gap is structural, not a Recurly flaw. Billing systems are systems of record: they store the current, authoritative version of commercial facts, plus an audit trail of changes. Agent memory is a different job. It stores the working context of the work itself: prior conversations, decisions and their reasons, open threads, and the running understanding that lets the next run continue instead of restart.

A subscription history export can tell a run that a plan changed on Tuesday. It cannot tell the run that the change was a negotiated save attempt, that the customer asked to be checked on in two weeks, or that the agent already explained the proration once and should not explain it again. Operators end up patching the gap with spreadsheets, prompt stuffing, and hand-written summary fields, and each patch lives inside one workflow, invisible to every other tool doing related work.

Giving the runs one shared memory

The fix is to put a memory layer beside the billing system rather than asking the billing system to be one. Vilix AI is built for exactly that role. It is cloud-hosted, so there is zero infrastructure to run: no database to provision, no vector store to maintain, no server to patch. An automation connects over MCP, the standard protocol AI tools use to reach external services, and the same memory is available everywhere that protocol reaches — the n8n or Make workflow running the recovery sequence, the chat assistant a support lead asks questions in, and the agent that drafts the outreach.

What gets stored matters as much as where. Vilix AI keeps full conversation history, not just extracted facts, so a run can recover the actual exchange where a customer explained their situation, not only a distilled note that they were unhappy. When the next run starts, it retrieves the relevant part of that history and begins briefed: this account already received an exception, this thread is waiting on support, this explanation was already given. Runs stop repeating each other because they finally share one agent state instead of each rebuilding a private guess from billing rows.

Control stays with the operator. Memory can be exported or deleted at any time, so the record of the work is never locked inside a tool. The free plan is free forever, which makes it practical to connect one workflow and watch whether the re-briefing disappears. For heavier use there is a 7-day Pro trial with no card required.

The short answer

Does Recurly remember between runs? The subscription record does: accounts, plans, invoices, transactions, and dunning steps persist in detail, down to every version of a changed subscription. The reasoning of the runs working on that record does not, and the next run pays for that gap in repeated offers, reopened questions, and restarted sequences. Keep Recurly as the system of record. Give the runs one shared memory of their own, and each run starts where the last one stopped.

Get Started for Free

Persistent memory across ChatGPT, Claude, and the AI tools you already use in Vilix AI.

Get Started for Free

Free forever, no credit card.

Keep reading
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 Lemon Squeezy Remember Between Runs? The Customer Record Holds. The Run Reasoning Does Not.

Does Lemon Squeezy Remember Between Runs? The Customer Record Holds. The Run Reasoning Does Not. Ask whether Lemon Squeezy remembers between runs and the answer splits cleanly in two. Lemon Squeezy remembers the commercial state of a customer relationship in detail, because that is its job as a merchant of record for digital products. 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 w

Does FastSpring Remember Between Runs? The Account Record Holds. The Run Reasoning Does Not.

Does FastSpring Remember Between Runs? The Account Record Holds. The Run Reasoning Does Not. Ask whether FastSpring remembers between runs and the answer splits cleanly in two. FastSpring remembers the commercial state of a customer relationship in detail, because that is its job as a merchant of record. 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 workflow