{
  "slug": "support-agent-customer-memory",
  "title": "How an AI Customer Support Agent Should Remember Customers Across Conversations",
  "description": "A memory policy for an AI customer support agent: what may be stored, who writes it, how remembered text is read as data, how stale facts are handled, how customers are kept apart, and how memory is shown and corrected.",
  "kind": "sub",
  "order": 6,
  "target_query": "how should an ai customer support agent remember customers across conversations",
  "secondary_queries": [
    "ai agent memory"
  ],
  "tags": [
    "customer-support",
    "memory",
    "privacy",
    "prompt-injection",
    "identity",
    "agents",
    "production"
  ],
  "published": "2026-10-22",
  "updated": "2026-10-25",
  "words": 1629,
  "estimated_tokens": 2167,
  "premium": false,
  "rights": {
    "access": "free",
    "note": "Editorial guides are always free and never part of the licensed corpus.",
    "license": "https://changegamer.ai/license.xml",
    "pricing": "https://changegamer.ai/api/pricing.json",
    "payment": "https://changegamer.ai/api/payment.json"
  },
  "license": "https://changegamer.ai/license.xml",
  "citation": "ChangeGamer (2026-10-22). How an AI Customer Support Agent Should Remember Customers Across Conversations. ChangeGamer. https://changegamer.ai/articles/support-agent-customer-memory (updated 2026-10-25).",
  "bibtex": "@misc{changegamer_support_agent_customer_memory, title = {How an AI Customer Support Agent Should Remember Customers Across Conversations}, publisher = {ChangeGamer}, year = {2026}, url = {https://changegamer.ai/articles/support-agent-customer-memory}, note = {Updated 2026-10-25}}",
  "canonical": "https://changegamer.ai/articles/support-agent-customer-memory",
  "markdown": "https://changegamer.ai/articles/support-agent-customer-memory.md",
  "takeaways": [
    "An AI customer support agent should remember customers through an explicit memory policy covering what may be stored, who writes it, how it is read, when it expires and who can see it, not through a default of keeping everything.",
    "As inference, remembered customer text should be read by a support agent as data about the customer and never as instructions, because the agent-memory-context reference treats retrieved memory as an untrusted input surface.",
    "As inference, a support agent should prefer live system-of-record state over remembered facts, and treat a remembered fact as a hint carrying a verified date that must be re-checked before it changes an answer.",
    "As inference, every memory read and write in a support agent should be keyed by an authenticated customer ID rather than by name, email text or session guess, so a lookup can never return another customer's record.",
    "ChangeGamer runs no support agent as of October 2026, so this memory policy has no first-hand data behind it; only statements naming a reference are sourced, and the rest is labelled inference."
  ],
  "outline": [
    {
      "depth": 2,
      "text": "What does the corpus establish about support memory?",
      "anchor": "what-does-the-corpus-establish-about-support-memory",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#what-does-the-corpus-establish-about-support-memory"
    },
    {
      "depth": 2,
      "text": "What may a support agent store about a customer?",
      "anchor": "what-may-a-support-agent-store-about-a-customer",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#what-may-a-support-agent-store-about-a-customer"
    },
    {
      "depth": 3,
      "text": "Who or what writes to memory?",
      "anchor": "who-or-what-writes-to-memory",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#who-or-what-writes-to-memory"
    },
    {
      "depth": 2,
      "text": "How should remembered content be read at answer time?",
      "anchor": "how-should-remembered-content-be-read-at-answer-time",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#how-should-remembered-content-be-read-at-answer-time"
    },
    {
      "depth": 3,
      "text": "Provenance: customer-authored versus system-verified",
      "anchor": "provenance-customer-authored-versus-system-verified",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#provenance-customer-authored-versus-system-verified"
    },
    {
      "depth": 2,
      "text": "How should a support agent handle stale facts?",
      "anchor": "how-should-a-support-agent-handle-stale-facts",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#how-should-a-support-agent-handle-stale-facts"
    },
    {
      "depth": 2,
      "text": "How do you stop one customer's memory leaking to another?",
      "anchor": "how-do-you-stop-one-customer-s-memory-leaking-to-another",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#how-do-you-stop-one-customer-s-memory-leaking-to-another"
    },
    {
      "depth": 2,
      "text": "How long should memory live, and who decides?",
      "anchor": "how-long-should-memory-live-and-who-decides",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#how-long-should-memory-live-and-who-decides"
    },
    {
      "depth": 2,
      "text": "Should customers and humans see what the agent remembers?",
      "anchor": "should-customers-and-humans-see-what-the-agent-remembers",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#should-customers-and-humans-see-what-the-agent-remembers"
    },
    {
      "depth": 2,
      "text": "What does ChangeGamer itself store about a request?",
      "anchor": "what-does-changegamer-itself-store-about-a-request",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#what-does-changegamer-itself-store-about-a-request"
    },
    {
      "depth": 2,
      "text": "What does the corpus not cover?",
      "anchor": "what-does-the-corpus-not-cover",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#what-does-the-corpus-not-cover"
    },
    {
      "depth": 2,
      "text": "Where to read next",
      "anchor": "where-to-read-next",
      "url": "https://changegamer.ai/articles/support-agent-customer-memory#where-to-read-next"
    }
  ],
  "faq": [
    {
      "question": "How should an AI customer support agent remember customers across conversations?",
      "answer": "Through a written memory policy, not a default. State what may be stored, which component writes it, how remembered content is read as data rather than instructions, when facts are re-checked against the system of record, and how customers and humans can see and correct it."
    },
    {
      "question": "Does an AI support agent remember previous conversations by itself?",
      "answer": "An AI support agent does not remember by itself: the agent-memory-context reference says every model call starts fresh, so apparent memory is context your system deliberately reassembles and re-injects on each call. If nothing is stored and retrieved, the agent knows nothing of earlier conversations."
    },
    {
      "question": "Should a support agent store what the customer says verbatim?",
      "answer": "As inference, store short structured facts and summaries rather than raw customer text. Verbatim text can carry instructions that resurface later, and the reference notes that summarization is lossy, so keep the transcript in the helpdesk as the record and the memory as a pointer to it."
    },
    {
      "question": "How long should a support agent keep customer memory?",
      "answer": "The data-privacy-for-agents reference covers erasure and deletion duties in general terms and says to set explicit retention limits for memory stores, but it gives no period for customer memory. Choose one per memory type, give it a named owner, and take legal advice for your jurisdiction."
    }
  ],
  "body": "An AI customer support agent should remember a customer only through an explicit policy that fixes what is stored, who writes it, how it is read, when it is re-checked and who can see it. This is the memory sub-article of the [AI customer support agent in production](/articles/ai-customer-support-agents-in-production) cluster. The pillar already maps memory types to support and lists the hazards, so neither is repeated; this article is the policy you write before any store exists.\n\nReading key: a statement naming a reference is sourced to it, and anything else is the author's inference. As of October 2026 ChangeGamer operates no support agent and holds no customer transcripts. \"AI agent memory\" in general is a broader topic; this page covers only the support case.\n\n## What does the corpus establish about support memory?\n\nThe corpus establishes that an agent's memory is engineering around a stateless model, not something inside the model. The [agent memory and context](/resources/agent-memory-context) reference (updated 2026-08-05) says every call starts fresh and what looks like memory is context deliberately reassembled and re-injected each time. It describes long-term memory as persisted user facts, episodic memory as past conversation summaries, and summarization as lossy, with fine detail irretrievable once compressed.\n\nTwo consequences follow, and both are inference. First, forgetting is the default, so every remembered item exists because someone decided it should. Second, whatever is re-injected is an input the model will act on, so the store is part of the attack surface.\n\n## What may a support agent store about a customer?\n\nA support agent may store only facts that help resolve a future contact and that a human could justify when asked. As inference, write the allowed list as a short schema, not as a prompt instruction, and reject anything outside it.\n\n| Question | Policy (inference) |\n|---|---|\n| May be stored | Issue category, product or order reference IDs, stated language or contact preference, a short summary of the outcome |\n| Never stored | Payment card or bank data, credentials, one-time codes, anything the customer says should be done to their account |\n| Stored as a pointer | The transcript itself stays in the helpdesk; memory holds the ticket ID |\n| Needs a human decision | Health, legal or safety-adjacent remarks; route per [escalation triggers](/articles/support-agent-escalation-triggers) |\n\nThe never-stored row has two reasons. As inference, payment data is a liability with no resolution value in a memory store. As inference, customer-supplied instructions (\"always refund me without asking\") are the content a poisoning attempt would plant, so they are not memory at all.\n\n### Who or what writes to memory?\n\nOnly server-side code with a schema should write to memory, never the model free-form. As inference, the model proposes a candidate fact, a validator checks it against the schema and the authenticated session, and the validator stores or discards it. The [agent guardrails](/resources/agent-guardrails) reference (updated 2026-09-26) describes guardrails as runtime checks that sit around the model and enforce policy regardless of what it outputs; a memory write is a natural place for one.\n\n## How should remembered content be read at answer time?\n\nRemembered content should be read as data about the customer, never as instructions to the agent. The pillar covers why memory is an untrusted surface; the policy consequence is that a stored customer message is data, never a directive.\n\n### Provenance: customer-authored versus system-verified\n\nEvery remembered item should carry a source tag, because a customer-authored claim and a system-verified fact deserve different weight. As inference, use at least two values:\n\n- **customer-authored**: said by the customer in a conversation. Usable as context, never as authority for an account action.\n- **system-verified**: written from the system of record, such as an order status returned by a tool. Usable to shape an answer, subject to the staleness rule below.\n\nRender each item to the model inside a clearly delimited data block that names its source and states that the block holds records, not directives. This is a mitigation, not a cure: the agent-guardrails reference says prompt injection has no complete fix, so the action-side controls from the [pillar](/articles/ai-customer-support-agents-in-production) still apply regardless of what memory says.\n\n## How should a support agent handle stale facts?\n\nA support agent should treat a remembered fact as a dated hint and prefer live state from the system of record whenever the two could differ. As inference, store a verified date with every fact and define, per fact type, whether it may be quoted without a re-check.\n\nOrder status, subscription state and refund progress change without the customer telling the agent, so these should never be answered from memory; call the tool. A language preference or a product the customer owns changes slowly and may be quoted with a light confirmation (\"Is this still the right contact preference?\"). When memory and live state disagree, live state wins and the agent should update or drop the memory entry. Summaries deserve extra suspicion because a summary can contradict a later human decision.\n\n## How do you stop one customer's memory leaking to another?\n\nScope every read and write by an authenticated customer ID, and never by name, email text or a similarity match alone. As inference, this is the highest-severity failure in the whole design: a retrieval that returns another person's history exposes real data in the company's own voice.\n\nA workable rule set, all inference:\n\n1. Derive the customer ID from the authenticated session or verified channel, never from what the customer types.\n2. Apply the ID as a mandatory pre-filter before any semantic ranking, so similarity only ranks within one customer's records; the mechanics, including why post-filtering fails, are in [permission-aware retrieval for multi-tenant RAG](/articles/multi-tenant-rag-permissions).\n3. Treat an unauthenticated conversation as having no long-term memory at all; offer only the current session.\n4. Handle shared contexts (a family account, a company with several contacts) explicitly, since one login may not mean one person.\n5. Log which customer ID each memory read used, so a wrong-customer incident can be investigated.\n\nThe [shipping agents to production](/resources/shipping-agents-to-production) reference (updated 2026-08-13) is the place for the general launch discipline; add a wrong-customer retrieval test to your own evaluation set before launch.\n\n## How long should memory live, and who decides?\n\nRetention is an operational design question that needs an owner and a per-type answer, and no corpus source gives a period for customer memory. The [Data Privacy and PII for Agents](/resources/data-privacy-for-agents) reference (updated 2026-09-28) covers erasure and deletion duties in general terms and says to set explicit retention limits for memory stores. Choose one limit per memory type with a named owner and take legal advice for your jurisdiction; no duration or legal claim is made here.\n\nAs inference, the questions to settle before launch are:\n\n- Which memory types exist, and what is the expiry rule for each (event-based, such as ticket closure, or time-based)?\n- What happens to memory when the customer closes their account or asks to be forgotten?\n- Does deleting a memory entry also remove derived summaries and embeddings?\n- Are backups and analytics copies in scope for the same deletion?\n- Who can approve an exception, and where is it recorded?\n\nWrite the answers as a one-page policy and test deletion end to end, not just the delete button.\n\n## Should customers and humans see what the agent remembers?\n\nYes: both the customer and the human agent should be able to see what is remembered and correct it, because unseen memory is unfixable memory. This is inference; the closest sourced parallel is the warm/cold handoff distinction in [customer support agents](/resources/customer-support-agents) (updated 2026-09-13), where a warm handoff passes transcript and structured state and a cold handoff drops the customer into a fresh queue with none of it.\n\nFor the human, surface the memory entries with their provenance and verified date in the console, beside the transcript, and, since the ticket is the system of record, carry the same entries as structured ticket fields, as [helpdesk integration](/articles/support-agent-helpdesk-integration) describes for gathered state. For the customer, let the agent state what it is relying on (\"I can see a ticket about a delayed order\") so a wrong entry is noticed in conversation, and give a path to correct or delete it. Record corrections as system-verified overrides rather than silent edits, and feed the wrong-memory cases into the review queue described in [quality measurement](/articles/support-agent-quality-measurement).\n\n## What does ChangeGamer itself store about a request?\n\nChangeGamer stores a fixed, small set of fields per request, and you can check it. As of 2026-10-25 the only analytics write in the site's worker is `logAccess` in `changegamer/worker/index.ts`, which records a category, the page slug, an outcome and the user-agent string cut to its first 100 characters, plus a count of one. That write carries no IP address and no request body. It is the allowed-list schema idea from the table above applied to traffic logs: the fields are fixed in code, so nothing outside them can be written. It says nothing about support agents, since ChangeGamer runs none. To verify it, search the worker for `writeDataPoint` and read the one call.\n\n## What does the corpus not cover?\n\nThe corpus gives no retention period for customer memory; the data-privacy reference covers erasure, deletion and retention limits only in general terms. It also does not cover consent wording, memory poisoning rates, or how any particular helpdesk or memory framework exposes per-customer scoping. It names frameworks and stores but does not rank them for support. Verify those in your own tooling and with your own legal advice before the policy above is implemented.\n\n## Where to read next\n\nStart with the pillar, [AI customer support agent in production](/articles/ai-customer-support-agents-in-production), for the whole plan. Then read [support agent escalation triggers](/articles/support-agent-escalation-triggers) for the topics memory must never decide alone.",
  "cluster": {
    "id": "support-agents-in-production",
    "title": "Support agents in production",
    "description": "How to run an AI customer support agent in production — escalation triggers and warm handoff, helpdesk integration, refund-policy enforcement, tone and brand voice, customer memory, CSAT and quality measurement, human review queues, multichannel deployment, wrong-answer incidents and replay testing — covering the support vertical only, not the horizontal guardrail and security mechanics owned by agent-guardrails-for-reliability and not the voice mechanics owned by voice-agents-in-production.",
    "status": "complete",
    "pillar": {
      "slug": "ai-customer-support-agents-in-production",
      "title": "AI Customer Support Agent in Production: Escalation, Policy, Tickets and Measurement",
      "description": "How to run an AI customer support agent in production: scope, escalation triggers, warm handoff, helpdesk integration, refund-policy enforcement, brand-voice guardrails, injection from customer text, memory, honest measurement and launch order.",
      "kind": "pillar",
      "order": 0,
      "html": "https://changegamer.ai/articles/ai-customer-support-agents-in-production",
      "markdown": "https://changegamer.ai/articles/ai-customer-support-agents-in-production.md",
      "json": "https://changegamer.ai/api/articles/ai-customer-support-agents-in-production.json"
    },
    "articles": [
      {
        "slug": "support-agent-escalation-triggers",
        "title": "Support Agent Escalation Triggers: When an AI Agent Should Hand Off to a Human",
        "description": "When an AI support agent should escalate to a human: seven triggers (five from the reference, two labelled inference), how to detect each, the false-positive and false-negative cost of each, which trigger wins when several fire, and what the human agent sees first.",
        "kind": "sub",
        "order": 1,
        "html": "https://changegamer.ai/articles/support-agent-escalation-triggers",
        "markdown": "https://changegamer.ai/articles/support-agent-escalation-triggers.md",
        "json": "https://changegamer.ai/api/articles/support-agent-escalation-triggers.json"
      },
      {
        "slug": "support-agent-tone-and-brand-voice",
        "title": "Support Agent Tone and Brand Voice: How to Keep an AI Customer Support Agent on Brand",
        "description": "How to keep an AI customer support agent on brand voice: what a tone specification contains, tone by situation (angry customer, refusal, policy denial), what an output check can and cannot enforce, persona-drift failure modes, and how to review tone.",
        "kind": "sub",
        "order": 2,
        "html": "https://changegamer.ai/articles/support-agent-tone-and-brand-voice",
        "markdown": "https://changegamer.ai/articles/support-agent-tone-and-brand-voice.md",
        "json": "https://changegamer.ai/api/articles/support-agent-tone-and-brand-voice.json"
      },
      {
        "slug": "support-agent-refund-policy-enforcement",
        "title": "Refund Policy Enforcement in an AI Support Agent: Keep the Decision Out of the Model",
        "description": "How to enforce a refund policy in an AI customer support agent: a deterministic policy check outside the model, why schema-valid arguments can still break policy, reading amounts from the order record, and test fixtures.",
        "kind": "sub",
        "order": 3,
        "html": "https://changegamer.ai/articles/support-agent-refund-policy-enforcement",
        "markdown": "https://changegamer.ai/articles/support-agent-refund-policy-enforcement.md",
        "json": "https://changegamer.ai/api/articles/support-agent-refund-policy-enforcement.json"
      },
      {
        "slug": "support-agent-helpdesk-integration",
        "title": "Integrating an AI Support Agent with a Helpdesk: Idempotent Tickets, Scoped Credentials, Narrow Tools",
        "description": "How to integrate an AI support agent with a helpdesk ticketing system: the ticket as system of record, write-back rules, idempotent create and update, least-privilege credentials, ticket ownership after handoff and a four-tool surface.",
        "kind": "sub",
        "order": 4,
        "html": "https://changegamer.ai/articles/support-agent-helpdesk-integration",
        "markdown": "https://changegamer.ai/articles/support-agent-helpdesk-integration.md",
        "json": "https://changegamer.ai/api/articles/support-agent-helpdesk-integration.json"
      },
      {
        "slug": "support-agent-quality-measurement",
        "title": "Measuring the Quality of an AI Customer Support Agent: Resolution Signals Beside Deflection",
        "description": "How to measure the quality of an AI customer support agent: explicit confirmation, a no-repeat-contact window and a post-resolution survey, each with its blind spot, read together with the escalation log and a verdict table.",
        "kind": "sub",
        "order": 5,
        "html": "https://changegamer.ai/articles/support-agent-quality-measurement",
        "markdown": "https://changegamer.ai/articles/support-agent-quality-measurement.md",
        "json": "https://changegamer.ai/api/articles/support-agent-quality-measurement.json"
      },
      {
        "slug": "support-agent-customer-memory",
        "title": "How an AI Customer Support Agent Should Remember Customers Across Conversations",
        "description": "A memory policy for an AI customer support agent: what may be stored, who writes it, how remembered text is read as data, how stale facts are handled, how customers are kept apart, and how memory is shown and corrected.",
        "kind": "sub",
        "order": 6,
        "html": "https://changegamer.ai/articles/support-agent-customer-memory",
        "markdown": "https://changegamer.ai/articles/support-agent-customer-memory.md",
        "json": "https://changegamer.ai/api/articles/support-agent-customer-memory.json"
      },
      {
        "slug": "support-agent-rules-based-bot-migration",
        "title": "How to Migrate From a Rules-Based Support Chatbot to an AI Support Agent",
        "description": "A migration sequence from a decision-tree support bot to a model-driven agent: inventory intents and escalation exits, shadow mode, per-intent cutover, the tree as fallback, parity measurement and rollback triggers.",
        "kind": "sub",
        "order": 7,
        "html": "https://changegamer.ai/articles/support-agent-rules-based-bot-migration",
        "markdown": "https://changegamer.ai/articles/support-agent-rules-based-bot-migration.md",
        "json": "https://changegamer.ai/api/articles/support-agent-rules-based-bot-migration.json"
      },
      {
        "slug": "support-agent-abusive-input-handling",
        "title": "How to Handle Abusive Customers With an AI Support Agent: Detect, De-escalate Once, Close or Escalate",
        "description": "How to handle abusive customers with an AI support agent: separate abuse of staff, off-topic input and ticket-text manipulation, then detect, de-escalate once, refuse briefly, close or escalate, and shield human reviewers.",
        "kind": "sub",
        "order": 8,
        "html": "https://changegamer.ai/articles/support-agent-abusive-input-handling",
        "markdown": "https://changegamer.ai/articles/support-agent-abusive-input-handling.md",
        "json": "https://changegamer.ai/api/articles/support-agent-abusive-input-handling.json"
      },
      {
        "slug": "support-agent-human-review-queues",
        "title": "How to Set Up a Human Review Queue for an AI Support Agent: Sample, Sign Off, Cap the Load",
        "description": "How to set up a human review queue for an AI support agent: what to sample, who reviews, a four-outcome decision set, a tie-break for disagreements, fixtures from corrections, and a cap on reviewer load.",
        "kind": "sub",
        "order": 9,
        "html": "https://changegamer.ai/articles/support-agent-human-review-queues",
        "markdown": "https://changegamer.ai/articles/support-agent-human-review-queues.md",
        "json": "https://changegamer.ai/api/articles/support-agent-human-review-queues.json"
      },
      {
        "slug": "support-agent-multichannel-deployment",
        "title": "How to Deploy an AI Support Agent Across Chat, Email and Voice: One Policy Core, Thin Channel Adapters",
        "description": "How to deploy an AI support agent across chat, email and voice: one shared policy core with a thin adapter per channel, what changes for email threads and voice, consistent escalation and memory, cross-channel identity, and channel order.",
        "kind": "sub",
        "order": 10,
        "html": "https://changegamer.ai/articles/support-agent-multichannel-deployment",
        "markdown": "https://changegamer.ai/articles/support-agent-multichannel-deployment.md",
        "json": "https://changegamer.ai/api/articles/support-agent-multichannel-deployment.json"
      },
      {
        "slug": "support-agent-wrong-answer-incidents",
        "title": "What to Do When an AI Support Agent Gives a Customer a Wrong Answer",
        "description": "What to do when an AI support agent gives a customer a wrong answer: classify the error, contain one intent, find every affected conversation in traces, correct the customer, weigh honouring or reversing the commitment, and add a fixture.",
        "kind": "sub",
        "order": 11,
        "html": "https://changegamer.ai/articles/support-agent-wrong-answer-incidents",
        "markdown": "https://changegamer.ai/articles/support-agent-wrong-answer-incidents.md",
        "json": "https://changegamer.ai/api/articles/support-agent-wrong-answer-incidents.json"
      },
      {
        "slug": "support-agent-replay-testing",
        "title": "How to Test an AI Support Agent With Historical Tickets: Build a Replay Set",
        "description": "How to test an AI support agent with historical tickets: pick resolved tickets per request type including escalations and refusals, redact them, assert on behaviour not wording, grade deterministically first, bridge to live traffic with shadow mode, and gate releases.",
        "kind": "sub",
        "order": 12,
        "html": "https://changegamer.ai/articles/support-agent-replay-testing",
        "markdown": "https://changegamer.ai/articles/support-agent-replay-testing.md",
        "json": "https://changegamer.ai/api/articles/support-agent-replay-testing.json"
      }
    ]
  },
  "navigation": {
    "pillar": {
      "slug": "ai-customer-support-agents-in-production",
      "title": "AI Customer Support Agent in Production: Escalation, Policy, Tickets and Measurement",
      "description": "How to run an AI customer support agent in production: scope, escalation triggers, warm handoff, helpdesk integration, refund-policy enforcement, brand-voice guardrails, injection from customer text, memory, honest measurement and launch order.",
      "kind": "pillar",
      "order": 0,
      "html": "https://changegamer.ai/articles/ai-customer-support-agents-in-production",
      "markdown": "https://changegamer.ai/articles/ai-customer-support-agents-in-production.md",
      "json": "https://changegamer.ai/api/articles/ai-customer-support-agents-in-production.json"
    },
    "previous": {
      "slug": "support-agent-quality-measurement",
      "title": "Measuring the Quality of an AI Customer Support Agent: Resolution Signals Beside Deflection",
      "description": "How to measure the quality of an AI customer support agent: explicit confirmation, a no-repeat-contact window and a post-resolution survey, each with its blind spot, read together with the escalation log and a verdict table.",
      "kind": "sub",
      "order": 5,
      "html": "https://changegamer.ai/articles/support-agent-quality-measurement",
      "markdown": "https://changegamer.ai/articles/support-agent-quality-measurement.md",
      "json": "https://changegamer.ai/api/articles/support-agent-quality-measurement.json"
    },
    "next": {
      "slug": "support-agent-rules-based-bot-migration",
      "title": "How to Migrate From a Rules-Based Support Chatbot to an AI Support Agent",
      "description": "A migration sequence from a decision-tree support bot to a model-driven agent: inventory intents and escalation exits, shadow mode, per-intent cutover, the tree as fallback, parity measurement and rollback triggers.",
      "kind": "sub",
      "order": 7,
      "html": "https://changegamer.ai/articles/support-agent-rules-based-bot-migration",
      "markdown": "https://changegamer.ai/articles/support-agent-rules-based-bot-migration.md",
      "json": "https://changegamer.ai/api/articles/support-agent-rules-based-bot-migration.json"
    }
  },
  "resources": [
    {
      "slug": "agent-memory-context",
      "html": "https://changegamer.ai/resources/agent-memory-context",
      "markdown": "https://changegamer.ai/resources/agent-memory-context.md",
      "json": "https://changegamer.ai/api/resources/agent-memory-context.json"
    },
    {
      "slug": "customer-support-agents",
      "html": "https://changegamer.ai/resources/customer-support-agents",
      "markdown": "https://changegamer.ai/resources/customer-support-agents.md",
      "json": "https://changegamer.ai/api/resources/customer-support-agents.json"
    },
    {
      "slug": "shipping-agents-to-production",
      "html": "https://changegamer.ai/resources/shipping-agents-to-production",
      "markdown": "https://changegamer.ai/resources/shipping-agents-to-production.md",
      "json": "https://changegamer.ai/api/resources/shipping-agents-to-production.json"
    },
    {
      "slug": "agent-guardrails",
      "html": "https://changegamer.ai/resources/agent-guardrails",
      "markdown": "https://changegamer.ai/resources/agent-guardrails.md",
      "json": "https://changegamer.ai/api/resources/agent-guardrails.json"
    },
    {
      "slug": "data-privacy-for-agents",
      "html": "https://changegamer.ai/resources/data-privacy-for-agents",
      "markdown": "https://changegamer.ai/resources/data-privacy-for-agents.md",
      "json": "https://changegamer.ai/api/resources/data-privacy-for-agents.json"
    }
  ]
}