What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
A “zero memories” result is a symptom, not a diagnosis. The vendor’s record may not have been saved or indexed, the search may be using the wrong user or tenant scope, a filter may exclude the record, or the agent may not have run retrieval at all. Trace the path in order: write → index → scope and filters → retrieval result → agent response. Without the memory backend, configuration, vendor identifier, and runtime trace, there is no reliable way to name the specific cause.
Start by checking whether the vendor memory exists
Look at the conversation, event, or record sent to the memory system—not just the agent’s final answer. Confirm that the submitted content includes the vendor’s exact name and, where relevant, aliases, domain, or stable identifier. If the platform extracts memories or metadata from conversations, check that its extraction strategy and metadata schema are configured. For example, Amazon Bedrock AgentCore Memory does not perform schema-based metadata extraction when a strategy has no metadata schema (AgentCore metadata documentation).
As an Amazon Associate I earn from qualifying purchases.
Then check whether the write or import was accepted and indexed. Review ingestion and indexing job status, validation errors, and any error sink. Salesforce recommends checking index completion and comparing source record counts with indexed records. Google Agent Retrieval documents that malformed imported records can be sent to an error sink while valid records continue, so a completed import does not necessarily mean every record was stored (Salesforce troubleshooting guide; Google Agent Retrieval import documentation).
Check scope and filters before changing retrieval
Match the write and read namespaces
Verify that the query uses the same user, client, tenant, or namespace as the write. A record stored under one scope will not appear in another. If the API resolves a namespace from a template, inspect the resolved value at runtime rather than assuming the placeholder expanded as intended.
#1 Best Overall
In Amazon Bedrock AgentCore Memory, the namespace separates primary entities, and retrieval configuration can narrow results by namespace, strategy ID, and metadata filters. Only declared indexed metadata keys are available for filtering. These are AgentCore-specific details, not universal rules for every memory product (AgentCore metadata documentation; AgentCore long-term memory documentation).
Temporarily relax metadata filters
Compare the production query with a broader query in the same namespace. Check that every filter field is indexed and that the stored value has the expected spelling and type. If filters are combined with AND logic, one mismatched condition is enough to hide an otherwise valid record. Salesforce specifically cautions that prefilters can prevent relevant content from being retrieved (Salesforce troubleshooting guide).
Use a broader query only to diagnose the issue; do not remove access controls or production filters as a permanent fix without confirming the intended data boundaries.
Test whether the query matches the vendor
Run the same search with the vendor’s literal name, common spelling variants, domain, or stable ID. If semantic or vector retrieval misses an exact name or code, compare it with keyword or hybrid retrieval on the same query and data. Salesforce recommends comparing hybrid and vector results in its Retriever Playground. Cloudflare Agent Memory describes recall as searching keyword indexes, topic keys, semantic vectors, and raw messages in parallel; that behavior is specific to Cloudflare’s product (Salesforce troubleshooting guide; Cloudflare Agent Memory documentation).
Inspect the returned records or chunks, not only whether the generated answer sounds right. A useful comparison is whether the exact vendor appears in a broad search, a filtered search, and each available retrieval mode.
Read the retrieval trace to distinguish no match from failure
Check the backend trace or per-source status. It should help answer three separate questions: did retrieval run, did the source return zero matching records, or did the source fail? These are different outcomes. Azure AI Search distinguishes a successful query with zero matches from a failed source; failures can involve malformed filters, configuration drift, permissions, throttling, timeouts, or temporary availability problems (Azure AI Search troubleshooting documentation).
Also verify whether the agent invoked or selected the retrieval action. Absence of a visible action in one planning trace is not always proof that retrieval did not happen: AWS documents that, in its Bedrock agentic retrieval flow, a new session may retrieve memory speculatively before planning, so the planning trace may not contain a memoryRetrieve action (AWS Bedrock agentic memory documentation).
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Separate retrieval from the final answer
If the trace contains the vendor memory, the failure is likely downstream of storage and retrieval. Compare the returned records with the context passed to the model and the final response. Check whether ranking, context limits, or output limits excluded the relevant content, and inspect the agent’s selected action and instructions. Salesforce recommends investigating agent, search/retrieval, and generation layers separately. Azure notes that a found document can still be absent from the final response if its content exceeds the output budget (Salesforce troubleshooting guide; Azure AI Search troubleshooting documentation).
Quick Recap
Match the symptom to the next check
| What you observe | Likely stage | Next check |
|---|---|---|
| Vendor is absent from a broad backend search | Write, extraction, validation, or indexing | Inspect the submitted event and write/import errors; compare source and indexed record counts. |
| Vendor exists in storage but not in the application query | Namespace, filters, strategy, or permissions | Use the same namespace with filters relaxed; verify indexed fields, values, and access rights. |
| Exact name or domain fails in semantic search | Query representation or retrieval mode | Try lexical/keyword and hybrid retrieval with the same query; compare returned records. |
| Trace reports a source error | Request or dependency failure | Inspect the source error for a malformed filter, configuration drift, authorization, throttling, timeout, or availability issue. |
| Trace contains the memory but the answer says there is none | Orchestration, ranking, context, or generation | Compare retrieved records with the prompt and final response; inspect the selected action and output limits. |
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
