Prompeteer Memory
Bring selected documents into a private knowledge base you can inspect and reuse.
Open MemoryExplore team workflows →
Illustrative Memory task
Use the selected product and brand-voice pages to outline a customer launch brief. Cite the source for each feature, availability and pricing claim. Follow the voice guidance, but treat example copy as style evidence rather than product fact. List any contradictions or missing details before drafting the final announcement.
Read the full example →
See which source reaches the task
Select the approved product note and inspect the excerpt returned for “CSV export.” A second query has no supporting passage. This reproducible example runs the same lexical-index functions used by Memory.
Reproducible example
Retrieve a passage from the selected source
Two authored document fixtures: an approved product note and past campaign copy. Actual production lexical-index results are shown for the explicitly selected product note.
Checked 19 October 2026 · product-workflows-v2 · no model calls
- Select Choose the current product note. Leave the old campaign outside the selected-source index.
- Retrieve The CSV export query returns the product excerpt with its page identifier.
- Missing facts The refund query returns no hits. The demonstration produces no answer to fill that gap.
- PassedQuery returns the approved export source
- PassedUnselected campaign copy never enters the index
- PassedAn unsupported query returns no lexical hits
Inspect inputs and recorded output
{
"input": {
"pages": [
{
"schemaVersion": 1,
"pageId": "release-facts",
"title": "Approved export release",
"content": "CSV export is available to existing customers. The approved release note does not state a launch date.",
"links": [],
"claims": [],
"sourceIds": [
"approved-release-note"
],
"updatedAt": "2026-10-19"
},
{
"schemaVersion": 1,
"pageId": "voice-example",
"title": "Past launch copy",
"content": "Our past launch offered a discount. This example describes a previous campaign and is not an approved fact about the current release.",
"links": [],
"claims": [],
"sourceIds": [
"past-campaign-copy"
],
"updatedAt": "2026-10-19"
}
],
"selectedPageIds": [
"release-facts"
],
"query": "CSV export",
"missingQuery": "refund eligibility"
},
"output": {
"selectedPageIds": [
"release-facts"
],
"retrievedPageIds": [
"release-facts"
],
"excerpts": [
{
"pageId": "release-facts",
"title": "Approved export release",
"snippet": "CSV export is available to existing customers. The approved release note does not state a launch date."
}
],
"missingQueryHits": 0
}
}
Production functions: buildKbLexicalIndex, queryKbLexicalIndex.
Download the input and result record, or replay all four examples locally with Node.js 20 or later.
Inputs and results (JSON)Offline replay (.mjs)File checksums
node replay-product-workflows-v2.mjs
The replay contains the relevant production functions. It needs no account, credentials, package install or network connection. Inspect the downloaded script before running it.
Selection here is explicit fixture input. This tests lexical retrieval, not production account authorization, storage, embedding retrieval, source compilation or answer generation. No model calls occur. Keep the source reference and inspect the original passage.
Source: Downloadable inputs, outputs and checks
Keep source material connected to the work
Memory organizes supported uploads into browsable knowledge pages. Inspect the source material behind a statement before carrying it into a new prompt or answer. Use the relevant material for the task rather than sending an entire document collection.
Supported connected workflows can search authorized excerpts and return grounded answers with page citations. They should identify insufficient evidence when the selected material cannot answer the question.
A launch brief grounded in product facts
Keep the launch instruction separate from the documents that establish what is true.
Illustrative Memory task
Use the selected product and brand-voice pages to outline a customer launch brief. Cite the source for each feature, availability and pricing claim. Follow the voice guidance, but treat example copy as style evidence rather than product fact. List any contradictions or missing details before drafting the final announcement.
A knowledge base needs maintenance
A stored source can be outdated, contradictory or incomplete. Review processing status, source dates and disputed statements. Do not treat successful upload as proof that every passage was understood.
Available upload, browsing and question-answering features depend on your plan and workspace policy. Check the current offer before building a workflow around a document limit.
Common questions
Does every connected tool receive my whole Memory?
Supported Memory retrieval is tied to the authenticated account and returns relevant authorized material rather than the entire collection by default.
Does Memory replace my original files?
Keep authoritative originals and ownership records. Memory is a working knowledge layer for selected material, not a replacement for your document system.
Sources and review
Published by Prompeteer. See the people responsible for the product and contact the team with corrections. The linked workflow artifacts describe exactly what was checked; illustrative examples are not customer results.
Reviewed September 14, 2026
- Supported connections and Memory access
- Current plans and availability
- Downloadable inputs, outputs and checks
Continue your workflow
Reuse what you already worked out →
Organize selected knowledge for future prompts and decisions. Separate durable facts from temporary task context and keep sources easy to inspect.
Make the knowledge easier to inspect →
Understand how selected sources become browsable, linked knowledge. Review provenance, contradictions and updates before using a wiki as AI context.
Turn a document into useful context →
Prepare selected files for contextual prompts and Memory. Review source quality, remove unrelated material and keep the requested output explicit.
Put the workflow to work.
Open MemoryExplore team workflows →
Source: https://prompeteer.ai/ai-memory