REM Labs · Zapier

Memory that sticks across every Zap

Zapier is incredible at fire-and-forget automation. It's also amnesiac by default — every run starts blank. Add a Webhooks by Zapier step pointed at REM Labs and every trigger becomes a durable, semantically-searchable memory your next Zap can recall.

Get API key Full docs

Free tier · No credit card · Works with every Zapier plan

Why Zapier + REM

Three things a persistent memory substrate gives your Zaps that they can't have on their own.

Continuity across triggers

Your Zap history only holds the last N runs. REM keeps everything forever — and lets a later Zap recall a Typeform response from 6 months ago with a semantic query, not a brittle record ID.

9 Dream Engine strategies

Overnight, REM summarizes your Zap firehose, links related events, dedupes duplicates, and flags anomalies. Your Monday morning Zap starts with the signal, not the noise.

Cross-tool portability

What your Zap stores, a LangChain agent queries. What an n8n flow writes, your next Zap reads. One REST endpoint, every automation tool on the market.

Install in 60 seconds

Zapier already ships the Webhooks app. You just point it at REM.

The official REM Labs Zapier app (with "Store Memory" and "Search Memory" actions) is in private beta pending Zapier's public-listing review. The Webhooks by Zapier pattern below is production-ready today and works on every Zapier plan. Request private-beta invite code in Discord.

Get your API key

Add a Webhooks by Zapier action

After your trigger step, add Webhooks by Zapier → POST. URL: https://remlabs.ai/v1/memory-set. Payload type: JSON. Add a custom header: Authorization: Bearer sk-rem-YOUR_KEY.

Map trigger fields into the payload

In the Data section, set namespace, content, and any metadata. Zapier's field-picker interpolates trigger values as usual — the whole thing stays visual.

POST https://remlabs.ai/v1/memory-set
Headers:
  Authorization: Bearer sk-rem-YOUR_KEY
  Content-Type: application/json

Data (JSON):
  namespace:  typeform-leads
  content:    {{Name}} ({{Email}}) said:
              "{{What's your biggest pain?}}"
  metadata:   { "source": "typeform",
                "submitted_at": "{{Submitted At}}" }

Common patterns

Three shapes: store on trigger, recall before ChatGPT/Claude steps, and a Schedule-by-Zapier Dream tick.

1. Trigger → store memory

Webhooks POST

Every new Slack message, Stripe charge, or Typeform submission becomes a durable memory. Map fields visually.

Trigger: New Stripe Charge
          |
          v
Webhooks by Zapier: POST
  URL:  https://remlabs.ai/v1/memory-set
  Auth: header "Authorization: Bearer sk-rem-..."
  Data (JSON):
    namespace:  stripe-events
    content:    Charge {{Amount}} {{Currency}} from
                customer {{Customer Email}}
                for {{Description}}
    metadata:   { "charge_id": "{{Id}}",
                  "customer_id": "{{Customer Id}}" }

2. Recall → prepend to ChatGPT/Claude step

Multi-step Zap

Before any ChatGPT or Anthropic Claude step, semantic-search REM and stitch the results into the prompt. Zero LLM changes required.

Trigger: New Slack DM to bot
          |
          v
Webhooks by Zapier: POST
  URL:  https://remlabs.ai/v1/memory-search-semantic
  Auth: header "Authorization: Bearer sk-rem-..."
  Data (JSON):
    namespace: slack-history
    query:     {{Text}}
    limit:     8
          |
          v
Formatter by Zapier: join results[].content with "\n- "
          |
          v
ChatGPT (or Claude): Conversation
  System: "Relevant memory:\n{{Formatter Output}}
           \n\nAnswer the user naturally."
  User:   {{Text}}
          |
          v
Slack: Send DM reply

3. Schedule → nightly Dream run

Schedule by Zapier

Trigger a Dream Engine consolidation on your own schedule. Get a summary DM back in Slack every morning.

Trigger: Schedule by Zapier
  Every Day at 03:00 local time
          |
          v
Webhooks by Zapier: POST
  URL:  https://remlabs.ai/v1/dream
  Auth: header "Authorization: Bearer sk-rem-..."
  Data (JSON):
    namespace:  stripe-events
    strategies: ["summarize", "link", "dedupe",
                 "score", "prune"]
          |
          v
Slack: "Dream run complete. {{delta.summary_count}}
        new summaries, {{delta.pruned}} stale
        memories removed. Top cluster:
        {{delta.top_cluster.title}}"

What you get

Everything a trigger-by-trigger automation tool leaves on the table.

Cross-Zap memory

One namespace can be read by every Zap you own. No more copy-pasting fields into Airtable just to look them up next week.

Semantic search in one step

Recall returns in a single webhook round-trip. Fuzzy "customer mentioned churn" queries, not brittle record-ID lookups.

Overnight consolidation

Dream Engine turns yesterday's 10,000 Zap runs into today's five useful summaries. Saves you tasks and saves your team reading time.

Portable everywhere

Same data powers Zaps, n8n flows, LangChain agents, Cursor MCP, Slack bots. REM is the seam between every tool you already run.