Recall, not retrain.
The memory layer for AI Agents.
Recalld gives your agents long-term memory that builds on new information. Write conversations and documents once; Recalld extracts facts, updates or supersedes older ones when new information arrives, and retrieves relevant facts for later questions. No need to build your own vector database or memory retrieval pipeline.
Three primitives. That's the whole API.
Stop bolting a vector database, a re-ranker, and a dedup pipeline together. Recalld is one endpoint for writing memory and two for reading it.
- add Stream in turns or documents. Recalld extracts atomic facts, anchors them in time, and reconciles them against what it already knows, updating, superseding, or merging automatically.
- recall Hybrid retrieval plus an LLM curation pass. Returns a short, ranked set of only the facts that answer the query.
- search Raw dense-vector lookup when you want the full candidate set and your own model does the filtering. No LLM in the loop.
# Write memory once
POST /v1/agents/{id}/memory
{ "content": "We migrated billing to Stripe on Tuesday." }
# Read back only what matters
POST /v1/agents/{id}/memory/recall
{ "query": "What payment provider do we use?" }
→ only what answers it, ~240 tokens
[{ "fact": "Billing runs on Stripe (since Tue)." }]Curation is the product.
Most memory tools hand your agent a pile of "relevant" chunks and let it sort out the mess, burning context and inviting hallucination. Recalld does the judgement for you.
search
rawVector similarity on the raw query, no LLM in the loop. Returns the top candidates and lets your own model do the filtering. Fast, cheap, predictable.
recall
curatedRuns a selection pass over the candidates and returns only the passages that actually answer the query. That pass runs on our infrastructure, not in your context window.
Same questions, same answerer, same judge, same memory. Recall scored 88.7% to search's 88.2%, on 6.7× fewer tokens. Anyone can buy accuracy with a bigger context window. The hard part is not needing one.
Recall was measured on 6 August 2026 and search on 1 September 2026. The engine is updated regularly, so current results may differ slightly.
Those saved tokens come off your model bill, and a DIY vector-database stack has costs of its own. See the full math →
Measured, not marketed.
Evaluated on LoCoMo (all 10 conversations, 1,540 non-adversarial questions) using the Agent Memory Benchmark harness, with its own answer and judge prompts. We didn't write the harness, the answer prompt, or the judge: Vectorize, another memory company, did. We ran it ourselves. Every change we made to the harness is a patch in our results repo, each marked with whether it can affect scoring.
Full breakdown: single-hop 91.7%, temporal 89.7%, open-domain 83.3%, multi-hop 80.5%.
Answerer gemini-3.1-pro-preview, judge gemini-2.5-flash-lite, the
same for every question. Recall's selection pass is an LLM call that runs
on our side, so it costs us compute, not your context window; the 243 tokens
are what lands in your agent. LoCoMo, August 2026. The harness's judge is
lenient: it often accepts "I don't know" as correct, so read accuracy as an upper
bound. We left it unchanged so the number compares with other results on the same
harness. Known limitations →
Don't take our word for it.
We didn't build the scoreboard we're standing on. Every number here comes from the Agent Memory Benchmark, an open harness maintained by another memory company, with its prompts and judge untouched. We ran it; our adapters, patches and raw reports are public. Clone it, add your keys, run it against us. If we're wrong, it'll say so.
Built for production agents.
Time-aware facts
Recalld can record when facts apply, so a past location and a current one remain distinct.
Fact updates
Recalld evaluates new information against stored facts and can update, supersede, or merge them.
Bring your own key
On Pro and Scale, point Recalld at your own model provider. You pay Google directly; we charge only for orchestration.
Active memory windows
Facts stay stored forever but stay searchable on a plan-tuned window. Nothing is ever deleted.
Regional data residency
Pick the EU or US region when you sign up. Your memory lives there, GDPR-compliant by default, and never moves.
Usage you can read
Credit-based billing with a dashboard in plain credits: ingestion, recall, and search broken out. Export to CSV.
Plug memory into any MCP client.
Recalld ships as a native MCP server. Point any MCP-compatible agent, IDE, or chat app at it and your assistant gets long-term memory as first-class tools. No SDK, no glue code, no bespoke integration.
- add The same primitives you already know, exposed as MCP tools. Your client calls them directly. Recalld handles extraction, reconciliation, and curation behind the scenes.
- recall The model decides when to remember and when to recall. Curated facts flow straight back into the context window, only what the query needs.
- scope Every connection is bound to one agent when you authorize it, so memory stays isolated per client and per tenant.
// mcp.json - add Recalld to any client
{
"mcpServers": {
"recalld": {
"url": "https://eu.recalld.ai/mcp"
}
}
}
// No key in the file. Your client authorizes over OAuth
// on first use, and the agent is bound to that connection.
// Your agent now has four tools:
memory_add - write a turn or document to memory
memory_recall - get back only the facts that matter
memory_search - raw candidate lookup, you filter
memory_filters - see what kinds, authors and paths existGDPR-compliant by design.
Memory is sensitive by definition. Recalld is built for the EU from the ground up. Your data stays in-region, under your control, with a clear paper trail for every byte you store.
EU data residency
Every fact, embedding, and log line is stored and processed inside the region you chose, EU or US. Nothing leaves it without your explicit choice.
Right to erasure
Delete an agent or account and every associated fact and vector is hard-purged, not soft-flagged. Erasure requests resolve in minutes, not tickets.
Encrypted end to end
TLS 1.3 in transit and AES-256 at rest, everywhere. Bring-your-own-key connections keep provider secrets sealed in an isolated vault.
Data processing agreement
A standard DPA is available for every plan. You remain the data controller; Recalld acts strictly as your processor under Article 28.
Transparent sub-processors
A public, versioned list of every sub-processor we use. You're notified before any change takes effect. No silent additions.
Portable on demand
Export every stored fact as JSON or CSV at any time. Your memory is yours: take it, audit it, or move it whenever you want.
Need our DPA, sub-processor list, or a security questionnaire filled in?
Simple, credit-based pricing.
Buy credits, spend them per request. Search is ~40× cheaper than curated recall, so most of your traffic costs almost nothing.
- Default model
- 30-day active memory
- 60 req/min
- Community support
- Default model
- 6-month active memory
- 120 req/min
- Email support · CSV export
- All models
- Unlimited active memory
- 600 req/min · BYOK
- Advanced analytics · Priority support
- All models · best $/credit
- Unlimited active memory
- 1,200 req/min · BYOK
- Advanced analytics · Priority support
Need one-off credits? Top-ups from $5. All plans include the full API.
Prices exclude VAT and sales tax, which are added at checkout where they apply.
Give your agent a memory.
Recall, not retrain. Start free in the EU or US region today.