Recalld
Recalld is a long-term memory layer for AI agents. It stores conversations and documents, reconciles new facts with existing memory, and retrieves relevant facts through an API or native MCP server.
Recalld is chat software teams evaluate for chat. Use this page to review pricing, integration signals, and the best alternatives before you commit.
Profile facts come from the vendor source. AiMatch labels unknown pricing or API details instead of estimating them.
Review official source →Quick Overview
Best for: Chat
What it does
Chat software for decision-makers comparing workflow fit and alternatives.
Best fit
Chat
Pricing snapshot
Freemium from $0
Next step
Compare Recalld with similar tools before you shortlist it.
Compare this tool before you shortlist it
Review alternatives, pricing posture, and workflow fit side by side.
Recalld is a long-term memory service for AI agents. Developers can write conversations or documents to an agent’s memory; Recalld extracts atomic facts, anchors them in time, and reconciles new information with what it already knows. Facts can be updated, superseded, or merged as information changes.
Agents can retrieve a curated set of facts with the recall endpoint, or use search to get raw vector-similarity candidates for their own filtering. Recalld is also available as a native MCP server, and Recalld Chat provides a chat interface that displays recalled facts and credit costs. Users choose an EU or US region when signing up.
Recalld is the memory layer for AI agents. Recall, not retrain. Persistent, curated memory that returns only what answers the question. 88.7% on the LoCoMo benchmark using 243 tokens per query.
Own this listing?
Claim this page for a one-time $29 to add pricing, features, screenshots, verified owner details, and a clearly labeled 30-day category position after the profile is live.
Claim this listing for $29Key Features
Fact extraction and reconciliation
The add operation accepts turns or documents, extracts atomic facts, anchors them in time, and can update, supersede, or merge facts when new information arrives.
Curated recall
The recall endpoint combines retrieval with an LLM curation pass and returns a short, ranked set of facts intended to answer the query.
Raw search
The search endpoint performs dense-vector lookup without an LLM in the loop, returning candidates for the agent’s own model to filter.
Native MCP server
Recalld exposes memory tools to MCP-compatible clients, including add, recall, search, and memory filters. Connections are bound to an agent when authorized.
Time-aware facts
Facts can retain information about when they apply, so past and current information can remain distinct.
Recalld Chat
A chat assistant built on Recalld’s memory engine that scopes memory by conversation and shows which facts were recalled and the credit cost.
Regional data residency and exports
Users select an EU or US region at signup. Stored facts can be exported as JSON or CSV.
Usage dashboard
Credit-based usage is broken out by ingestion, recall, and search in a dashboard.
Pricing
The free plan costs $0 and includes 15,000 credits per month, plus 50,000 extra credits in the first month. It includes the default model, 30-day active memory, and a 60-requests-per-minute limit. The page says no card is required.
Free
$0- 15,000 credits per month
- 50,000 extra credits in the first month
- Default model
- 30-day active memory
Starter
$10/month- 200,000 credits per month
- Default model
- 6-month active memory
- 120 requests per minute
Pro
$29/month- 870,000 credits per month
- All models
- Unlimited active memory
- 600 requests per minute
Scale
$99/month- 3.96 million credits per month
- All models
- Unlimited active memory
- 1,200 requests per minute
Use Cases
Add long-term memory to AI agents
Agent developers can store conversations and documents, then retrieve relevant facts in later interactions without building their own vector database and memory retrieval pipeline.
Connect agent memory through MCP
Teams using an MCP-compatible agent, IDE, or chat app can authorize Recalld as a server and provide memory tools to the client.
Use raw retrieval with a custom filtering model
Developers who want to handle candidate selection themselves can use search to retrieve raw vector-similarity results without Recalld’s curation pass.
Chat with scoped memory
Users can use Recalld Chat to have conversations that recall facts, keep memory scoped to each conversation, and show which facts informed replies.
Integrations
Model Context Protocol (MCP)
Recalld provides a native MCP server for compatible agents, IDEs, and chat apps. The page names Claude, Cursor, and VS Code as supported MCP hosts.
Google model provider
On Pro and Scale, users can bring their own key for a model provider; the page says users pay Google directly for model use.
Benefits
Limitations
Frequently Asked Questions
Can I use Recalld without building my own memory retrieval pipeline?
Can I choose where my data is stored?
Can I use my own model provider key?
Can I export stored memory?
Getting Started
- 1 Create an account and start on the free tier; the page says no card is required.
- 2 Choose the EU or US region when signing up. The region cannot be changed later.
- 3 Connect through the API or add Recalld as an MCP server in a compatible client; MCP authorization occurs over OAuth on first use.
- 4 Write a turn or document to memory, then use recall for curated facts or search for raw candidates.
Support
Community support
Listed for the Free plan.
Email support
Listed for the Starter plan.
Priority support
Listed for the Pro and Scale plans.
Documentation
The page links to product documentation and an API reference.
Security and compliance resources
The page offers a DPA, a sub-processor list, and a security questionnaire on request.
API
The page links to API reference documentation and shows POST endpoints for adding memory at /v1/agents/{id}/memory and recalling it at /v1/agents/{id}/memory/recall. It also describes a search endpoint and MCP tools for add, recall, search, and filters.
Plan limits listed on the pricing page: Free 60 requests per minute, Starter 120, Pro 600, and Scale 1,200.
Compare Recalld with similar tools
See how it stacks up against alternatives
Related Tools
View all 65 →
Chat & Ask AI
Chat & Ask AI is an advanced multi-model AI assistant and platform offering web search, document and link analysis, image generation, YouTube summarization, and chat features powered by a wide range of LLMs and AI technologies.
Quickblox
QuickBlox UI Kit is a pre-built, customizable chat UI component library and messaging solution from QuickBlox that accelerates building messenger apps across platforms, with built-in AI features (Answer Assist, Translate, Rephrase) and integration with QuickBlox SDKs and Server API.
DeepSeek Online
DeepSeek Online provides free, no-registration access to DeepSeek-V3, a 671B-parameter open-source language model available as an online demo and as downloadable source code for local installation and integration.
Premium Alternatives
pandachat-ai
PandaChat runs supervised AI customer operations for complex Adria/DACH e‑commerce stacks: a managed team of specialists and supervisors that connects to your systems, handles multilingual 24/7 support across channels, and sends daily briefings with value reports.
Uncensored
Uncensored AI is a chat and developer platform that provides access to a broad catalog of powerful, minimally filtered AI models for chat, voice, vision, code, and API access, with privacy features like auto-redaction and developer tooling.
Promptbuilder
Prompt Builder is a web app that generates, optimizes, tests, and manages AI prompts tuned to specific models (ChatGPT, Gemini, Claude, Grok, etc.), with a prompt library, optimizer, built-in chat assistant, and versioning for prompt workflows.