Total Recall iconTotal Recall

saas Freemium

Persistent memory for AI agents that enables cross-session context retention with local-first retrieval

98.0% LongMemEval
96.8% Recall@10
97% Token Savings

Overview

Total Recall provides persistent memory for AI agents, enabling them to retain context across sessions rather than starting fresh each conversation. The system captures narrative, decision, and pattern memories, then surfaces relevant context automatically or on-demand without bloating the active prompt. It uses local-first retrieval with zero LLM calls at search time, achieving deterministic performance.

The Verdict

Who Should Use Total Recall?

Best For

  • AI agent developers needing session continuity
  • Coding agents that need project context
  • Research assistants tracking long-term work
  • Teams wanting shared agent memory
  • Privacy-focused (local retrieval)

Not Ideal For

  • Simple chatbot use cases
  • Short, one-off conversations
  • Production at scale (still in beta)
  • Those needing immediate access (waitlist)

What's Great

  • Zero LLM calls at search time (local retrieval)
  • High benchmark scores (98% LongMemEval)
  • Progressive prompt integration (memory added only when needed)
  • Searchable dashboard for work history
  • Proactive memory surfacing when relevant
  • Significant token savings (up to 97%)

Watch Out For

  • Currently in closed beta (waitlist only)
  • No public pricing yet
  • Limited production track record
  • Single-developer project

Pricing

View all features & details

Memory Types

  • Narrative memory
  • Decision memory
  • Pattern memory
  • Cross-session persistence

Key Capabilities

  • Local-first retrieval
  • Zero LLM calls for search
  • Searchable dashboard
  • Proactive memory surfacing
  • Progressive prompt integration

Benchmarks

  • 98.0% LongMemEval accuracy
  • 96.8% recall@10
  • 98.2% correct in top 10 results
  • Up to 97% token reduction

Use Cases

  • Coding assistants
  • Planning agents
  • Research assistants
  • Debugging workflows

Benchmarks

98.0%
LongMemEval
Answered 490/500 questions correctly
96.8%
Recall@10
Deterministic local retrieval accuracy
97%
Token Reduction
Savings through out-of-prompt storage

How It Compares

Feature Total Recall Mem0 Zep
Local Retrieval Yes, zero LLM calls No No
Cross-Session Memory Yes Yes Yes
Progressive Loading Yes No Limited
Self-Hosted Option Unknown Yes Yes
Production Ready Beta Yes Yes

User Reviews

Loading reviews...