MemoSift

Open-source core.
Platform when you're ready.

The context engine is open-source. The cloud platform adds persistent memory, semantic search, compliance reporting, and team management when your team needs it.

Open Source

$0/forever
  • Sync compression pipeline (<50ms)
  • Security scanning (flag/redact/block)
  • In-memory persistence (dev mode)
  • 3 LLM providers (OpenAI, Anthropic, Google)
  • 5 agent tool definitions
  • Apache 2.0 licensed
Popular

Pro

$39/mo
  • Everything in Open Source
  • Neon DB persistent memory (pgvector)
  • Redis caching layer
  • Session history & replay
  • Observability dashboard
  • Email support

Team

$99/mo
  • Everything in Pro
  • Neo4j relationship graph
  • Three-tier memory synthesis
  • Compliance audit trail
  • Team management & RBAC
  • Priority support

Enterprise

Custom
  • Everything in Team
  • HIPAA / PCI-DSS / SOX policies
  • Customer-managed encryption (BYOK)
  • Private deployment option
  • Custom domain hints & extractors
  • Dedicated support + SLA
Platform Comparison

Feature Matrix

CapabilityOpen SourceProTeamEnterprise
Context EngineFullFullFullFull
LLM Providers3333 + Custom
Agent Tools5555 + Custom
Memory StorageIn-memoryNeon DBNeon DBNeon DB + BYOK
Relationship Graph——Neo4jNeo4j
Observability—FullFull + TeamFull + Custom
Security ScanningLocal onlyCloud + AlertsCloud + AlertsCustom policies
Compliance Audit——GDPR, SOXHIPAA, PCI-DSS, SOX
Semantic SearchText onlypgvectorpgvectorpgvector + Custom
Session History—30 days90 daysImmutable forever

Getting started with the Pro tier

The asynchronous extraction + reconciliation pipeline that powers track, recall, compress, and explore is hosted at https://dev.memosift.com. To enable the Pro features:

  1. Sign up at memosift.com.
  2. Create a project from the dashboard.
  3. Create an API key inside that project (it starts with msk_).
  4. Configure your SDK to point at the cloud at https://dev.memosift.com.
  5. Pass the key when constructing MemoSift(...).
  6. You're live — track, recall, compress, explore all work.

Mode A — free forever

The Inspector primitives — classify, extract_metadata, scan— run locally inside the SDK with no cloud dependency, no API key, no rate limits. They're MIT-licensed and free to use forever for compliance scanning, telemetry, and routing decisions.

Read about Inspector mode

70%+ Token Savings

Pointer replacement + summary injection removes redundant tool results. Combined with dedup and context injection, enterprise agents processing 1 billion tokens/month save over $100K/year. GPT-5.4-nano extraction costs ~$0.001 per turn.

OPUS_SAVINGS: ~$10/MTok

SONNET_SAVINGS: ~$2/MTok

EXTRACTION_COST: ~$0.001/turn (nano)

Security-First Architecture

Security scanning runs before LLM extraction. Redacted content never reaches the model. Every memory is classified: public, internal, or redacted. Cloud platform offers AES-256 encryption, zero retention, and optional customer-managed keys.

SCAN_MODES: FLAG / REDACT / BLOCK

MEMORY_LEVELS: PUBLIC / INTERNAL / REDACTED

ENCRYPTION: AES-256-GCM + TLS 1.3