Vectorize
company · United StatesScores
1 product on the map — 1 open.
Openness
5 high confidence- license
- MIT (engine, API server, control plane, clients and integrations)
- source
- public
- core features withheld
- no — the 5,403-path tree at 55dde52 holds the engine, API server, control plane, extensions and clients with no ee/, enterprise/, commercial/ or proprietary/ path, and Memory Defense and audit logs, listed on the Enterprise plan, ship in the open engine; Enterprise adds private deployment, support and the separate Memlake storage product
- commercial
- Hindsight Cloud and Hindsight Enterprise
Hindsight's memory engine, API server, control plane and clients are all in the public repository, and Vectorize says self-hosting carries no usage limits or restrictions. Its paid offerings are a managed cloud billed on usage and an enterprise deployment with private hosting, support and a separate storage product, not features cut from the open engine.
- https://raw.githubusercontent.com/vectorize-io/hindsight/main/hindsight-api-slim/pyproject.toml recorded 2026-11-20
The engine package manifest declares name "hindsight-api-slim" and license "MIT".
- https://raw.githubusercontent.com/vectorize-io/hindsight/main/LICENSE recorded 2026-11-20
The root LICENSE is the MIT License text, "Copyright (c) 2025 Vectorize AI, Inc.".
- https://raw.githubusercontent.com/vectorize-io/hindsight/main/README.md recorded 2026-11-20
The README installs the server by Docker, pip install hindsight-api or Helm, documents Memory Defense as an opt-in per-bank policy and webhooks in the open server, and offers Hindsight Cloud as the hosted option; it ends "MIT — see LICENSE".
- https://vectorize.io/pricing recorded 2026-11-20
The pricing page says "Hindsight is MIT licensed with no usage limits, no telemetry, and no restrictions", sells Hindsight Cloud on token usage, and sells Enterprise for private cloud, BYOC or on-premises deployment, SSO, Memlake and support.
Adoption
2 low confidenceAdoption is read from monthly PyPI downloads of hindsight-api-slim, the engine package that both hindsight-api and the hindsight-all bundle install. The recommended Docker image is published on GitHub's registry, which reports no download count, and the client packages count connections to any Hindsight server, the hosted cloud included, so neither is added.
- https://pypi.org/pypi/hindsight-api-slim/json recorded 2026-11-20
PyPI lists hindsight-api-slim, "Hindsight: Agent Memory That Learns", under the MIT license.
- https://pypistats.org/api/packages/hindsight-api-slim/recent recorded 2026-11-20
98,470 downloads of hindsight-api-slim on PyPI in the trailing 30 days.
- https://raw.githubusercontent.com/vectorize-io/hindsight/main/hindsight-api/pyproject.toml recorded 2026-11-20
The hindsight-api package depends on "hindsight-api-slim[all]==1.2.3", the engine package.
Capability
5 medium confidenceHindsight performs every part of the cycle itself, from extracting facts and entities to consolidating them into observations, decaying older memories in ranking and retrieving across several typed stores and a graph.
- https://hindsight.vectorize.io/developer/observations recorded 2026-11-20
The observations page describes deduplicated beliefs that evolve as new evidence supports, contradicts or extends them, and a section on observation lifecycle and invalidation.
- https://hindsight.vectorize.io/developer/retrieval recorded 2026-11-20
The retrieval page applies a recency boost with "Linear decay over 365 days from the memory's occurrence date" alongside temporal proximity and evidence strength.
- https://raw.githubusercontent.com/vectorize-io/hindsight/main/README.md recorded 2026-11-20
The README says retain "uses an LLM to extract key facts, temporal data, entities, and relationships", lists world facts, experiences, observations and mental models, and describes isolated banks and four parallel recall strategies.