---
title: "past.dev vs Hindsight"
description: "Compare Hindsight memory and past.dev by licence, the operations each exposes, deployment, and what a recall returns."
canonical: https://past.dev/vs/hindsight
last-updated: 2026-10-26
---
# past.dev vs Hindsight

Source: https://past.dev/vs/hindsight

Hindsight is an open-source agent memory system from Vectorize, under MIT, with a hosted option. It exposes retain, recall and reflect, and ships as a Docker image, a Python server, or a library embedded in your process. past.dev provides a managed or self-hosted JSON API for timestamped source ingestion and grouped, ranked recall with dated source excerpts. Neither includes an agent runtime.

## What Hindsight is

[Hindsight](https://github.com/vectorize-io/hindsight) is an MIT-licensed agent memory system built by Vectorize, with [its own documentation](https://hindsight.vectorize.io). It organises long-term memory into four networks, world, experience, observation and opinion, and exposes three operations: retain to store, recall to search, and reflect to generate a disposition-aware response. Its retrieval runs vector search, keyword matching, graph traversal and temporal filtering in parallel rather than picking one.

There are three ways to run it: a Docker image, `pip install hindsight-api` for the server, and `pip install hindsight-all` to embed it in a Python process with no server at all. Its documentation names PostgreSQL with pgvector or Oracle AI Database 23ai as the store. Hindsight Cloud is the hosted option.

## What past.dev is

past.dev accepts raw text and its original timestamp without a required schema, and the timestamp can be backdated. The same public routes are available through the managed service and the self-host bundle.

`/recall` returns ranked documents with occurrence dates and source excerpts under a token budget, and the application decides whether those documents establish an answer. That is the clearest difference between the two: reflect writes the response itself, where `/recall` returns the evidence and leaves the wording to your model. Which you want depends on whether the answer is yours to write.

## Architecture differences

|  | Hindsight | past.dev |
| --- | --- | --- |
| Licence | MIT | Commercial, with a self-host image |
| What you operate | A Docker image, a Python server, or nothing when embedded | Nothing on the managed service; the published deployment bundle when self-hosted |
| Storage underneath | PostgreSQL with pgvector, or Oracle AI Database 23ai | Not part of the public API contract |
| Operations | retain, recall, reflect | Timestamped ingestion, and `/recall` |
| What a read returns | Searched memories, and with reflect a generated response | Ranked documents with occurrence dates and source excerpts |
| Abstention | Your code decides | Your code evaluates returned documents |
| Agent runtime | None | None |
| Cost visibility | Your infrastructure bill, plus model usage | Accounted per call on the console dashboard |

## How to choose

Choose based on the requirements below.

- **Choose Hindsight** if you want the engine under MIT, if you want it embedded in a Python process with no service to run, or if your storage has to be Oracle.
- **Choose past.dev** if you want dated source excerpts behind every recall, identities and audiences, per-call cost on a dashboard, and no licence to track.
- **Both** are memory services rather than agent frameworks, and leave your agent loop where it is. [What is agent memory?](/what-is-agent-memory) covers what such a service has to do.

## What this page does not claim

This page contains no head-to-head scores. Our evaluation uses BEAM and LoCoMo with community-corrected answer keys, with full-context and plain-RAG baselines and several judges. The method is published on [benchmarks](/benchmarks). We will publish product comparisons after documenting a reproducible configuration for the other product.

## Related

- [What is agent memory?](https://past.dev/what-is-agent-memory)
- [past.dev vs mem0](https://past.dev/vs/mem0)
- [past.dev vs Graphiti](https://past.dev/vs/graphiti)
- [Memory API overview](https://past.dev/docs/memory-api/overview)
- [How we measure memory](https://past.dev/benchmarks)