Hebbrix
Outcome-weighted memory layer that keeps what worked, not just what sounds related
Hebbrix is the rare memory layer built around outcome feedback, not just semantic resemblance. If your agent keeps repeating failures because retrieval ignores whether a past action worked, this is worth a serious trial. The free tier is generous, and the One-line OpenAI drop-in makes adoption painless. Alternatives like Mem0 or a plain vector DB lack the outcome-signal loop that compounds learning.
Verified 2d ago · liveness 78/100 · cite: rightaichoice.com/tools/hebbrix
- Developers building production agents that need to learn from what worked, not just what sounds similar
- Teams using LangChain/LangGraph wanting drop-in persistent memory with knowledge graph capabilities
- Customer support bots that must remember user history and adapt based on outcomes
- Voice agents needing to recall preferences and avoid repeating failures over long periods
- Simple chatbots that don't need cross-session memory (overkill)
- Users seeking a fully free unlimited memory solution (credit caps bite)
- Teams needing on-premise deployment without sales contact (no self-hosted option listed)
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Hebbrix if you don't need cross-session memory, want a fully free unlimited solution, require on-prem without enterprise sales, or can't tolerate credit-based billing with potential overage costs.
Exceeding your credit allowance on Free or Starter pauses memory writes until the next cycle, which can silently degrade your agent's memory-dependent features.
Hebbrix's freemium tier (1,000 credits/mo) is generous for trials and small projects. Starter at $19/mo fits early production with 25K credits. Pro at $99/mo is cost-competitive with Mem0's paid tiers when you need outcome-weighted memory and knowledge graph. Scale at $399/mo suits high-volume agents. Enterprise is custom.
In short
Hebbrix — Outcome-weighted memory layer that keeps what worked, not just what sounds related. Best for Developers building production agents that need to learn from what worked, not just what sounds similar, Teams using LangChain/LangGraph wanting drop-in persistent memory with knowledge graph capabilities, Customer support bots that must remember user history and adapt based on outcomes. Free to start; paid plans from $19/mo.
What's new in Hebbrix
Checked 8 days agoAcross the latest 5 updates: 4 feature updates and 1 changelog entry.
API v109: Evidence-Bound Retrieval Safety
All retrieval modes now share one safety envelope with grounded/degraded states and evidence memory IDs, plus fail-closed claim grounding.
API 1.0.0: Unified Retrieval and Abuse Controls
Unified memory retrieval across search, chat, and cache; operational metrics moved behind private networking; renewable sessions and signup abuse controls.
Deterministic Retrieval and Unified Write Readiness
All memory writes now follow a truthful readiness contract with 200/201/202 responses; corrections get status endpoints and idempotency.
Smart Memory Ingestion Pipeline
infer:true mode extracts atomic facts with a 3-tier pipeline; memory worthiness classifier rejects noise; uses gpt-5-nano for inference.
5-Layer Hybrid Search Engine
Search now combines semantic, BM25, graph, importance, and recency signals with optional cross-encoder reranking and score explanations.
Viability Score
How well maintained and how widely used is Hebbrix? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: September 2026
How we score →Key Features
- Outcome-weighted recall based on reported success/failure
- 3-tier memory (short, mid, long-term) with automatic promotion
- 5-layer hybrid search: semantic, BM25, graph, temporal, importance
- Cross-encoder reranking (ONNX) for high precision
- Knowledge graph extraction with temporal reasoning and contradiction detection
- Smart memory ingestion with infer:true and noise-rejection classifier
- SEARCH_MIN_SCORE setting to filter weak matches
- Document & media upload (PDF, DOCX, audio, video) up to 100MB
- Automatic transcription and chunking
- User profiles with automatic fact extraction
- Memory decay on usage, recency, importance, corrections
- OpenAI-compatible chat endpoint (change base URL only)
- REST API with batch create up to 100 memories
- Versioned memory and provenance tracking
- Portable across models (BYOK support)
About Hebbrix
Hebbrix is a persistent memory layer for AI agents that replaces similarity-based retrieval with outcome-weighted recall. Instead of surfacing documents that merely sound related, it prioritizes memories that led to successful outcomes and lets dead ends decay. Built for developers running production agents, it integrates by changing one line of code in your existing OpenAI client, automatically injecting relevant context into each request. This makes it a practical choice for teams building customer support bots, voice agents, or any long-running AI system that needs to compound learning over time. At its core, Hebbrix uses a 3-tier cognitive memory (short, medium, and long-term) with automatic promotion based on access frequency and importance. Its 5-layer hybrid search engine combines semantic vectors, BM25, knowledge graph traversal, importance scoring, and recency boosting, with an optional cross-encoder reranker for high-precision workloads. The system also auto-extracts knowledge graphs from natural language without any schema setup, enabling temporal reasoning and contradiction detection. A self-improving retrieval system applies six quality checks and reinforcement learning after each interaction. Recent updates have added a smart memory ingestion pipeline with noise-rejection classification and automatic fact extraction, plus a SEARCH_MIN_SCORE setting to filter weak matches. Hebbrix supports document and media upload (PDF, DOCX, audio, video) up to 100MB, with automatic chunking and transcription. It offers a generous free tier (1,000 credits/month), escalating paid plans, and enterprise options with SOC 2/HIPAA compliance in progress. Compared to vector databases or simple key-value stores, Hebbrix focuses on outcome success over similarity, making it a strong fit for agents that need to remember what worked and avoid repeating failures. Its model-agnostic design means memory remains portable if you switch LLMs.
Behind the Verdict
The pitch is refreshingly blunt: your agent remembers what sounds related, not what actually worked. Hebbrix scores memories by whether acting on them led to a good result, and that feedback loop is what separates it from a dozen memory layers that are just pretty vector stores. For a team debugging why a support bot keeps giving the same wrong answer, the outcome signal is the difference between a patch and a fix. We'd reach for this when your agent's mistakes are costly and repeatable—customer support, voice agents, sales follow-ups. The free tier gives you 1,000 credits a month, enough to kick the tires without a card, and the drop-in OpenAI-compatible endpoint means you can be live in an afternoon. No migration, no rewrite. Where it bites: credits are a real constraint at scale. 1,000 credits is roughly 300 chat turns for a hobby project; a production assistant serving thousands of daily users will blow through the free tier fast. You'll be on Pro ($99/mo) or Scale ($399/mo) quickly, and overage billing is only available on those tiers. If you're running high frequency, low-value calls, the credit model adds a tax that a flat-rate vector DB won't. Compared to Mem0, which we've reviewed, Hebbrix is the more opinionated system. Mem0 is a solid plug-and-play memory that's good for general recall; Hebbrix is built around the outcome question and carries more machinery—knowledge graphs, temporal reasoning, contradiction detection. That's power when you need it, but it's also more to understand. The docs help, but you'll spend an afternoon with the concepts. One more caveat: the team is small (two engineers), so enterprise buyers should expect to lean on the sales process for SLA and compliance details—SOC 2 is still in progress. For most teams, though, the
Researching Hebbrix? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Hebbrix actually fits — and what changes day-one when you adopt it.
You install the Hebbrix SDK, create a collection, and store past resolved tickets. You point your OpenAI client at Hebbrix's base URL.
Outcome: The bot recalls past resolutions and successful outcomes, reducing repeat complaints and freeing your team.
You use the Python SDK to store call transcripts and user preferences, and enable infer:true for automatic fact extraction.
Outcome: The assistant remembers callers' preferences and past interactions, delivering personalized responses without re-prompting.
You connect Hebbrix via the LangChain integration, replacing your previous vector store without changing your agent logic.
Outcome: Your agent now uses outcome-weighted recall, so it surfaces what worked in past runs rather than merely related text, improving decision quality.
Use Cases
- Store user preferences across sessions to personalize responses without re-prompting.
- Enable customer support agents to recall past issues and resolutions automatically.
- Build a knowledge base for internal tools that learns from successful outcomes.
- Create voice assistants that remember user context across calls.
- Power sales agents that track prospect interactions and adapt pitches accordingly.
Models Under the Hood
as of 2026-08-28
Limitations
- Free and Starter plans pause memory writes when exceeding credits, but reads continue.
- Pro and Scale plans can enable overage billing to avoid interruption.
- Exports of memory and knowledge graph are available on any plan, and memory is portable across models.
as of 2026-08-26
Verification history
We have re-verified Hebbrix 7 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 7 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Hebbrix tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Free
$0/mo
Ideal for
Solo developers or small projects trying out Hebbrix, with 1,000 credits/month—enough for about 300 chat turns—to test outcome-weighted memory without cost.
What this tier adds
Starting tier with adaptive hybrid search and profile extraction, but limited to 1K credits and community support.
Starter
$19/mo
Ideal for
Teams shipping their first product with 25K credits/month, needing knowledge graph and the ability to bring your own model key.
What this tier adds
Adds knowledge graph, BYOK, and 25x more credits than Free.
Pro
$99/mo
Ideal for
Production agents with 200K credits/month, needing the Outcome Memory API, overage billing, and priority support.
What this tier adds
Adds the core outcome-weighting functionality and overage billing.
Scale
$399/mo
Ideal for
High-volume workloads and multi-tenant platforms with 1M credits/month, needing advanced analytics and higher rate limits.
What this tier adds
Adds advanced analytics and higher rate limits on top of Pro.
Enterprise
Custom
Ideal for
Large teams with compliance and scale needs, requiring SSO, custom terms, and solutions engineering.
What this tier adds
Custom pricing with pay-as-you-go, BYOK any model, security review, and enterprise-grade controls.
Where the pricing makes sense
The company stage and team size where Hebbrix's pricing actually pencils out — and where peers do it cheaper.
Hebbrix's freemium tier (1,000 credits/mo) is generous for trials and small projects. Starter at $19/mo fits early production with 25K credits. Pro at $99/mo is cost-competitive with Mem0's paid tiers when you need outcome-weighted memory and knowledge graph. Scale at $399/mo suits high-volume agents. Enterprise is custom.
Setup time & first value
How long it actually takes to get something useful out of Hebbrix — broken out by persona, not the marketing-page minute.
For developers using the OpenAI drop-in: minutes. Change the base URL and you're done. For LangChain/LangGraph or MCP integrations: under an hour to wire up and test. For smart ingestion and knowledge graph tuning: allow a few hours to refine extraction and thresholds like SEARCH_MIN_SCORE.
Switching to or from Hebbrix
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Mem0 or a vector DB: Use Hebbrix's bulk import to bring in existing memories and start querying via the hybrid search, with outcome-weighting as a new layer.
- ↗To any LLM provider: Export memories and the knowledge graph anytime and import into another memory store; Hebbrix guarantees portability.
Integrations
Resources & Guides
Tutorials & Learning
Official links
Featured Head-to-Head Comparisons
Hebbrix vs Spider Cloud
Spider Cloud and Hebbrix solve different problems. If your AI agent needs to fetch fresh web data for RAG or scraping, Spider Cloud’s Rust engine and Browser AI commands are unmatched. If you need persistent, self-improving memory for agents (customer support, voice, etc.), Hebbrix’s outcome-weighted recall and knowledge graph are a game-changer. Choose based on your primary need: external data or internal memory.
Hebbrix vs Presto Voice
Presto Voice and Hebbrix serve entirely different markets: Presto is a turnkey drive-thru voice AI solution for QSR chains, while Hebbrix is a developer-focused memory layer for building any AI agent. Choose Presto if you run a multi-location fast-food chain and want proven revenue lift. Choose Hebbrix if you're an AI developer needing persistent, outcome-weighted memory that integrates via one line of code.
Hebbrix vs Temporal Ai
Temporal is the clear choice if you need durable, fault-tolerant orchestration for complex multi-step workflows, especially AI agents that must survive crashes. Hebbrix excels at adding smart, persistent memory to existing agents, but if you already need orchestration, Temporal's memory is built-in. For a pure memory layer on top of a separate orchestrator, Hebbrix is drop-in easy. For everything-in-one, go Temporal.
Popular in Agent Memory & Runtimes
Frequently Asked Questions
Categories
Used Hebbrix? Help shape our editorial sentiment research.