Tenure
Tenure is an open-source, MIT-licensed AI memory and policy platform that provides a belief-state proxy between AI coding clients and LLM providers, serving engineering teams and individual developers with freemium self-hosted and paid enterprise tiers.
- Company typePrivate
- Founded2026
- Headquarters—
- Headcount1–10
- GTM typeB2B and B2C
- OfferingSoftware
What Tenure does
Tenure is a private, MIT-licensed AI infrastructure company that sells a governed context platform for AI agents, primarily serving software engineering teams and individual developers using AI coding assistants. The company was founded in 2026 and operates with 1-10 employees under a bootstrapped, open-source-first model. Its product sits as a transparent proxy between AI clients (Cline, Continue, Windsurf, Cursor, Open WebUI, OpenClaw, mobile) and upstream LLM providers (Anthropic, OpenAI, AWS Bedrock, LiteLLM, Ollama), replacing raw chat history with a structured belief store.
The core technology uses typed, scoped, versioned belief objects (Preference, Decision, Entity, Relation, Open Question) retrieved through alias-weighted BM25 search with hard architectural scope isolation. The company reports 1.00 mean retrieval precision and a 0.00 drift score on its published PrecisionMemBench (arXiv 2605.11325, HuggingFace dataset), compared with 0.06-0.22 precision and 0.81-0.94 drift for Mem0, Zep, Supermemory, GBrain, Agentmemory, and Hindsight, with sub-15ms p50 retrieval latency. The product portfolio includes the core Tenure platform, a free self-hosted Tenure Community tier, a paid Tenure Team enterprise tier (Helm chart, OIDC/SSO, SCIM 2.0), the GroundEval policy authoring tool, and the PrecisionMemBench benchmark suite. Security features include MongoDB client-side field-level encryption and a runtime gateway that blocks unauthorized agent tool calls before they reach business systems, with EU AI Act compliance positioning for regulated buyers.
Tenure makes money through a freemium model: the Community edition is MIT-licensed and free (single Docker container, no account), while the Team/Enterprise tier is a subscription product delivering OIDC/SSO, SCIM provisioning, per-user PAT management, audit trails, and multi-tenant isolation. The /pricing page exists but actual Team/Enterprise price points are not disclosed. Distribution is fully self-serve via GitHub, Docker/Helm, VS Code Marketplace, Open VSX, ClawHub, and npm, with no sales team, no paid advertising, and developer-first marketing anchored on benchmark results and an arXiv research paper.
Tenure firmographics
Firmographics- Name
- Tenure
- Legal name
- Tenure
- Website
- https://tenureai.dev
- Company type
- Private
- Founded year
- 2026
- Operating status
- Operating
- Headcount range
- 1–10 employees
- Short description
- Tenure is an open-source, MIT-licensed AI memory and policy platform that provides a belief-state proxy between AI coding clients and LLM providers, serving engineering teams and individual developers with freemium self-hosted and paid enterprise tiers.
- Ownership category
- akta.pro rank
Tenure industry classification
Industry- Product category
- AI Memory Infrastructure
- NAICS
- Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (5182)
- SIC
- Services-Prepackaged Software (7372)
- akta.pro primary industry
- Model Hosting, Serving & Inference Platforms (HDAAACAB)
- akta.pro secondary industries
- Identity & Access Management (IAM) Administration Services (BPAEABAJ), User Lifecycle Provisioning & SCIM Automation (HDAEAJAK)
Keywords
Tenure business model
Business model- GTM type
- B2B and B2C
- Offering type
- Software
- Cost components
- Personnel, Technology or R&D, Infrastructure, Marketing or Sales, Operations
Revenue model
- Tenure Community: MIT-licensed free download for individual developers. Single Docker container, no account required, runs locally on localhost.
- Tenure Team (Enterprise): Helm chart deployment for engineering teams with OIDC/SSO integration, SCIM provisioning, per-user PAT management, org-wide audit trails, orientation tax dashboard, and team/org multi-tenant isolation. Enterprise governance layer on top of the core memory proxy.
Go-to-market motion1 record
Distribution channels6 records
Marketing channels7 records
Tenure product offering
Product offeringCore offering
Tenure is a governed context platform for AI agents that operates as a memory proxy sitting between AI clients (VS Code, Open WebUI, OpenClaw, mobile chat) and upstream LLM providers. It extracts structured, typed, scoped, versioned beliefs from conversations and injects the right belief at each turn using alias-weighted BM25 retrieval with hard scope isolation, achieving 1.0 retrieval precision versus 0.06-0.22 for competitor memory systems. A runtime gateway authenticates agents, evaluates preconditions against policy, and allows, blocks, or escalates consequential AI actions before they reach business systems, with full audit trails per turn. Tenure is sold as a free MIT-licensed Tenure Community self-hosted Docker edition and a paid Tenure Team enterprise Helm chart deployment with OIDC/SSO and SCIM 2.0; the separate GroundEval product turns observed agent behavior into reviewable, enforceable policy.
Product overview
Tenure is a governed context platform for AI agents that provides precise memory management and policy enforcement. The product portfolio consists of the core Tenure platform (proxy-based memory and policy gateway), Tenure Community (free self-hosted tier), Tenure Team (enterprise Kubernetes deployment with OIDC/SCIM), and GroundEval (policy authoring tool). The architecture uses structured beliefs as the unit of memory rather than vector embeddings, with retrieval achieving 1.0 precision versus competitors averaging 0.06-0.22. Core capabilities include belief extraction from conversations, alias-weighted BM25 retrieval with hard scope isolation, runtime gateway enforcement with capability filtering, and audit trails. Tenure ships as a single Docker container (Community) or Helm chart (Teams) with no cloud dependency.
Differentiator
Problem solved
Functional benefit
Brands
- Tenure Community: Free self-hosted version for individual developers and small teams with local deployment
- GroundEval
- Tenure Team
Products and services
- Tenure Governed context platform that provides precise belief-based memory retrieval, runtime policy enforcement, and institutional knowledge injection for AI agents. Sits as a proxy between AI clients (VS Code, Open WebUI, OpenClaw, mobile) and upstream LLM providers (OpenAI, Anthropic, Bedrock, Ollama, LiteLLM), serving engineering teams and individual developers who need persistent project memory, hard scope isolation between projects/teams/users, and a runtime gateway that allows, blocks, or escalates AI actions against business systems before they execute.
- GroundEval Policy evaluation tool that turns observed agent behavior into reviewable, enforceable policy. Implements an observe-draft-review-publish-enforce workflow: observe agent capabilities, draft policy candidates, review invoked permissions, publish to tokens, and enforce on every request via the Tenure runtime gateway.
- PrecisionMemBench Open-source benchmark suite (MIT licensed) for evaluating LLM memory systems on retrieval precision. Contains 89 test cases across static and session evaluations and publishes reproducible comparison results (Tenure 1.00 precision vs. Mem0 0.06, Zep 0.09, Supermemory 0.22, Agentmemory 0.17, GBrain 0.14, Hindsight 0.06). Dataset hosted on HuggingFace.
Quantifiable outcome
- 1.00 mean retrieval precision vs. 0.06 for Mem0, 0.09 for Zep, 0.22 for Supermemory, 0.06 for Hindsight on PrecisionMemBench (arXiv:2605.11325)
- +5 more outcomes
Companies that use Tenure
Customer profileSegments3 records
Ideal customer profiles3 records
Tenure technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
Integration17 records
AI capability6 records
Feature7 records
Tenure partnerships and signals
Strategic signalPartnerships
Four partnerships are on record, tiered core and minor.
- VS Code MarketplacecoreTenure distributes its VS Code extension through the official VS Code Marketplace. The extension (tenureai.tenure) syncs workspace scope on every file switch and provides a beliefs panel in the sidebar. Works with any AI client inside VS Code supporting custom base URLs (Cline, Continue, Windsurf, Cursor).
- OpenClaw / ClawHubcoreTenure is distributed as a native plugin on ClawHub (@tenureai/openclaw-plugin) and npm (@tenureai/openclaw-plugin). OpenClaw users install by saying 'Install Tenure' in chat. Plugin includes automatic per-agent memory isolation for multi-agent deployments.
- Open WebUIcoreTenure acts as a transparent proxy for Open WebUI. Users set Tenure's base URL (http://localhost:5757/v1) as the OpenAI API endpoint in Open WebUI settings. No plugins or workflow changes required; Open WebUI sees standard OpenAI-format responses.
- HuggingFace (PrecisionMemBench Dataset)minorTenure's evaluation dataset is hosted on HuggingFace (tenurehq/precisionmembench). GBrain independently ran the benchmark and published results. The live leaderboard is available on HuggingFace Spaces (tenurehq/precisionmembench).
Scale indicators8 records
Recent moves6 records
Expansion highlights6 records
Tenure competitors and assessment
Company assessmentDirect peers
- Mem0: Mem0 is the most direct competitor in the LLM memory layer, providing persistent, self-improving memory for AI agents and assistants. Tenure explicitly benchmarks against Mem0 (0.06 precision) and operates in the same developer-installable middleware category.
- Zep (getzep.com): Zep is a long-context memory server for AI agents, also published on PrecisionMemBench (0.09 precision). It offers session/thread memory with a similar proxy-style API and targets the same engineering teams integrating LLMs into production.
- Letta (formerly MemGPT): Letta is an open-source framework and cloud service for stateful AI agents with long-term memory. Like Tenure, it targets engineering teams building memory-augmented agents and emphasizes persistent context across sessions.
- Supermemory: Supermemory is an AI memory API service for chatbots and agents, also benchmarked on PrecisionMemBench (0.22 precision). It overlaps directly with Tenure's belief-store memory retrieval use case and consumer/individual-developer pricing tier.
- Hindsight (vector-memory AI): Hindsight is a memory system explicitly compared on Tenure's PrecisionMemBench page (0.06 precision). It targets the same LLM-agent memory use case, making it a directly comparable vendor despite being earlier stage.
Broad incumbents
- OpenAI Memory (ChatGPT / API memory features): OpenAI ships native memory features inside ChatGPT and the Assistants API. As a foundation-model provider, it represents both a partner (Tenure proxies to it) and the most significant competitive risk of memory being absorbed into the upstream model product.
- LangChain (memory modules): LangChain offers memory abstractions inside its broader agent/framework portfolio. While not a focused memory platform, it competes for the same wallet by packaging memory features alongside orchestration, tool use, and model routing.
- Anthropic (Claude project context / Artifacts): Anthropic supports prompt caching and project-level context as platform features. Tenure integrates with Anthropic via prompt caching, but Anthropic also expands native memory-like capabilities that compete for the same governance and persistence budget.
Others
- Pinecone: Pinecone is a managed vector database that underlies many AI memory systems, including competitors Tenure is positioning against. It is an adjacent/alternative infrastructure approach to retrieval — vector similarity versus Tenure's BM25 — rather than a direct substitute product.
- Okta (SCIM / Identity Governance): Okta is a primary identity provider that Tenure integrates with via SCIM 2.0. While not a memory or governance-platform competitor, Okta's Workflows and identity-governance roadmap represents an adjacent player that could absorb AI-agent governance into a broader IAM suite.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat6 records
Key risks6 records
Key highlights7 records
Customer concentration
Tenure social profiles
Digital presenceTenure financial estimates
Financial estimateRevenue estimate
Valuation estimate
Tenure leadership team
Management profileNumber of profiles
Tenure funding detail
Funding detailFunding overview
Funding rounds
Investors
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Tenure M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Tenure
What does Tenure do?
Tenure is a governed context platform for AI agents that operates as a memory proxy sitting between AI clients (VS Code, Open WebUI, OpenClaw, mobile chat) and upstream LLM providers. It extracts structured, typed, scoped, versioned beliefs from conversations and injects the right belief at each turn using alias-weighted BM25 retrieval with hard scope isolation, achieving 1.0 retrieval precision versus 0.06-0.22 for competitor memory systems. A runtime gateway authenticates agents, evaluates preconditions against policy, and allows, blocks, or escalates consequential AI actions before they reach business systems, with full audit trails per turn. Tenure is sold as a free MIT-licensed Tenure Community self-hosted Docker edition and a paid Tenure Team enterprise Helm chart deployment with OIDC/SSO and SCIM 2.0; the separate GroundEval product turns observed agent behavior into reviewable, enforceable policy.
Is Tenure a public or private company?
Tenure is a private company. It is classified as founder individual operated bootstrapped and is currently operating.
When was Tenure founded?
Tenure was founded in 2026. It employs 1 to 10 people.
How does Tenure make money?
Two revenue lines are on record. Tenure Community is the primary driver. The others are tenure Team (Enterprise).
Who are Tenure's main competitors?
Direct peers on record are Mem0, Zep (getzep.com), Letta (formerly MemGPT), Supermemory and Hindsight (vector-memory AI). Broad incumbents are OpenAI Memory (ChatGPT / API memory features), LangChain (memory modules) and Anthropic (Claude project context / Artifacts). Others are Pinecone and Okta (SCIM / Identity Governance).
Does Tenure have an API?
Yes. Tenure exposes an OpenAI-compatible API for proxy routing and belief retrieval. The API serves multiple functions: chat completions proxy (/v1/chat/completions), message endpoints (/v1/messages), model listing (/v1/models), WebSocket belief extraction routes, backup/export-import endpoints (/v1/backup/export, /v1/backup/import), SCIM provisioning endpoint (/scim/v2), admin API for audit trails, and token management. Authentication uses bearer tokens for Personal Access Tokens (PATs) and trusted headers for SSO proxy authentication. PATs are scoped to prevent admin route access. Developer documentation is at tenureai.dev/docs.
What industry is Tenure in?
Tenure's product category is AI Memory Infrastructure. Its primary akta.pro industry code is HDAAACAB, Model Hosting, Serving & Inference Platforms, with a secondary code of BPAEABAJ, Identity & Access Management (IAM) Administration Services. Its NAICS code is 5182 and its SIC code is 7372.