What Engram does

Engram is a San Francisco-based AI startup founded in 2025 that builds context-learning AI models designed to internalize and continuously study an organization's proprietary data — including GitHub, Slack, Notion, and document repositories — rather than relying on public internet training data. Its core technology is a proprietary continual-learning training pipeline that the company describes as spending "the equivalent of hundreds of years" of compute studying a single customer's context, currently retraining daily with a stated roadmap to hourly and then minute-level updates. The product surface centers on an API for agents that learn on very large shared knowledge workspaces; the company claims 10x to 100x token efficiency versus traditional context-gathering approaches because its models no longer need to re-read documents at inference time. Go-to-market is API-first and partnership-led, with three disclosed enterprise design-partner relationships: Notion (Custom Agents for large Notion workspaces), Harvey (models that internalize law-firm knowledge for precedent search), and Microsoft (a pilot inside M365 for cost-efficient, customized enterprise agents). Revenue mechanics are usage-based API pricing, though public pricing is not disclosed; the company is venture-backed and pre-commercial at scale, having raised $98M from General Catalyst, Kleiner Perkins, and Sequoia alongside a syndicate that includes Factory, Modern, Amplify, Neo, and SV Angel, with individual advisors Andrej Karpathy, Pieter Abbeel, and Assaf Rappaport.

Name

Engram

Legal name

Engram

Website

https://engram.com

Company type

Private

Founded year

2025

Operating status

Operating

Headcount range

11–50 employees

Short description

Engram builds continually learning AI models that internalize enterprise context from tools like GitHub, Slack, Notion, and Microsoft 365, delivering an API for agents with 10x–100x token efficiency for enterprise customers.

Ownership category

akta.pro rank

Product category

Enterprise AI Infrastructure

NAICS

Software Publishers (5132), Custom Computer Programming Services (541511), Computer Systems Design Services (541512)

SIC

Services-Computer Programming, Data Processing, Etc. (7370), Services-Computer Integrated Systems Design (7373)

akta.pro primary industry

Agents & Autonomous Workflows (Tool Use, Planning, Multi-Agent) (HDAAACAF)

akta.pro secondary industries

Enterprise Foundation Model Integration & APIs (Connectors, Governance, Deployment) (HDAAACAO), Generative AI & LLM Solutions Services (RAG, Agents, Copilots) (BPAEAHAG), Enterprise AI Data & Knowledge Platforms (Vector Databases, Knowledge Graphs) (HDAEANAH)

Keywords

• Contextual AI models
• Continual learning AI
• Enterprise AI memory
• AI agent API
• Knowledge base internalization

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Personnel, Infrastructure, Operations, Marketing or Sales

Revenue model

1. API Access for Agents: First product is an API for agents that learn on very large shared knowledge workspaces. Revenue generated through API usage and enterprise licensing arrangements with partners.

Go-to-market motion1 record

Distribution channels2 records

Marketing channels3 records

Core offering

Engram builds context-learning AI models and provides an API that lets enterprise AI agents internalize and continually learn from large organizational knowledge workspaces such as GitHub, Slack, Notion, and Microsoft M365. Instead of re-reading documents per session, Engram's models absorb company data through daily retraining, enabling token-efficient, deeply contextual agents for productivity, legal, and enterprise software use cases.

Product overview

Engram offers an AI platform centered on its API product for agents that learn from user context and organizational knowledge. The company builds models that study and internalize context from sources like GitHub, Slack, and Notion, achieving 10x-100x token efficiency compared to traditional approaches. Through strategic partnerships, Engram delivers specialized solutions: Custom Agents for Notion workspaces, specialized models for Harvey's legal use cases, and M365-integrated models for Microsoft enterprise customers. The platform focuses on continual learning where models get better over time through daily (moving to hourly, then minute-by-minute) retraining on organizational data.

Differentiator

Problem solved

Functional benefit

Products and services

• Engram API for Agents A developer API for building AI agents that learn on very large shared knowledge workspaces. The API enables models to internalize organizational context from sources such as GitHub, Slack, and Notion, delivering token-efficient agents that understand specific domains without repeated context retrieval. Targeted at enterprise software developers and platform teams.

Quantifiable outcome

• 10x or even 100x more token-efficient than traditional context-gathering approaches
• +1 more outcomes

Named customers3 records

Segments2 records

Ideal customer profiles1 record

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

AI capability4 records

Feature4 records

Partnerships

Three partnerships are on record, tiered core.

• NotioncoreStrategic or Co-development PartnerDesign partnership with Notion to build Custom Agents that understand large Notion workspaces. This is one of Engram's earliest and most important partnerships, demonstrating the company's contextual AI capabilities within productivity tools.
• HarveycoreStrategic or Co-development PartnerDeveloping models that internalize the knowledge of an entire law firm and can search and find precedents across many client matters. Enables sophisticated legal AI applications with deep institutional knowledge.
• MicrosoftcoreStrategic or Co-development PartnerPiloting Engram models inside Microsoft 365 to deliver cost-efficient, customized agents for enterprise customers. Represents a major distribution opportunity through Microsoft's enterprise customer base.

Scale indicators1 record

Recent moves6 records

Expansion highlights6 records

Direct peers

• Mem0: Mem0 builds a persistent memory layer for LLM applications and AI agents, allowing models to remember, learn, and update user preferences over time. It is the closest direct competitor to Engram, pursuing a very similar memory-as-a-service thesis for production agent deployments.
• Zep: Zep offers a long-term memory store and context-assembly API for AI assistants and agents, focused on reducing token cost and improving personalization. It competes directly with Engram's API-first enterprise memory product on both cost-efficiency and retrieval-quality dimensions.
• Letta: Letta (formerly ChatGPT Memory / Zep's sibling product line) provides a memory service and framework for stateful AI agents with long-term memory and continual learning. It targets the same enterprise agent memory problem that Engram addresses, with overlapping developer-tool positioning.

Emerging players

• Cognee: Cognee builds an open-source knowledge layer for LLMs using knowledge graphs and structured memory to ground agent responses in enterprise data. It overlaps with Engram's enterprise-context-internalization thesis but emphasizes graph-based retrieval rather than continual training.
• LlamaIndex: LlamaIndex provides a data framework and orchestration layer for connecting LLMs to enterprise documents and long-running context. It is a partial competitor to Engram's enterprise-context product, with overlapping agent-memory and retrieval use cases.

Others

• Chroma: Chroma develops the open-source vector database most commonly used to power RAG and agent memory in production AI systems. It is enabling infrastructure for the broader memory-layer category in which Engram competes.
• Pinecone: Pinecone provides a managed vector database that underpins many enterprise memory and RAG systems. It is adjacent infrastructure that Engram's enterprise customers may use alongside or instead of Engram's training-based context layer.

Broad incumbents

• LangChain: LangChain offers a broad agent orchestration framework and LangGraph platform for building production AI agents, including memory modules that overlap with Engram's API. It is a wider incumbent in the enterprise agent stack rather than a focused memory-layer specialist.
• OpenAI: OpenAI ships long-term memory and custom GPT features inside ChatGPT and its API, addressing the same enterprise-context problem that Engram targets. As a frontier-model incumbent, it represents both a partner ecosystem and the most significant competitive threat to a separate memory layer.
• Anthropic: Anthropic's Claude offers long-context windows and project-level memory features for enterprise customers, addressing the continual-context problem from within a foundation model rather than as a separate layer. It competes for the same enterprise memory budget as Engram.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/EngramLab
• LinkedInlinkedin.com/company/engram-lab

Revenue estimate

Valuation estimate

Number of profiles

Profiles1 record

Funding overview

Funding rounds1 record

Investors8 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Engram