September 1, 2026

Workflow & Agents|Index 05

Almanac Unveils AI Agent with Persistent Company Memory

A new AI agent builds personal and company-specific knowledge bases from connected accounts, enabling proactive task management and long-horizon workflows without losing context.

Via
AITECH TOKYO Editors
Dateline
August 31, 2026
Date
August 31, 2026
Time
6 min read
Almanac Unveils AI Agent with Persistent Company Memory

Tagline

AI agent builds persistent knowledge for proactive tasks.

Who & Why

For a Tokyo-based project manager overseeing complex, multi-stakeholder projects who needs an AI to proactively manage follow-ups and draft communications based on accumulated company knowledge.

vs. Existing

Almanac differs from standard LLM chat interfaces like ChatGPT or Notion AI by building a persistent, pre-compiled knowledge base from connected accounts, allowing for long-horizon task management and proactive suggestions beyond single-session queries.

Tokyo Take

While promising for context-aware automation, Almanac's immediate utility for Tokyo professionals hinges on robust Japanese language processing for its "wiki" creation and integration with local SaaS tools. Without transparent JPY pricing and demonstrated compatibility with common Japanese business applications, its adoption will likely be slow, initially appealing to global-minded startups.

Almanac is an AI agent designed to act as a company-specific “brain,” integrating with various business and personal accounts to build persistent knowledge bases. It aims to overcome the common challenge of creating context-aware AI assistants that retain information across sessions and tasks.

The system connects to services like Gmail, Calendar, and other business tools, then organizes this incoming information into two distinct wikis: a personal one for individual context and a company-wide wiki detailing projects, roadmaps, and blockers. This foundational knowledge layer is key to its functionality.

Its creators, a team of three IIT and Harvard alumni, emphasize their unique approach to memory and context. They assert that most AI assistants treat memory as an afterthought, while Almanac invests significant compute upfront in “pre-compiling” this knowledge base.

This pre-compiled knowledge enables two primary capabilities. First, a proactive agent that can identify and suggest tasks for automation. Users may wake to notifications like:

> "I already prepared a draft of your fundraising pitch deck, want to take a look?"

Second, Almanac supports long-horizon tasks that span days or involve human dependencies. Unlike session-bound agents that forget context after completion, Almanac can track ongoing projects, follow up on emails, or chase documents over extended periods, remembering the full thread.

Real-world applications observed by the founders include managing dog-rescue operations, researching complex strategies, and building marketing campaigns without repeatedly re-explaining business context. The agent's persistent memory allows for continuity across diverse workflows.

Regarding privacy, Almanac states it only accesses explicitly connected accounts. OAuth credentials are held by a third-party connection provider, and Almanac itself stores only the wiki content and the source citations, not raw account data. Specific pricing plans are not detailed in the announcement, though a seven-day trial is offered.

The service is currently live and positioned for companies seeking an agent capable of executing complex, multi-stage tasks.

The Briefing

World AI tech, read from Tokyo. Once a week, in Japanese.

Each Friday: the five global AI tech stories Japanese business professionals should know about this week, translated and read through a Tokyo lens — what it means for Japan, what to act on, what to keep watching.

We respect your inbox. Unsubscribe anytime.