AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Prime Big Deal Days · Oct 6–7Offer from Amazon

Get tech for your team delivered free — and shop member deals

  • Fast, free delivery on millions of items
  • Access to Prime Big Deal Days deals on October 6–7
  • Prime Video, Amazon Music and more included
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

Earendil and the Pi community have released Pi 1.0 alongside Pi Durable, an experimental framework for building long-running agent applications. Pi Durable is designed to persist conversations and tasks across restarts and support different storage and execution environments; the supplied report does not establish production reliability or broad deployment.

Earendil and the Pi community have shipped Pi 1.0 and introduced Pi Durable, an experimental framework for building AI agents intended to run for long periods, resume after failures and operate across different environments. The release matters because it extends the Pi project beyond its existing terminal-based coding agent, while the company says that agent itself is not being replaced.

Earendil describes Pi Durable as a framework for agentic applications, including coding agents, rather than a new version of the Pi coding agent. The existing agent is designed to run on a user’s local or remote machine in a terminal, with one person directing it. Pi Durable is intended for applications that may need to keep conversations running, accept input through different interfaces and allow multiple people to steer the same agent.

The report defines a harness as storage plus the machinery to run conversations with language models, provide tools and supply the environments where those tools operate. Pi Durable includes memory, SQLite and JSONL storage options, along with a conformance suite and benchmarks for developers implementing their own storage backend. Its interfaces are intended to let developers connect other storage systems and execution environments, including environments running remotely from the harness.

Earendil says the project can run wherever a JavaScript runtime is available. The release includes a Node execution environment for tools that need local files or a shell. The company says the SQLite and JSONL storage implementations do not use Node APIs and can be adapted for Bun or Cloudflare Durable Objects. The source report also puts the codebase at about 15,000 lines, excluding tests; that is the company’s description, not an independent measure of maintainability or performance.

At a glance
announcementWhen: Announced alongside Pi 1.0; the source…
The developmentEarendil and the Pi community shipped Pi 1.0 and introduced Pi Durable, an experimental framework for durable, long-running agents.

A Separate Path for Persistent Agents

Pi Durable addresses a design problem that appears when an agent needs to operate beyond a single interactive session: its conversation state and tasks must persist, while its tools may need access to different machines or execution environments. A framework that makes those pieces configurable could give developers more control over where an agent runs and how it connects to storage and tools.

The release also gives Earendil room to test approaches to durable, multi-user agent applications without changing the existing Pi coding agent. The company says lessons from Pi Durable may later inform that agent if they prove useful. For developers, that separation is relevant: Pi Durable is an experimental building block, not a claim that the current Pi coding agent has gained all of these capabilities or that the framework is ready for every production workload.

Amazon

JavaScript persistent storage solutions

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

How Pi Durable Extends Pi

Pi 1.0 is presented by Earendil as a milestone after what it calls extensive hardening, maintenance and active development. The company characterizes the current Pi coding agent as a terminal tool for one person, where a user can inspect what happened if the process stops and ask it to continue. That product focus is unchanged by the Pi Durable announcement.

Pi Durable shares code with the coding agent, including pi-ai, and is intended to carry over Pi’s stated principles of minimalism and malleability. In the report, malleability means developers can adapt the harness, storage and execution environment to different applications. The framework is not limited to coding: Earendil describes it as a base for building varied agentic applications, with coding agents among the possible uses.

“Pi Durable was built specifically for long-running, durable, and malleable agents that can run anywhere.”

— Earendil, in its Pi Durable announcement

Amazon

SQLite database for JavaScript

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Reliability Claims Await Testing

Pi Durable is explicitly described as experimental. The announcement outlines intended behavior, supported storage options and interfaces, but does not provide independent test results, uptime figures, failure-recovery benchmarks or evidence of use at production scale. It is not clear from the supplied report how the framework handles every form of simultaneous access, what operational limits apply, or what safeguards are provided for sensitive agent data.

The announcement also does not specify a calendar release date, licensing terms, a support policy or a product roadmap. Although it describes the framework as suitable for agents that can run anywhere a JavaScript runtime is available, actual compatibility will depend on the runtime, adapters and execution environment used. These details should not be inferred from the broad portability goal.

Amazon

Node.js environment for AI agents

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Community Testing and Development

Earendil is inviting the Pi community to help develop Pi Durable and says it wants to make it a strong durable agent harness. Developers can evaluate the included memory, SQLite and JSONL backends, examine the storage and execution interfaces, and test whether the framework fits their own applications. The announcement does not set out a specific next release date or testing milestone.

The next meaningful evidence will come from implementation details, documented limitations and results from real-world use. Earendil says ideas developed through Pi Durable may flow back into the Pi coding agent if they prove valuable; it has not said which features, if any, will be adopted or when that might happen.

Amazon

JSONL storage tools for developers

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

What is Pi Durable?

Pi Durable is an experimental framework from Earendil for building long-running AI agent applications. It provides a harness for conversations, model calls, tools, storage and execution environments.

Does Pi Durable replace the Pi coding agent?

No. Earendil says the existing terminal-based Pi coding agent is not being replaced. Pi Durable is a separate framework that can also be used to build coding agents.

What storage options are included?

The announcement lists memory, SQLite and JSONL storage. It also describes a small storage interface and a conformance suite and benchmarks for developers building other backends.

Has Pi Durable been proven reliable in production?

The supplied announcement does not provide independent reliability results or production-scale evidence. Earendil calls the framework experimental, so its stated ability to recover from failures should be treated as a design goal pending further testing.

Source: hn

HALLOWEEN

Halloween Picks

As an affiliate, we earn on qualifying purchases.

You May Also Like

Tracking Food Trends: Lahori Zeera Gains Global Attention

GDELT’s media monitoring gave the Lahori Zeera trend an 88/100 signal score as global coverage of the Pakistani drink surges. What is confirmed, and what isn’t.

Best AI Automation Software For Productivity Compared

Compare Zapier and Make for AI productivity automation, including setup, flexibility, maintenance, pricing value, and who should choose each.

TALA Is Open-Source

TALA has announced it is now open-source, allowing broader access and collaboration. Details are still emerging, but this marks a significant shift.

How to Choose AI Automation Software For Productivity

Choose an AI automation tool, connect apps, build a safe workflow, test it, and measure whether it saves time.