I

Idler

AI agents automate back-office workflows

Software Engineer

Full-TimeUpdated on 10/23/2026
$125k - $400k/yr+ Meaningful equity + Relocation assistance
Entry, Junior, Mid
San Francisco, CA, USA
In PersonIn-person role in San Francisco.
Software EngineeringBackend Engineering

About the job

Requirements
  • Have 1–10 years of experience writing excellent production software; new graduates are welcome.
  • Have experience with full-stack applications in a popular programming language.
  • Create robust systems under time pressure through meticulous problem solving.
Responsibilities
  • Design and build scalable systems that generate reinforcement learning environments.
  • Create agentic quality-assurance systems to validate environment quality and fairness.
  • Work directly with AI researchers at leading labs to understand what makes training data effective.
  • Handle petabyte-scale extract, transform, and load processes and orchestrate millions of concurrent jobs.
  • Support new product lines as the company expands beyond coding environments.
  • Manage the company's security posture and compliance requirements.

About the company

Idler provides AI agents that automate back-office and operational tasks for businesses. Its platform combines document parsing, decision logic, and agent-driven execution to manage end-to-end workflows such as order fulfillment, payment collection, and service scheduling, including PO processing and invoicing. It configures automations that update trackers, send emails, and coordinate with vendors and clients, with human intervention only when needed. It starts with a four-week pilot to map workflows, set up and tune automations, and measure impact, followed by a tailored plan to scale. The goal is to reduce manual work and boost efficiency for organizations that rely heavily on spreadsheets, emails, and vendor portals.

Company Size

11-50

Company Stage

Seed

Total Funding

$9.1M

Headquarters

San Francisco, California

Founded

2025

Get referred to Idler

See people who can refer or advise you

Simplify Jobs

Simplify's Take

What believers are saying

  • Paradigm led Idler's $9 million seed on August 20, 2026.
  • Post-training Nemotron 3 Nano improved Finance Agent Benchmark from 30.0% to 41.8%.
  • YC lists a twelve-person team and an active San Francisco company on September 7, 2026.

What critics are saying

  • Idler's credibility depends on unnamed frontier-lab customers; none are publicly disclosed.
  • Selling environments to frontier labs limits demand to a handful of buyers worldwide.
  • If frontier labs build internal environments, Idler loses its only market.

What makes Idler unique

  • August 19, 2026 launch: proprietary operating-data benchmarks, not generic chatbot wrappers.
  • ShelfLife uses a live retailer's complete operations across nine stores and three platforms.
  • YC S25 founders bring Dark Forest, Microsoft, PayMongo, and Census Bureau experience.

Help us improve and share your feedback! Did you find this helpful?

Benefits

  • Health Insurance
  • Meal Benefits
  • Relocation Assistance
  • Paid Vacation
  • Company Social Events

Company News

The Adventurer
Aug 20th, 2026
idler raises $9M led by Paradigm to sell evals and RL environments to frontier AI labs

idler launched on 19 August as a frontier data research lab with a $9 million seed round led by Paradigm, alongside Y Combinator, Long Journey VC and angels. The company sells evaluations, benchmarks and reinforcement learning environments to labs training frontier models. The timing suggests a real market. Z.ai described synthesizing RL environments end to end for GLM-5.3, and Harvey built LAB environments before post-training Tenet — both announcements made the same week. Paradigm's involvement marks a notable rotation of crypto capital into AI infrastructure. The business model targets perhaps 15 organisations worldwide, making it high-margin but low-volume. The announcement contains no product details, methodology or named customers. For a company whose credibility depends on client names, this absence is significant. The company also does not address the independence question raised by grading systems its own customers are measured against.