ResearchOSby Stratum
Join first 5 labs →
Design partner — 5 labs, CU Boulder

Your lab has a memory problem.

When a postdoc graduates, their institutional knowledge leaves with them. HPC jobs fail overnight and nobody knows until morning. Literature piles up faster than anyone can read it.

ResearchOS is a persistent AI agent that lives in your lab’s Slack, monitors your experiments, and accumulates the institutional knowledge your lab generates — permanently.

Apply for early accessSee how it works ↓

First 3 months free. Then $200–$500/month. No lock-in.

ResearchOS — Heinz Lab — Live Feed
09:42⬡LAMMPS job #4821 completed — 2.3M steps, 14.7h. Carbon nanotube tensile simulation converged.
09:38◈3 new papers on ArXiv matching 'force field + carbon nanotube mechanics' — 1 high relevance flagged.
09:31◆Dr. Chen asks about April surface adsorption params. ResearchOS retrieves: 0.08 M after artifact at 0.1 M (Aug 3 session).
09:15⬡VASP job #5103 stalled at 47% — memory limit on Alpine. Alert sent. Job rescheduled automatically.
08:55◇Weekly digest ready — 7 experiments active, 2 papers under review, 1 HPC anomaly resolved.
——ResearchOS ready_
Currently running inHeinz Lab, CU Boulder
Setup time< 30 minutes
Price$200–500/month
Integrates withSlack + HPC clusters
The problem

Research labs lose knowledge three ways. ResearchOS stops all three.

01

Knowledge walks out the door

A postdoc leaves after three years. Their calibration notes, failed-experiment reasoning, and institutional know-how leave with them. The next student starts over.

02

HPC jobs fail silently

Simulation jobs run overnight and die at 3am. Nobody knows until morning. Two days of compute time, gone. You find out when you manually check the queue.

03

Literature is a second job

Your field spans three arXiv categories. Staying current takes hours per week. You read papers. Then you forget what you read and why it mattered.

How it works

Setup once. It grows from there.

1

Connect Slack

Add ResearchOS to your lab's Slack workspace. Takes 20 minutes. No engineering required.

We handle the infrastructure. You get a @researchos bot in your workspace.

2

Seed with context

Share your lab's protocols, published papers, and shared docs. ResearchOS reads, indexes, and begins building your lab's knowledge base.

Secure. Your data stays in your dedicated instance.

3

Connect your cluster

Optional: point ResearchOS at your HPC cluster (SLURM, PBS). It monitors job status and alerts you in Slack.

Alpine, Summit, your department's cluster — any SLURM-compatible system.

4

Ask it anything

Lab members @-mention ResearchOS in Slack. It answers using your lab's specific knowledge, not the internet.

Every answer adds to the memory. Every conversation makes it smarter.

Use cases

What labs actually use it for.

Institutional memory
@lab@researchos what concentration did we use for the June surface adsorption run, and why did we change it?
@rosAug 3 session: moved from 0.1 M → 0.08 M after artifact at higher concentration. See notes from Dr. Chen + Prof. Heinz Slack thread. Full context in lab log #4421.
HPC monitoring
@labLAMMPS job #4821 has been running 14.2 hours on Alpine. Expected completion: 2–3 hours.
@rosJob completed at 14.7h. 2.3M steps converged. Output files in /scratch/heinz-lab/cnt-tensile-04/. Archiving to shared drive now.
Literature alerts
@lab3 new papers on carbon nanotube mechanics (ArXiv, ACS Nano) since Monday. 1 flagged high relevance.
@ros"Machine-learned force fields for single-walled CNT under biaxial strain" — methods overlap with your Q2 tensile series. Recommend review before next group meeting.

Also used for

◆Grant section drafts from lab activity data
◆New graduate student onboarding
◆Weekly PI digest — no status reports needed
◆Protocol documentation as experiments run
◆Cross-departmental lab coordination
◆Conference paper methods reconstruction
◆Equipment booking and maintenance logs
◆Literature synthesis across 3+ domains

“The agent knows our calibration history going back eighteen months. When a new student asks why we moved from 0.1 M to 0.08 M in August, it pulls up the exact reasoning from the Slack thread where that decision was made. That context used to leave with the postdoc.”

H

Prof. Hendrik Heinz

PI, Heinz Lab — CU Boulder (CHBE / MSE)

Pricing

Less than a part-time RA. Available 24/7.

Design partner pricing: first 3 months free for the first 5 CU Boulder labs. Standard pricing kicks in at month 4. Cancel anytime.

Lab

$200/month

Up to 10 lab members

  • ✓1 ResearchOS agent
  • ✓Slack integration
  • ✓HPC job monitoring
  • ✓Literature tracking (3 domains)
  • ✓6-month memory retention
  • ✓Lab onboarding session
Start for free
Most common

Research Group

$350/month

Up to 25 members

  • ✓Everything in Lab
  • ✓Equipment scheduling (LabLink)
  • ✓Cross-project context
  • ✓18-month memory retention
  • ✓Priority support
  • ✓Monthly PI digest
Start for free

Multi-Lab

$500/month

Up to 3 labs

  • ✓Everything in Research Group
  • ✓Cross-lab knowledge sharing
  • ✓Centralized PI dashboard
  • ✓36-month retention
  • ✓Custom integrations
  • ✓Dedicated onboarding
Talk to us

Academic pricing adjustments available for NSF/NIH-constrained budgets. SLURM integration, equipment scheduling (LabLink), and custom domain domains available on all plans. Your data lives on dedicated infrastructure — never shared across labs.

Design partner program — CU Boulder cohort

Join the first 5 labs. Shape the product.

The first cohort gets 3 months free, direct access to the founding team, and shapes what ResearchOS becomes. Slots are limited to 5 CU Boulder labs. If your lab runs more experiments than it can track, this is for you.

Or email sean@onstratum.com directly — one paragraph about your lab and your biggest knowledge management problem.