# Talaria, the research harness

> For computational scientists whose week goes to validation instead of science.

Status: Private beta today; Open source December 5; Hosted alpha December 5, invite only.

Ideas are what moves science forward, but validation is expensive and slow: data wrangling, sysadmin, re-implementing baselines, manual checking. Talaria is a multi-agent, multi-foundation-model harness that runs that machinery. You direct it; it handles the coding, the environments, the cluster runs, and the record keeping, with the scientist in the loop and HPC as the guardrail. A mech suit for the researcher, not an auto-scientist that replaces you.

[Request Talaria access](https://talariasci.com/signup) / [Read the architecture](https://talariasci.com/blog/the-talaria-architecture) / [Contact us](https://talariasci.com/contact)

![Figure 1. The Talaria architecture. The five-stage Research Loop runs on top of the System of Record; below, the rigor skills augment that process and can extend the record, the inspection skills give you a view into the record at various levels, and the maintenance skills ensure the record is always clean. Capture is cheap, every claim-hardening write passes a gate that checks it against the record's rules, so every hardened claim stays traceable and inspectable, and the adversarial panel rides the gates as claims harden; a failed stage kicks back to a previous part of the loop. The record compounds over time as the process unfolds.](https://talariasci.com/images/blog/the-talaria-architecture/fig1-architecture.png)

## How you would use it

A session is a loop. You bring an idea; the harness hardens it into a pre-registered plan, builds and reviews the code, runs it on your compute, and proposes findings grounded in the results. Each stage is a skill you can read and edit.

1. **/idea** captures the thought with its provenance. Capture is cheap; nothing is gated here.
2. **/plan** freezes it into a pre-registered hypothesis: configurations, seeds, metrics, and what it takes to validate or refute it.
3. **/implement** builds and reviews the experiment code.
4. **/run** dispatches the work to your compute: a university cluster, a national-lab allocation, a cloud, or your laptop, one config block apart.
5. **/synthesize** proposes findings from the results. A panel of isolated reviewers attacks them before they register, and the findings seed the next idea.

![Figure 2. Who owns what. The harness you download (the core, a field's packs, the skills, the docs) is open and we maintain it; your instance (the record and your experiment code) is private and yours; the published literature is the commons.](https://talariasci.com/images/blog/the-talaria-architecture/fig3-property-line.png)

## What you get

- A system of record: every result traces to the commit, config, seeds, environment, and cost that produced it, so an audit is a lookup and a reproduction is one command.
- Rigor that scales with the claim: a what-if costs a conversation; only the claims you want to publish pass the full depth of review and reproduction.
- Your models: a frontier coding agent drives it, the review panels fan out across the models you choose, and the record compounds whichever driver or model you swap in.
- Twenty skills in five groups: the research loop, extra rigor, record inspection, maintenance, and the harness's own upkeep.

## Who it is for

- Computational physicists, applied mathematicians, and simulation-driven engineers whose experiments run on solvers and clusters: the researcher whose domain expertise no model has, and whose week still goes to work no scientist should do by hand.
- We start with PDEs, optimization, and the compute-heavy corners of applied math, on problems that validate on a single GPU first. The open core is free for everyone; the private beta starts with working research groups.

## What stays yours

- Your instance, its record, and the models built for you are yours.
- The open core runs where your data lives and sends no telemetry.
- The hosted alpha keeps your experiment records on our infrastructure; the enterprise deployment runs in your cloud, with the data retention you set.

## How to get it

- Open core, December 5: the harness from the terminal, under Apache 2.0, on your laptop, your cluster, or a cloud, driven by the frontier coding agent you bring a key for, free.
- Hosted, December 5, an invite-only alpha: sessions in a managed sandbox, the record and its DAG view, the frontier models we supply, and your own compute through the credentials you upload. The sandbox runs on our infrastructure and keeps your experiment records there; the models built for you are yours. Teams, a pack registry, and compute we provision for you follow in the beta. Includes TauOpt hosted. The open core is free for everyone; hosted compute and models are priced at cost for academic work and cost-plus otherwise.
- Enterprise, coming soon: the platform in your cloud with single sign-on, role-based access, audit logs, and the data retention you set, plus the people to run it.

Read: [The architecture of the Talaria research harness](https://talariasci.com/blog/the-talaria-architecture), [A system of record for computational science](https://talariasci.com/blog/the-system-of-record), [Why I'm building Talaria Scientific](https://talariasci.com/blog/why-im-building-talaria)

## Get access, or get in touch

Talaria is in private beta with a small group of researchers, and TauOpt is in closed beta. For product access, use the request forms. For enterprise, partnerships, or press, use the contact page. [Request Talaria access](https://talariasci.com/signup) / [Request TauOpt access](https://tauopt.com/#request-access) / [Contact us](https://talariasci.com/contact)

Read: [Talaria: the architecture](https://talariasci.com/blog/the-talaria-architecture), [Talaria: the blog](https://talariasci.com/blog), [TauOpt: documentation](https://tauopt.com/docs/)

---

Source: https://talariasci.com/talaria