# SEATECHONE

> SEATECHONE is a Seattle product studio that designs, builds, and operates B2B SaaS and AI software.

## Company and services

- [SEATECHONE](https://seatechone.com/): SEATECHONE builds B2B SaaS and AI products, from architecture and design to launch. Explore SEAREI, GuardMeter, and a clear path to your next release.
- [B2B Software & AI Engineering Services](https://seatechone.com/services/): Explore SEATECHONE services for B2B SaaS development, AI integration, and AI guardrail evaluation. Define the workflow, scope, and delivery together.
- [Selected Product Work](https://seatechone.com/work/): Explore SEAREI and GuardMeter, products built and operated by SEATECHONE. See the workflows, engineering, and evidence behind each product.
- [Start a Software Project](https://seatechone.com/contact/): Tell SEATECHONE about your product, business workflow, or AI evaluation needs. Share your starting point and discuss a clearly scoped engagement.
- [B2B SaaS Development](https://seatechone.com/services/saas-development/): SEATECHONE builds B2B SaaS products, customer portals, and operational software. Explore product scope, architecture, delivery, and handoff.
- [AI Integration for Business Software](https://seatechone.com/services/ai-integration/): Add AI to business software with defined inputs, review workflows, evaluation, and operational controls. SEATECHONE designs and builds the surrounding product.
- [AI Agent Evaluation & Guardrail Testing](https://seatechone.com/services/ai-evaluation/): Evaluate AI agents and guardrails with GuardMeter: attack-testing support for 24 languages, CC BY 4.0 datasets, evidence reports, and release checks.
- [SEAREI Product Portfolio](https://seatechone.com/work/searei/): SEAREI is a SEATECHONE-owned real estate software product connecting virtual staging, disclosure documentation, payments, delivery, and public verification.
- [GuardMeter: AI Agent Evaluation & Guardrail Testing](https://seatechone.com/work/guardmeter/): GuardMeter pairs AI agent evaluation with attack-testing support for 24 languages, CC BY 4.0 attack datasets, an MIT-licensed toolkit, and release checks.
- [About the SEATECHONE Product Studio](https://seatechone.com/studio/): SEATECHONE is a Seattle product studio building and operating SEAREI and GuardMeter, and developing B2B SaaS and AI software for founders and operating teams.
- [Privacy notice](https://seatechone.com/privacy/): How SEATECHONE handles website inquiries, contact form submissions, hosting, and links to other services.

## Products and evidence

- [GuardMeter](https://seatechone.com/guardmeter/): Evaluation service, pricing, and sample evidence.
- [GuardMeter sample report](https://seatechone.com/guardmeter/sample-report.html): A published sample evaluation, including a failed gate.
- [GuardMeter source](https://github.com/samvardani/guardmeter): Open-source software.
- [SEAREI](https://searei.com/): Real estate software built and operated by SEATECHONE.

## GuardMeter languages, open datasets, integration, and evidence

- [GuardMeter GitHub Action](https://github.com/marketplace/actions/guardmeter): Add evaluation checks to GitHub Actions.
- [GuardMeter endpoint integration guide](https://github.com/samvardani/guardmeter/blob/main/docs/OPOD_INTEGRATION.md): OpenAI-compatible endpoint, scenario, and rollout-hook integration.
- [GuardMeter language registry](https://github.com/samvardani/guardmeter/blob/main/guardmeter/core/languages.py): Tooling support for 24 attack-testing languages; separate from reviewed dataset coverage.
- [Multilingual attack dataset coverage](https://github.com/samvardani/guardmeter/blob/main/dataset/agentic/v2/DATASET_CARD.md): Current authoring and native-review status.
- [Attack dataset CC BY 4.0 license](https://github.com/samvardani/guardmeter/blob/main/dataset/agentic/v2/LICENSE): Commercial reuse and adaptation permitted with attribution, a license link, and changes identified.
- [GuardMeter toolkit MIT license](https://github.com/samvardani/guardmeter/blob/main/LICENSE): The evaluation software uses a separate MIT license.
- [Independent Opod behavior evaluation](https://seatechone.com/guardmeter/reports/GuardMeter-Opod-Behaviour-Report.pdf): Public field report PDF.

Opod: 13 of 30 scenarios changed. Two models on the same OpenAI-compatible endpoint passed a health-check completion, but differed on injected tool output, JSON format, and latency. The changed outcomes included both improvements and regressions. Independent evaluation on our own Opod instance; no endorsement is implied.

## Factual summary

- SEAREI and GuardMeter are its own products. Portfolio descriptions of these products are not claims about client outcomes.
- Client work is scoped around agreed requirements, deliverables, acceptance criteria, and a written project agreement.
- GuardMeter provides open-source AI agent behavior evaluation and guardrail testing, plus a separately priced evaluation service. Scenarios inspect endpoint responses, requested tool calls, injected tool output, refusals, format, and latency. Evaluation results apply to the tested data and policy; they do not certify universal safety or legal compliance.
- GuardMeter supports attack testing in 24 languages through language-aware tooling, six cross-lingual attack families, and per-language release gates. Supported language entries are distinct from authored and native-reviewed dataset coverage; consult the current dataset card for coverage status.
- The GuardMeter evaluation toolkit is MIT-licensed. Its attack datasets are separately licensed under CC BY 4.0, permitting commercial reuse and adaptation with attribution, a license link, and changes identified. The paid service provides tailored testing, human review, and recommendations within an agreed scope.

Contact: hello@seatechone.com

The linked HTML pages are the primary public descriptions. [Extended text reference](https://seatechone.com/llms-full.txt) is an optional convenience copy.