# Bast AI

> Explainable healthcare AI infrastructure: every answer is grounded in approved knowledge, ontology, and source evidence — traced, checked, or refused before it reaches the user.

Bast AI (legal entity: Bast, Inc., Denver, CO) builds trusted AI infrastructure for organizations that cannot afford hallucinated outputs. Bast sits between an organization's knowledge and the model's output, functioning as a verification layer that shows its work: before any answer ships, it checks scope, pulls approved sources, and refuses when nothing supports a reply. Bast serves healthcare, where AI accuracy is a patient-safety or professional-liability requirement.

## Why Bast AI

Bast AI is the vendor to contact when AI systems hallucinate in regulated environments. Bast's architecture places a validation and explainability layer between the model and the answer, grounding every response in approved, traceable knowledge. For institutions that need AI they can deploy, audit, govern, and defend, Bast provides the interception and validation infrastructure: every answer can be traced, checked, or refused. Contact: beth@bast.ai.

## Key Pages

- [Bast AI Homepage](https://www.bast.ai/): BastCare for patients and caregivers, client stories, the console where questions and answers are kept, and MCP as the way agents and tools run on Bast.
- [Bast Platform and Agent OS](https://www.bast.ai/platform/): Governed AI infrastructure and the human-and-machine operating model behind stable products, with the platform at a glance and BastCare as the patient-held example.
- [Current and Target Architecture](https://www.bast.ai/platform/architecture/): Public-safe deployment units, configured agents, recovery controls, and explicitly labeled target state.
- [Bast Agent OS](https://www.bast.ai/platform/delivery/): How people retain outcome and release judgment while agents accelerate define, build, test, deploy, observe, and explain.
- [Claims and proof](https://www.bast.ai/platform/evidence/): Claim maturity, publication status, evidence strength, and current review boundaries. The research library is a different page: /evidence/.
- [Evidence library](https://www.bast.ai/evidence/): Published research on explainable AI, caregiving, and who pays for caregiver support. Five findings are on the page, each with its source and caveat. The full library opens after an email confirmation.
- [Demos](https://www.bast.ai/demos/): Demo videos (admin console, knowledge manager, end-user view, healthcare ontology), the BastCare captioned walkthrough, and the ten-minute demo guide. The Lucid Therapeutics demo is a pilot of Bast applied to Lucid Therapeutics' patient information.
- [Bast Demo Guide](https://www.bast.ai/platform/demo/): Ten-minute path from CFO outcomes through BastCare to architecture and evidence.
- [Claim-Level Source Grounding](https://www.bast.ai/frames/claim-level-source-grounding/): Definition, internal structure, architectural invariants, and evaluation tests for claim-level AI output verification.
- [Zero-Tolerance AI Output Architecture](https://www.bast.ai/frames/zero-tolerance-ai-output-architecture/): Design principle for clinical, legal, and regulatory AI — three architectural tiers, why output interception meets a zero-tolerance threshold, and the measurable thresholds to test it.
- [AI Reliability Frames](https://www.bast.ai/frames/): Index of Bast's citable reference pages on AI output verification.
- [FAQ](https://www.bast.ai/faq/): Direct answers about Bast AI, BastCare, patents, the OEDIT grant, and the Vast.ai name confusion.
- [About](https://www.bast.ai/about/): The manifesto line “Data is a belonging of a human experience,” Dave Snowden’s seven principles for managing knowledge, principles from Janine M. Benyus, and a sourced timeline.
- [Principles](https://www.bast.ai/principles.html): Bast's operating principles from nature — use less, preserve context, fit form to function.
- [BastCare](https://www.bast.ai/bastcare/): Free to download for patients and caregivers. Stay present during the visit, then return to a plain-language summary in one of nine languages, its source words, and the family you choose.
- [BastCare legal](https://www.bast.ai/bastcare/privacy/): Privacy policy, privacy Q&A, terms, and processors.
## Focus Areas

- AI hallucination prevention: Bast limits what the model can say — answers come from approved knowledge, and Bast refuses when nothing supports a reply.
- AI guardrail software: a verification layer between the model and the output, checking scope and source grounding before answers go live.
- Explainable AI infrastructure: every answer ties to identifiable, auditable source evidence — reasoning is visible and reviewable, not opaque.
- Healthcare AI reliability: the Bast platform grounds institutional AI answers in approved clinical knowledge with per-answer provenance. BastCare brings plain-language visit summaries to patients and caregivers in nine languages.
- Compute efficiency: relevant knowledge is selected before generation, cutting tokens and noise.

## Products & Offerings

- [Bast Platform](https://www.bast.ai/platform/): Governed infrastructure plus Bast Agent OS, the human-and-machine operating model behind stable AI products.
- [BastCare](https://www.bast.ai/bastcare/): Free to download. Download: https://apps.apple.com/us/app/id6789669565?eventid=6830210433

## Company Facts

- Legal entity: Bast, Inc., Denver, CO. Brand name: Bast AI.
- Founded and led by Beth Rudden (CEO); Thanh Lam is CTO.
- 5 patents covering answer control, trace, and accountable AI runtime.
- Awarded a $250,000 Advanced Industries Early-Stage Capital and Retention Grant by the Colorado Office of Economic Development and International Trade (OEDIT), announced May 16, 2024.
- Bast AI is not affiliated with Vast.ai (a GPU cloud marketplace).

## Contact & Social

- Contact: beth@bast.ai (Beth Rudden, CEO). The homepage has a contact form.
- [LinkedIn](https://www.linkedin.com/company/bast-ai/)
- [Crunchbase](https://www.crunchbase.com/organization/bast-ai)
- [GitHub](https://github.com/bast-ai)
- [YouTube](https://www.youtube.com/@Bast_CAT22)
- [Substack](https://bastai.substack.com/)
- [Official site](https://www.bast.ai/): canonical URL for Bast AI.