> ## Documentation Index
> Fetch the complete documentation index at: https://docs.flowx.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Research

> FlowX.AI public research: the border open-source guardrails library, open models on Hugging Face, and the research paper series.

FlowX.AI publishes research as open tools, open models, and papers. This page indexes what is public; everything below links to its original home.

<Info>
  These are research artifacts and open-source tools published by FlowX.AI, not shipped platform features. For platform capabilities, see [AI in FlowX](./ai-in-flowx).
</Info>

***

## border: open-source LLM guardrails

[border](https://border.flowx.ai) is an open-source Python library (Apache-2.0) that checks LLM inputs and outputs and signs an evidence record for each check. Evidence records hold hashes, the resolved policy hash, and the model revision that produced each finding, never the user's text. It runs on CPU with the network interface down, and every classifier is scored in 26 languages.

```bash theme={"dark"}
pip install flowx-border
```

<CardGroup cols={3}>
  <Card title="border.flowx.ai" icon="shield-check" href="https://border.flowx.ai">
    Project site and documentation
  </Card>

  <Card title="GitHub" icon="github" href="https://github.com/flowx-ai/border">
    Source, Apache-2.0
  </Card>

  <Card title="Benchmarks" icon="gauge-high" href="https://border.flowx.ai/benchmarks">
    Published latency and accuracy numbers
  </Card>
</CardGroup>

***

## Open models on Hugging Face

The [flowxai organization on Hugging Face](https://huggingface.co/flowxai) publishes 50 Apache-2.0 models focused on on-device AI for regulated industries: PII detection (`piiguard`, `cee-pii`), scam and fraud classification (`scam-guard`), content moderation, and regulated-advice detection, alongside two public benchmarks (`cee-pii-bench`, `scamguardbench`).

<Card title="huggingface.co/flowxai" icon="face-smile" href="https://huggingface.co/flowxai">
  All models and benchmark datasets
</Card>

***

## Paper series

Seven technical papers on enterprise AI agents, published at [flowx.ai/research](https://www.flowx.ai/research):

| Paper | Scope |
| - | - |
| **VERA** | ROI measurement for production agents |
| **ORNA** | Agent self-adaptation |
| **HALO** | Hallucination prevention |
| **GAVEL** | Runtime governance and audit enforcement |
| **SIFT** | Self-teaching document classification |
| **RAILS** | Deterministic agent workflows |
| **MNEMĒ** | Structured memory with provenance tracking |

<Card title="flowx.ai/research" icon="book-open" href="https://www.flowx.ai/research">
  The paper series, with arXiv links where available
</Card>

***

## Browser-agent reliability benchmark

[Browser Agents Don't Fail on Capability. They Fail on Consistency](https://www.flowx.ai/blog/browser-agents-dont-fail-on-capability-they-fail-on-consistency) benchmarks nine LLMs driving the open-source [browser-use](https://github.com/browser-use/browser-use) library through nine controlled web tasks, five repetitions each: 405 runs in total. Tasks cover multi-page forms, iframes, shadow DOM elements, scrolling, and error recovery. Every outcome is verified against the target site's recorded submissions and state changes, not against the agent's own success claim.

Key findings:

* Several models passed all 45 of their runs, and the models that did span a 26.5x per-run cost difference: price does not predict reliability.
* Some models reported success even on runs that failed, with overconfidence rates reaching 26.7%.
* Checking only whether the requested action happened misses over-actions, such as an agent submitting the same claim twice; outcome-state verification catches them.

<Card title="Read the benchmark" icon="globe" href="https://www.flowx.ai/blog/browser-agents-dont-fail-on-capability-they-fail-on-consistency">
  Full methodology and per-model results on the FlowX.AI blog
</Card>


## Related topics

- [Deep Research node](/5.9/docs/platform-deep-dive/integrations/deep-research-node.md)
- [FlowX.AI 5.12.0 Release Notes](/release-notes/v5.x/v5.12.0-september-2026/v5.12.0-september-2026.md)
- [Deployment guidelines v5.12](/release-notes/v5.x/v5.12.0-september-2026/deployment-guidelines-v5.12.md)
- [Browser Automation](/5.9/ai-platform/agent-builder/browser-automation.md)
- [AI in FlowX](/5.9/ai-platform/ai-in-flowx.md)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.