- Independent AI Assurance Tools- 


Protecting Reputations,
​
Accelerating Safer Suppliers.

---

Situation and Background to AI Risk & Liabilities

---

-  Visibility of Operational and Governance Risks-

Problem:  Your organisation already has approved and unapproved use of AI in its operations.  Daily loss of intellectual property, information governance breaches, and increased legal liabilities are not uncommon.

Solution:  PRIDAR  - A supplier self report tool that surfaces the risks AI suppliers place on your organisation

---

Problem:  Preparation for AI Governance (e.g. IS0 42001) is rate limited by the availability of consultants

Solution:  Adam  - A tool that you can use to get ready for ISO 42001 without consultants

---

Problem: As a GenAI User you recognise that they do not have control over LLMs, and that they are not static. 
So their prompt/LLM combinations do not always work and and you often need to react quickly to remedy the situation

Solution: PLIM - automatically identifies when to switch a Prompt or LLM out for another

---

Risk Mitigation Examples

Below are some of the robust, low cost methods your suppliers can use to mitigate the risks they place on your operations. 

​Each one of them can be embodied into existing supplier's systems. Common systems used being:

​ ​Jira/Atlassian, Linear, Plane, Azure DevOps/Microsoft, Monday.com, GitHub/Microsoft, GitLab, Bitbucket/Atlassian, CircleCI, Datadog, New Relic, Grafana/Grafana Labs, PagerDuty, Sentry/Functional Software, Slack/Salesforce, Confluence/Atlassian, Notion/Notion Labs, Docker, Terraform/HashiCorp, Figma, FigJam/Figma, Miro.

---

Problem:  Regulators and Insurers want suppliers to have an auditable history of Human-GenAI interactions during the process of creating code

Solution: Marc - automatically creating an auditable history of developer-GenAI code commitments on Git

---

Problem: AI performance can change quickly so the methods for managing it need to be responsive

Solution: AutoDeclare ML Tracker enables you to implement model observability

---

---

Problem: to comply with ISO 42001 you need an AI Management System for your organisation and they can be costly to design and implement

Solution: Eve - an online AI Management system that has been designed to enable compliance with ISO 42001

---

Problem:  GenAI can be used to decrease the cost of processing information.  But humans need to be designed into the loop at the correct time to comply with regulations.  This can be a costly and time consuming process to design. 

​Solution:  AutoHILL - an automated method of designing GenAI systems with humans-in-the-loop that complies with regulations.

The example below is a Car Insurance Split claim liability System designed with AutoHIL

Problem: in healthcare many sources of data can be used to predict clinical outcomes, but these need to be explained to users, clinicians and the regulator so that they can agree to their use.

Solution:  VitaGraph - a method of graphically explaining the interdependency of data, analysis and inferences associated with every clinical decision