Skip to content
Try a demo Talk to us
Search
WAVE CONNECT DEVELOP

How the AI plugin works inside your current tools

Your people are already pasting reports into an assistant. This is the solution your security team can approve: the same convenience, with permissions enforced by us. 
MCP-example
img-wave-header-mobile
HOW IT WORKS

How the AI plugin builds an answer

The AI plugin (MCP) is part of Wave Connect Develop. Responses are defined by that person's own assessment profile, built by our platform rather than by the model. 

A person completes a behavioral assessment, rated 27 out of 30 in review by the British Psychological Society. That profile is the only thing the AI plugin draws on. 

Wave scene mockup
YOUR AI TOOLS

Works with the assistant you already use

The AI plugin is built on MCP, the open standard leading AI assistants now support. Nothing new for your people to log into, and nothing new for your IT team to run. 
openai-logo-1 Claude_AI_logo microsoft-copilot-logo openai-logo-1 Claude_AI_logo microsoft-copilot-logo
ACCESS CONTROL

Who sees what

Three levels of access, set by us rather than by the assistant. Each one is enforced on our side, so what a person can ask about is the same in the plugin as it is in the platform. 

ANSWER RESTRICTIONS

Four measures, not one verdict

1

Someone with no completed assessment

Depending on their access, a user can see who has and has not completed. What the plugin will not do is answer questions about a person without a completed assessment. Nothing is inferred from a job title or a name. 
2

Anyone beyond your access

The plugin refuses, on our side, whoever is asking. A question about your own manager gets the same answer as a question about a peer in another team. 
3

A group that is too small

Every group view needs at least three completed people, applied to each filter. Narrowing a filter until one person is left does not work: the plugin declines and explains why. 
4

Who is the best of these people

The plugin declines. A profile describes how someone works, not how well they work, so there is no best to worst to give. 
SECURITY AND DATA

The answers your security team asks for

 Most of our clients work under strict compliance requirements, which is why every question has an answer. These are the ones that come up first. 

We do not train models on your people

Your data is not used to train any AI model of ours. What your assistant provider does with what passes through it is set by the agreement you already hold with them, the same as any other document your people use. 

Inside the certifications you already ask for

We hold ISO 27001 and SOC 2, the two independent security standards a risk team asks about. The plugin sits inside that scope rather than beside it, so it is not a separate exception for anyone to approve. 

Nobody else in the chain

No third party handles this data on our side. Once an answer reaches your assistant it is handled by the assistant you have already approved, under the agreement you already hold with them. 

Encrypted in both directions

The data is unreadable to anyone intercepting it while it moves between systems, and unreadable while it is stored.  

Where your data goes

Profile insight passes into your assistant so it can appear in the conversation, and nowhere else. It is read to answer a question, and that is the end of it. 

Help & Support

Frequently asked questions

If it is not here, ask us. You will get a fast answer from an expert, not a form reply. 

The leading AI assistants all work with it, because the plugin is built on MCP and they support that standard. ChatGPT, Copilot and Claude are the three we have tested.  

Get Started

Put it in front of your security team

Tell us what your reviewers need and we will bring the right people to the call, including the documents that do not sit on a web page.