What talks to your customers, what publishes on your behalf, what it costs.
For customer service and marketing: Ojja keeps assistants, voice agents, call analysis and generated content in a single view, whatever the vendor, with the AI disclosure, emotion inference, content marking and what is billed per outcome.
Request a readoutTell your customers they are talking to an AI
Customer service and marketing questions
- Our assistant answers our customers on our behalf: on which tool, which model, and who set it up?
- Do our customers know they are talking to an AI?
- Does the call analysis tool infer the emotions of our customers, or of our agents?
- Our assistant is billed per resolved conversation: how high can the bill go?
- The images generated for our campaigns: does the vendor mark them?
- Does our customers' data leave the European Union?
Ojja does not answer for you. It gives you the facts to answer.
What you see in Ojja Watch
- The “Customers and public” view. Everything that talks to your customers or publishes on your behalf, whatever the vendor: assistants, voice agents, call analysis, generated content. On each line, the AI disclosure, emotion inference, content marking and the linked frameworks.
- The AI disclosure. For each assistant and each voice agent, the disclosure that tells the person they are dealing with an AI: provided by the vendor or not, on by default or not, can be turned off by your administrator or not. When the vendor says nothing: “not documented”.
- Emotion inference. For each feature that analyzes calls or conversations: whether it infers emotions according to the vendor, whose (your customers' or your contact center agents'), from what (voice, face or text). “Not documented” as long as the vendor says nothing.
- Generated content. The features that generate text, images, sound or video, and what the vendor documents about marking that content.
- The assistants you built. An assistant built by your teams on an AI vendor's API has its line, like the others, with a point to review with your counsel: who is its provider within the meaning of the European regulation on artificial intelligence.
- Outcome-based billing. Each feature's billing mode: included in the license, usage-based, in credits, or per outcome, per resolved conversation. And the cap, when the vendor offers one; otherwise, the line joins what is billed with no cap.
- Where your customers' data goes. Where it is processed, training on your data (yes, no, or not documented by the vendor), the date of the vendor's last modified terms.
- What is coming. The updates announced by your vendors, with their effective date, before they apply.
When the vendor opens read access, Ojja reads your settings. Otherwise, Ojja documents its features from its release notes and from what you declare, and each line says where it comes from.
When a tool evaluates your agents
- When a contact center tool evaluates your agents, from their calls or their replies, it also concerns your HR: its features enter the “People” view, with the frameworks the catalog links to employee evaluation.
- Ojja gives no figure per agent.
The frameworks the catalog links
- The frameworks the catalog links to these features, depending on what they do: the European regulation on artificial intelligence, Article 50, applicable since August 2, 2026, and its Article 5(1)(f), on inferring emotions in the workplace, with the points to review with your counsel.
Fictional example
A company has put an assistant on its website and a call analysis tool in its contact center. The assistant's fact sheet says, according to the vendor: AI disclosure provided, on by default, can be turned off by the administrator; billed per outcome, per resolved conversation, with no documented cap. The analysis tool's fact sheet says it infers customers' emotions from their voice; for agents, “not documented”. Both lines enter the “Customers and public” view, with the frameworks the catalog links to them. The company reviews the points with its counsel, then decides.
What you get out of it
- The fact sheet before you buy: the AI fact sheet of the candidate software, in the same format as the others, and the equivalent you already pay for.
- A client's AI questionnaire, filled in from what is read.
- The state of your tools on a given date. On any date since your first readout: what was active, what the vendor said about it, what you had decided. The means to answer a customer, a lawyer or an inspection.
- The COMEX page, every quarter, with Ojja Watch Committee.
What Ojja never does
- Ojja reads no conversation, no message, no ticket, no recording, no transcript, no campaign and no published content.
- Ojja does not say whether your assistant answers correctly: it does not read its answers.
- Ojja does not say whether a use is permitted: it gives the facts, and the points to review with your counsel.
- Ojja never concludes: no risk level, no role, no assessment of whether a text is respected.
Ojja does not provide legal advice: it records and documents; how the law applies to your situation is for your counsel to determine.
For agencies and integrators
Do you set up and configure these tools at your clients? Your profile has its own page: what is on at their end, the AI disclosure and what they are billed.