AI is a central part of our Product Experience Platform. We’ve built Screeb around AI to help our customers being more efficient, save time and spot improvement opportunities in their digital experiences sooner. AI at Screeb is here to help you increase the adoption of your new features and your success.
At Screeb, AI is not a feature. It’s deeply linked into all the areas of our platform.
TL;DR
AI is deeply integrated into Screeb's Product Experience Platform
Screeb uses AI for both content creation and analysis
AI-powered features include survey drafting, session replay analysis, funnel analysis, and smart tagging
Screeb utilizes the Mistral Small model for enhanced security and GDPR compliance. We don’t use OpenAI or any other American provider at all.
No private user information is shared with AI providers
AI features in Screeb offer time-saving benefits and expertise in prompt engineering
Screeb's approach ensures better security and privacy compared to using general AI tools
If you want more detailed legal information and find our DPA explaining precisely which type of data we use and with whom we share them, you can visit this page.
What do you use AI for?
We use AI both to help you create and analyse content:
Content Creation
When you create a survey or a message, you can ask our AI to help you draft a scenario. You say what your survey or message should be about and our AI will create a draft that you’ll be able to edit before launching.
Also, when you’re creating new questions in your surveys, you can ask our AI to help you write it.
Content Analysis
At Screeb, AI is mostly used to analyse the various insights you can generate with our platform.
Session Replay
One of the main problem with Session Replay is knowing which session to watch and spending a lot of time watching irrelevant sessions. Our AI helps you spot priority sessions to watch and then gives you a summary of each session so you don’t need to watch all of them. In a blink, you can read what happened during this session. Also, it gives you a global summary of what’s happening in sessions.
Funnel Analysis
Our AI analyses how people behave in you key journeys to give you explanation and recommandations.
Smart Tagging
You can set a series of tags you want to add to the feedback you’ll receive. Our AI analyses the content and meaning of feedback to see if one (or more) of your tags should be added to them. Then, you can filter all our reports per tag or be alerted in Slack when a feedback with a specific tag is received.
Emotion Analysis
For each feedback you receive, our AI will rate them from 0 to 5 on 4 emotions: joy, fear, anger and sadness. Then we generate a graph in your reports for you to visualise the emotions of your users, and you can filter all our reports on emotions.
Content Analysis
This is currently our main report based on AI. It helps you save a lot of time by analyzing your feedback in real time. On this report you have multiple information:
Summary and key findings
Opportunities
Strenghts
Weaknesses
Trends
Recurrent topics
Q&A: questions you will find answers to in your feedback
Next steps: recommandations of things you should do to improve your user experience
Benchmark
This report enables you to compare your main metrics to our other customers. With AI, we compare data, give your recommandations and help you understand sooner what you should focus on.
Weekly Briefing
Weekly briefings gather all the insights you collected with Screeb (sessions, funnel, feedback, interactions with in-app messages & product tours) and summarize everything in a single email for you to have an overview of what happened in your product this week. All this gathering an summarizing is done by AI. Also, our AI writes recommandations of actions based on those insights.
You can add various recipients with different roles in your organization. The content of the email will be tailored to their role, thanks to AI.
Which AI model are your using?
For security and privacy reasons, and to better comply with the GDPR, we chose to use the Mistral Small model made and hosted by Mistral.ai.
This choice was the best to be able to quickly deploy new AI-powered features while ensuring that your data is not shared with an American company and while respecting our budget and costs-constraints.
Also, if today we are using the Mistral API so we can focus on building the features, choosing a Mistral model also gives us the opportunity to host them ourself in the future since their models are open sourced.
We don’t use any OpenAI model or API at Screeb.
Also, we chose to use a Large Language Model (LLM) and not specialized models since most of the features we want to integrate to Screeb are doable with a LLM and since it’s easier for us to maintain, compared to having multiple models to host ourselves.
What about security and privacy?
Mistral
Mistral AI offers APIs for building SaaS features with several notable characteristics around privacy and compliance. As a European company based in France, Mistral AI operates under EU jurisdiction and designs its services to align with GDPR requirements. They implement data minimization practices and provide transparency about their data handling procedures.
Also, their contract ensure that no data that we share with them are used to train their model, which is not the case for OpenAI for instance.
Finally, as said above, in the future we’ll be able to host their model ourself, ensuring even more the security and privacy of your data, which is not possible with most of the other AI models providers.
We don’t send private information to Mistral
Even if we’re working with a European partner hosting their data in Europe, we make sure not to share any private data with them. No user information or property are shared with Mistral. We only share raw data (events, feedback…) without information about who created those data.
Session Replay
While we also respect our ‘no private information shared with Mistral’ policy for Session Replay, some actual screen content could contain private information. For instance, if on the homepage of your product you display the name of your user, that information contained in the recording could be share with Mistral, for summarizing purpose.
You can prevent that by opting for a complete text masking in your recordings.
What’s the point of using AI-powered features in Screeb instead of just using ChatGPT?
Time saving
Letting us do the analysis work on your behalf is the best way to ensure that you’ll save a lot of time. You won’t need to export data, you won’t need to iterate on finding the right prompt and won’t spend any time analyzing the results to make sure everything is right, since ChatGPT and other AI tools can hallucinate.
Prompt engineering expertise for customer insights
We spend a lot of time working on our prompts so you don’t have to do it. Crafting the right prompts for reports like the ones we have in Screeb require a strong expertise both in customer insights and prompt engineering.
Better security and privacy
If you choose to use ChatGPT (or any other tools like that), you’ll have to share all your customers data with them. Based on the tool you’ll use, it could mean that those data will be used to train their model and won’t be protected. Also, most of those tools are made by American companies and hosted by American cloud providers, meaning that your data would leave the EU which is not GDPR compliant.
What can I set to have more tailored results for my business?
For now, our reports are pre-built and designed to help all of our users. We don’t offer features to customized them or our prompts for the moment.
Yet, if you have specific needs for a tailored report based on AI, please tell us!
Details on how each of our AI-powered features are working
Details on how each of our AI-powered features are working
Session Replay
Session Replay
Sessions Summary
Sessions Summary
We give all the session summaries to the AI and ask it to summarize them again into a single paragraph.
Session Summary
Session Summary
We transform our raw session file into a more comprehensive one, including the name of the elements on which there have been a click (for instance) and other types of information to give the AI as much context as possible. Then, we ask the AI to describe what’s happening without giving too much descriptive details not just to have a list of what’s happening but to have an overview of the most important actions.
Funnel Analysis
Funnel Analysis
Funnel Summary
Funnel Summary
We send the raw funnel data to the AI, with the event names, conversion rates and drop off rates and we ask it to analyze those data and formulate recommandation on how to improve them.
Survey
Survey
Survey Creation
Survey Creation
We share best practices of what’s make a good survey with the AI, based on what has been working well for our customers + we detailed the various formats of questions we have. Then, we send it the topic of your survey you enter into Screeb and ask it to create a scenario for a new survey that will help you gather the feedback you want.
Question Suggestion
Question Suggestion
We look at the type of question you want to ask and ask the AI for the best way to ask this question in an online survey.
Emotion Analysis
Emotion Analysis
For each feedback you receive, our AI will rate them from 0 to 5 on 4 emotions: joy, fear, anger and sadness.
Smart Tagging
Smart Tagging
Our AI analyses the content and meaning of feedback you receive to see if one (or more) of your tags should be added to them.
Content Analysis
Content Analysis
We use a unique prompt for each section displayed on this page.
Summary & Findings: we ask the AI to summarize all the feedback you received.
Opportunities: we ask our AI to look for opportunities of improvements to meet your users expectations
Strengths: we ask our AI to list the strengths of your app, based on feedback
Weakness: we ask our AI to list the weeknesses of your app, based on feedback
Trends: we ask our AI to spot new trends (good or bad ones) in the feedback
Keywords & Highlights: we ask our AI to list the 10 most recurring topics in your feedback. This may not be actual keywords but topics, categories of feedback. Here, we also ask the AI to list 3 representative feedback. By reading those 3 feedback, you can have a good understanding of what people are saying.
Q&A: We ask our AI to imagine 4 questions you may have and that you’ll find answers to in your feedback.
Next Steps: This is a list of 10 improvements you should make to your app to meet your users expectations, based on the feedback you received.
Benchmark
Benchmark
Observation: For each metrics, we ask the AI to make an observation of your performances compared to those of your competitors and those of the top performer in your industry, among our customers. This way, you can see what you should do to improve your performances.
How to Analyze your metrics: In this panel, we ask the AI to make a full analysis of the data we display on this page. Our goal is to make the data easier to interpret and the decisions to make more straightforward.
Weekly Brief
Weekly Brief
We ask our AI to analyse all the insights you collected with Screeb during the week: Sessions, Funnels, Surveys and In-App Messages. From that, we ask it to make a global summary so you have a clear overview of what’s happening in your product. Then, based on those data, we ask the AI to give you recommandations of things to do.
Both for the summary and the recommandations, we ask the AI to write the email for the role of the recipient. So a Product Manager and a CPO, or a CEO won’t receive the same email but an email tailored to their role.