Skip to content

Conversation Starters: speed up adoption of your AI agents.

Bring your assistant into the heart of key journeys. Show targeted, contextual invitations to support users at the right time, remove a blocker or trigger a conversion.

  • 10× adoptionwith contextual invitations
  • No developmentSet up like a survey
  • Web and mobileiOS, Android, Flutter, React Native

Why your users don’t use your AI assistant.

On average, only 2% of your users open your AI agent. Adoption stalls as long as it’s relegated to a simple widget at the bottom of the screen. To drive usage, the answer isn’t a more visible button, but inviting the user at the exact moment they need it.

The widget waits for someone to think of opening it.

2%of users spontaneously open a conversational agentMarket finding

The invitation arrives at the moment they need it.

10×agent adoption, with contextual invitationsSeen at our customers
  • Contextual, proactive CTAA popover, a button or suggestions, triggered as soon as a friction or hesitation signal is detected in the journey.
  • Tailored behavioral targetingTriggered by profile, page, behavior or segment, so you only reach the users concerned.
  • Three formatsThe Proactive Starter as a popover, the Embedded Starter built into the page, the Integrated Starter as suggestions in your interface.
  • Built-in impact measurementOnce the conversation starts, Screeb checks whether it resolved the request and built trust.

Trigger the right conversation and measure its impact, in four steps.

A well-placed invitation beats a more visible button. A conversation starter is set up like an in-app message, no development needed.

Choose the signal

When and to whom to show it

A page, a period of inactivity, a page exit, a segment or a profile: the conversation starter is targeted with the same rules as an in-app message, and a control group can be set aside for A/B testing.

On the web and in your apps alike.

Conversation starters are built on Screeb’s in-app messages: web, iOS, Android, Flutter and React Native. In an app, the SDK shows the same popover over the screen, attached to the relevant element and sized to the phone’s width.

  • Web
  • iOS
  • Android
  • Flutter
  • React Native
  • The right element on the screenThe popover points to the row, button or tab that raises a question.
  • The button opens your agentWith the suggested question, just like on the web.

More conversations, and their content to analyze.

A conversation starter drives up your agent’s conversation volume. Agent Analytics then analyzes them one by one: what the user was looking for, whether they got it, and how the exchange left them feeling.

  • The outcome of each conversationResolved, partially resolved, unresolved, abandoned or transferred.
  • Satisfaction and frustrationConversation by conversation, and how they evolve with each agent version.
  • Conversations started by a conversation starterA dedicated filter isolates them in the conversation list.
Discover Agent Analytics

From unsupported request to new use case.

That’s the full loop. Agent Analytics shows you what your users ask for that the agent can’t do yet. You build the use case. And the day it’s available, a conversation starter announces it to the very people who asked for it.

  1. 1Spot the request

    The requests the agent can’t handle, ranked by volume, with the users concerned. Here, booking an in-store test ride.

  2. 2Announce the new use case

    A conversation starter targeted at the users who had asked for it, on launch day. Analyze, build, promote: measurement feeds adoption directly.

Frequently asked questions

Your questions.

How is this different from a classic chat widget?

The widget waits for the user to think of opening it. A conversation starter triggers the invitation at the right time, based on a detected friction signal, instead of waiting for the user to take the initiative.

Can I target who sees the invitation?

Yes, by segment, page or behavior, with the same targeting rules as a survey.

Can I measure the effect of these invitations on adoption?

Yes, by cross-referencing with Agent Analytics and the TRUE Score™, to see whether the conversation started this way resolved the request and built trust.

Which signals can trigger an invitation?

A profile, a page, a behavior or a segment, as well as a friction or hesitation signal detected in the journey.

Which conversation starter formats are available?

Three templates: the Proactive Starter, a popover that invites the user to try your agent; the Embedded Starter, a link or button built into your page; the Integrated Starter, AI suggestions shown in your interface.

How does the conversation starter open the agent?

Through its button’s actions, no development needed: a click triggered on your agent’s launcher, the suggested question set in its input field, or a custom event your code listens for.

Multiply the adoption of your AI assistants, starting today.

In 30 minutes, see a contextual invitation land on your own product.

Shall we place your first invitation together?