
In-app AI copilot that guides users & does tasks in SaaS

Tandem AI is an in-app copilot embedded directly inside web-based SaaS products to guide users and complete actions within the interface. It is designed for product, growth, customer success, and support teams that want to reduce friction in key workflows such as onboarding, feature adoption, and self-service support without rebuilding their product.
The copilot appears as a side panel in the product UI, understands on-screen context, and can take actions such as filling forms, clicking buttons, navigating, and orchestrating multi-step processes. Tandem AI is used by companies including Qonto, Aircall, and Sellsy. The company was founded by Christophe Barre and Manuel Darcemont and raised a $3.8M seed round led by Tribe Capital (July 2025).
Teams add a lightweight JavaScript snippet once to their application. The copilot then becomes available as a side panel across the product, without requiring backend changes or SDK-heavy integrations. Product teams configure where the copilot appears, define tone and guardrails, and upload knowledge or playbooks. From there, they can iterate without engineering involvement.
At runtime, Tandem AI analyzes on-screen elements and user context to provide targeted guidance or to perform actions on behalf of the user. It can drive multi-step processes, proactively trigger at friction points, or defer to the product’s native UI when appropriate. If the product UI changes, Tandem can adapt, attempt self-repair (for example, when a button selector changes), or gracefully fall back and notify the team. When needed, it escalates to human support and passes along relevant context.
Tandem AI addresses common friction in both customer-facing and internal software scenarios:
Reported outcomes from the vendor’s site:
| Metric | Value | Context |
|---|---|---|
| Ticket deflection | 70% | In-context answers before tickets reach support |
| Onboarding speed | 3x faster | Guided setup and multi-step assistance |
| Conversion lift | 20% | Faster path to activation and aha moments |
| Time-to-first-value | Minutes | Reduced from days via automated workflows |
Security and compliance overview:
| Area | Details |
|---|---|
| Compliance | SOC 2 Type II, GDPR |
| Encryption | AES-256 |
| Data handling | Client-side operation; vendor states user data is not stored; configurable redaction for sensitive fields |
Comparison with common alternatives:
| Alternative | Typical limitation | How Tandem AI differs |
|---|---|---|
| AI chatbots | Can answer from docs but do not see on-screen context | Sees the current UI state and can take actions to resolve issues |
| Tours and guides | Scripted and linear; break when users deviate | Responds dynamically to what is on screen and user flow |
| Building in-house | High engineering effort and maintenance | Single-script deployment; no backend changes; no-code iteration |
| Live support | Scales with headcount and cost | Automates predictable workflows; escalates only when needed |