Ambient AI overview

Case study · Product Strategy

Ambient AI

Making AI agent and device configuration easier through one connected platform.

Role
Lead Product Designer
Contribution
End-to-end product design
Focus
UX strategy, information architecture, user journeys, interaction design, UI design, prototyping and developer collaboration
Platform
Enterprise AI configuration and orchestration platform
The Management Portal: the central workspace for agents, skills, workflows, orchestration and deployment.

The challenge was to make this complexity manageable without removing the flexibility required by an enterprise platform.

03 The design objective

Create one platform where users can configure, validate and deploy agents and connected devices through a clear, structured workflow.

04 Design approach

My approach

  1. Gateway / MAO management, one of the capabilities brought together in the Management Portal.
  2. The Agent workspace: a persistent conversational entry point on the left, the configuration workspace on the right.
  3. Skill Registry: launch parameters, visibility rules, trigger profiles and templates.
    Workflow Builder: composing steps, configuring parameters and understanding dependencies.
  4. Pre-flight validation: blockers and warnings, with a direct path to the affected configuration.
  5. Publishing and deployment: versioned snapshots, deployment status and execution logs.
Ambient AI end-to-end user flow: 1. Set up — Sign in to Management Portal, Open Agent workspace, Create or select an agent, Review agent details and dependencies; 2. Configure — Configure experts and skills, Set launch parameters and trigger profiles, Build or update workflow, Configure orchestrator and gateway dependencies; 3. Validate — Review templates and mappings, Run pre-flight checks, Resolve blockers and warnings, Test configuration and inspect execution; 4. Deploy & operate — Publish a versioned configuration, Deploy to target device / environment, Monitor execution logs and agent status, Retry failures and maintain newer versions
Fig. 1 Conceptual journey from documented platform workflows; exact role permissions and deployment APIs are implementation-dependent.
Ambient AI product architecture: User-facing layer — Management Portal, Agent workspace, Skill Registry, Workflow Builder, Gateway / MAO; Configuration layer — Agent identity and settings, Experts and skills, Trigger profiles and launch parameters, Workflow graph and templates, Orchestrator configuration; Release & runtime — Versioned configuration, Validation and pre-flight, Publish and deployment control, Device / environment connection, Execution logs and runtime status
Fig. 2 Conceptual product architecture, not a verified infrastructure topology. Gateway / MAO is represented as a platform dependency.

07 Decisions

Key design decisions

  1. 01

    Centralised configuration

    Bring the important configuration capabilities into one platform instead of making users manage them through disconnected experiences.

  2. 02

    Agent as the entry point

    Use the Agent workspace as the starting point while keeping specialised configuration tools available in context.

  3. 03

    Visual workflow composition

    Represent workflows as connected steps so users can understand sequence, dependencies and configuration at a glance.

  4. 04

    Decision-oriented validation

    Prioritise blockers, warnings and direct navigation to the affected configuration instead of overwhelming users with technical information.

  5. 05

    Version-aware publishing

    Preserve published workflow versions and make draft changes explicit.

  6. 06

    Visible execution feedback

    Make deployment progress, failed agents and execution logs easier to inspect and act on.

Intro film. Hello World. I’m Nishanth, Lead Product Designer. I make things that help others do their thing. Even when it’s complex, I make it clear. +15% Best engagement lift, 5+ Years designing products, 04 Design systems built or shaped. I design it in Figma, then ship it in code.