AI Products & Agents
AI products ask designers to solve new problems: how people talk to software, how they review what a model produced, and how they stay in control of automated work. These agencies show that thinking in named projects, from assistants inside business tools to generative creative products.
What we looked for
Projects where the agency documents its interface work on an assistant, conversational flow, agent workspace or generative tool, rather than general AI branding or a marketing site. We focus on how the interface handles prompt construction, model output inspection, human-in-the-loop approval, and multi-step agent supervision.
- Prompt & Context UX. How users scope instructions, attach domain context, and refine generated responses without losing thread state.
- Explainability & Supervision. Interfaces that let operators inspect citations, confidence signals, and autonomous agent actions.
- Generative Canvas Flows. Creative and analytical workspaces that blend direct manipulation with model-driven generation.
Agencies
Showing 11 agencies across all topics
No agencies in this collection match that filter.
Fuselab Creative
McLean, Virginia, United States
AI copilots and assistants · Voice AI and conversational experiences · AI workspaces and agent supervision
Relevant project: Stardog Voicebox
Conversational AI/data workspace. Despite its name, the case documents chat, not speech UX.
View Stardog Voicebox case study ↗ (opens in new tab)Portfolio reviewed:
Futurice
Helsinki; European offices, Finland
AI copilots and assistants
Relevant project: Neste
Conversational knowledge assistant designed and built with Recordly.
View Neste case study ↗ (opens in new tab)Portfolio reviewed:
Netguru
Poznan, Poland
AI copilots and assistants
Relevant project: ITI
Dental-professional knowledge assistant with explicit product-design scope.
View ITI case study ↗ (opens in new tab)Portfolio reviewed:
Pixeldarts
Berlin, Germany
AI copilots and assistants
Relevant project: Piry.ai
AI reading companion, summaries, notes and gamified reading.
View Piry.ai case study ↗ (opens in new tab)Portfolio reviewed:
hedgehog lab
Newcastle upon Tyne, United Kingdom
AI copilots and assistants
Relevant project: Conflict Dynamics / Akord AI
Research contributionAI peacebuilding copilot and user research. NGO client, not a government-contract claim.
The agency's documented role here is research or discovery work rather than a complete shipped interface.
View Conflict Dynamics / Akord AI case study ↗ (opens in new tab)Portfolio reviewed:
UX studio
Budapest, Hungary
AI copilots and assistants
Relevant project: Inquisita
Prototype / demoResearch and clickable prototype for a legal discovery-response assistant.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View Inquisita case study ↗ (opens in new tab)Portfolio reviewed:
AI workspaces and agent supervision
Relevant project: Oversee.ai
Prototype / demoAI-model monitoring dashboard, PoC and investor-demo MVP; not proof of agent orchestration.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View Oversee.ai case study ↗ (opens in new tab)Portfolio reviewed:
Eleken
Kyiv, Ukraine
AI copilots and assistants
Relevant project: Privado Dining
AI copilot inside an event-management product.
View Privado Dining case study ↗ (opens in new tab)Portfolio reviewed:
AI workspaces and agent supervision
Relevant project: Siena
Commerce support-agent workspace, automation testing and evaluation controls.
View Siena case study ↗ (opens in new tab)Portfolio reviewed:
Generative creative tools
Relevant project: Modia
AI editorial content generation and refinement workflows.
View Modia case study ↗ (opens in new tab)Portfolio reviewed:
Lazarev.agency
San Francisco, United States
AI copilots and assistants
Relevant project: Accern Rhea
AI research workflows combining conversation, data visualization, and report creation. Agency case study; funding and business-impact claims not independently verified.
View Accern Rhea case study ↗ (opens in new tab)Portfolio reviewed:
Voice AI and conversational experiences · Generative creative tools
Relevant project: Elva
Voice-first AI video editing; branding, UX, onboarding and agentic editing flow.
View Elva case study ↗ (opens in new tab)Portfolio reviewed:
Bakken & Bæck
Oslo; European offices, Norway
Voice AI and conversational experiences
Relevant project: IKEA Voice
Prototype / demoTechnical prototype and voice-interaction definition for IKEA HomeSmart.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View IKEA Voice case study ↗ (opens in new tab)Portfolio reviewed:
Generative creative tools
Relevant project: World Labs / Marble
Product interfaces for generating and exploring 3D worlds.
View World Labs / Marble case study ↗ (opens in new tab)Portfolio reviewed:
Cieden
Lviv; US presence, Ukraine
Voice AI and conversational experiences
Relevant project: Multimodal banking / anonymized client
Prototype / demoWorking voice/text loan-journey prototype; not a confirmed live banking release.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View Multimodal banking / anonymized client case study ↗ (opens in new tab)Portfolio reviewed:
AI workspaces and agent supervision
Relevant project: Tandom
Prototype / demoClickable MVP for chat-based agent creation, run visibility, triggers and integrations.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View Tandom case study ↗ (opens in new tab)Portfolio reviewed:
Clay
San Francisco, United States
Generative creative tools
Relevant project: Eden
AI-assisted visual home exploration and image generation; no separate copilot claim.
View Eden case study ↗ (opens in new tab)Portfolio reviewed:
Fuselab Creative
McLean, Virginia, United States
Relevant project: Stardog Voicebox
Conversational AI/data workspace. Despite its name, the case documents chat, not speech UX.
View Stardog Voicebox case study ↗ (opens in new tab)Portfolio reviewed:
Futurice
Helsinki; European offices, Finland
Relevant project: Neste
Conversational knowledge assistant designed and built with Recordly.
View Neste case study ↗ (opens in new tab)Portfolio reviewed:
Netguru
Poznan, Poland
Relevant project: ITI
Dental-professional knowledge assistant with explicit product-design scope.
View ITI case study ↗ (opens in new tab)Portfolio reviewed:
Pixeldarts
Berlin, Germany
Relevant project: Piry.ai
AI reading companion, summaries, notes and gamified reading.
View Piry.ai case study ↗ (opens in new tab)Portfolio reviewed:
hedgehog lab
Newcastle upon Tyne, United Kingdom
Relevant project: Conflict Dynamics / Akord AI
Research contributionAI peacebuilding copilot and user research. NGO client, not a government-contract claim.
The agency's documented role here is research or discovery work rather than a complete shipped interface.
View Conflict Dynamics / Akord AI case study ↗ (opens in new tab)Portfolio reviewed:
UX studio
Budapest, Hungary
Relevant project: Inquisita
Prototype / demoResearch and clickable prototype for a legal discovery-response assistant.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View Inquisita case study ↗ (opens in new tab)Portfolio reviewed:
Eleken
Kyiv, Ukraine
Relevant project: Privado Dining
AI copilot inside an event-management product.
View Privado Dining case study ↗ (opens in new tab)Portfolio reviewed:
Lazarev.agency
San Francisco, United States
Relevant project: Accern Rhea
AI research workflows combining conversation, data visualization, and report creation. Agency case study; funding and business-impact claims not independently verified.
View Accern Rhea case study ↗ (opens in new tab)Portfolio reviewed:
Lazarev.agency
San Francisco, United States
Relevant project: Elva
Voice-first AI video editing; branding, UX, onboarding and agentic editing flow.
View Elva case study ↗ (opens in new tab)Portfolio reviewed:
Bakken & Bæck
Oslo; European offices, Norway
Relevant project: IKEA Voice
Prototype / demoTechnical prototype and voice-interaction definition for IKEA HomeSmart.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View IKEA Voice case study ↗ (opens in new tab)Portfolio reviewed:
Cieden
Lviv; US presence, Ukraine
Relevant project: Multimodal banking / anonymized client
Prototype / demoWorking voice/text loan-journey prototype; not a confirmed live banking release.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View Multimodal banking / anonymized client case study ↗ (opens in new tab)Portfolio reviewed:
Fuselab Creative
McLean, Virginia, United States
Relevant project: Stardog Voicebox
Conversational AI/data workspace. Despite its name, the case documents chat, not speech UX.
View Stardog Voicebox case study ↗ (opens in new tab)Portfolio reviewed:
Fuselab Creative
McLean, Virginia, United States
Relevant project: Stardog Voicebox
Conversational AI/data workspace. Despite its name, the case documents chat, not speech UX.
View Stardog Voicebox case study ↗ (opens in new tab)Portfolio reviewed:
Cieden
Lviv; US presence, Ukraine
Relevant project: Tandom
Prototype / demoClickable MVP for chat-based agent creation, run visibility, triggers and integrations.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View Tandom case study ↗ (opens in new tab)Portfolio reviewed:
UX studio
Budapest, Hungary
Relevant project: Oversee.ai
Prototype / demoAI-model monitoring dashboard, PoC and investor-demo MVP; not proof of agent orchestration.
Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
View Oversee.ai case study ↗ (opens in new tab)Portfolio reviewed:
Eleken
Kyiv, Ukraine
Relevant project: Siena
Commerce support-agent workspace, automation testing and evaluation controls.
View Siena case study ↗ (opens in new tab)Portfolio reviewed:
Eleken
Kyiv, Ukraine
Relevant project: Modia
AI editorial content generation and refinement workflows.
View Modia case study ↗ (opens in new tab)Portfolio reviewed:
Lazarev.agency
San Francisco, United States
Relevant project: Elva
Voice-first AI video editing; branding, UX, onboarding and agentic editing flow.
View Elva case study ↗ (opens in new tab)Portfolio reviewed:
Clay
San Francisco, United States
Relevant project: Eden
AI-assisted visual home exploration and image generation; no separate copilot claim.
View Eden case study ↗ (opens in new tab)Portfolio reviewed:
Bakken & Bæck
Oslo; European offices, Norway
Relevant project: World Labs / Marble
Product interfaces for generating and exploring 3D worlds.
View World Labs / Marble case study ↗ (opens in new tab)Portfolio reviewed:
Evidence labels
- Research contribution: The agency's documented role here is research or discovery work rather than a complete shipped interface.
- Prototype / demo: Client concept, prototype/demo, or MVP-design qualification — not necessarily a deployed production system.
Scope notes
- Conversational AI is not treated as spoken-voice design unless the project says so. Fuselab Creative appears under conversational experiences for a typed interface.
- Clay is placed under generative creative tools only, and UX studio’s Oversee.ai is an AI-monitoring proof of concept.
Frequently asked questions
What types of AI product interfaces are covered in this collection?
This collection covers four published topics: AI copilots and assistants, Voice AI and conversational experiences, AI workspaces and agent supervision, and Generative creative tools. Every agency placement links directly to a named product case study rather than a general AI marketing page.
How do you distinguish between AI copilots and agent supervision workspaces?
Copilot placements focus on embedded assistants that help users draft, query, or summarize inside an existing workflow (such as Futurice’s Neste knowledge assistant or Eleken’s Privado Dining copilot). Agent supervision placements focus on workspaces where operators configure triggers, monitor multi-step runs, or evaluate automated support and model outputs (such as Eleken’s Siena or Cieden’s Tandom MVP).
Does placement under Voice AI and conversational experiences always mean spoken-voice UX?
No. We distinguish between spoken-voice interaction (such as Lazarev.agency’s Elva voice-first video editor and Bakken & Bæck’s IKEA Voice prototype) and typed conversational interfaces (such as Fuselab Creative’s Stardog Voicebox, which documents a chat and data workspace rather than speech UX).
Why do some AI placements carry a Prototype / demo label?
Emerging AI capabilities are frequently validated through interactive proofs of concept or investor-demo MVPs before full production rollout. Projects such as UX studio’s Inquisita and Oversee.ai, Bakken & Bæck’s IKEA Voice, and Cieden’s Tandom carry the Prototype / demo badge so buyers can see the exact scope of the engagement.
Which agencies in this collection have documented generative creative tool experience?
Eleken (Modia editorial generation), Lazarev.agency (Elva AI video editing), Clay (Eden visual home exploration and image generation), and Bakken & Bæck (World Labs / Marble 3D world generation) are listed under Generative creative tools.
How can I filter this collection to compare studios for a specific AI workflow?
Use the topic pills above the agency grid to switch between All topics and any of the four AI sub-topics in their curated editorial order, or use the search box to filter by project name, capability keyword, or agency location.
Related collections
How we select agencies for AI Products & Agents
Every studio in this collection is backed by a named portfolio project reviewed against our inclusion criteria. See our selection methodology or report a correction.