
Allan Wilson
President - Team Alert
"I was really impressed with how much they cared about our product."
Every product team eventually faces the same question: voice, image, or multimodal – which one actually fits? Get it wrong and you waste engineering cycles, confuse your users, and ship a feature that erodes trust instead of building it. Discover the right modality for your product with Boldare – where over 20 years of design craft meets AI-native product thinking.
See the companies that trusted Boldare to get it done.











GPT-4o does it. Your competitor just announced it. The board asked about it. So voice goes on the roadmap – without anyone asking whether users actually came to your product to speak. Features built on trend logic instead of user context add complexity without adding value. And complexity, once shipped, is expensive to undo.
A voice feature tested in a quiet meeting room behaves very differently in a busy office, a hospital ward, or a moving car. A modal choice that ignores where users actually are – hands occupied, screen out of reach, ambient noise at full volume – ships a feature that fails exactly when users need it most.
Some modalities require trust before users will engage with them. Voice input in a health app. Image upload in a financial product. These aren't just UX decisions – they're trust decisions. A product that asks for a new level of intimacy before it's earned the right to ask will see users back away, quietly and permanently.
Every modality has failure modes. Voice fails in noise. Image recognition fails in low light. Multimodal input fails when one signal is ambiguous. If the fallback wasn't designed before the feature shipped, users hit a dead end – and dead ends don't get second chances.
Some teams arrive with a modality decision already half-made.
Some are starting from zero.
Some have shipped and suspect something is off.
The three packages below cover all three moments – with a clear deliverable at the end of each.
Modality scorecard
Every current modal choice in your product run through all 5 tests. Each one rated pass, stall, or fail – with the reasoning behind every score.
Gap and risk report
A prioritised list of where your modality decisions are weakest, what the user-facing risk is, and which gaps are worth fixing first.
Written recommendation
A clear, actionable brief your team can take straight into the next product sprint – no interpretation required.
Wondering if your product's modality choices hold up? – Ask AI
Modality decision document
The chosen modality (or combination of modalities) with the full rationale behind it. Built to be shared with design, engineering, and stakeholders without additional explanation.
Failure mode plan
For every modal choice made in the session: what breaks, when it breaks, and how the product degrades gracefully when it does.
Next steps brief
A prioritised set of actions your team takes out of the room – whether that's a design sprint, a prototype, or a further research plan.
Wondering if your team is ready for the workshop? – Ask AI
Working prototype
A functional interface prototype built around your product's specific context of use. Voice, image, text, or any combination that the workshop identified as the right fit – not a generic demo, but something that reflects how your users actually interact.
Voice interface strategy and interaction guidelines
A documented set of voice interaction patterns for your product: what the interface says, when it listens, how it confirms, and how it handles silence or misrecognition.
Failure mode fallbacks
Every failure scenario designed in advance – so when voice doesn't work, the product still does.
User validation report
The prototype tested with real users in the environments where the feature will actually be used. What worked, what didn't, and which modal choices earned the most trust in practice.
Build recommendation
A clear go/no-go brief for each modality in the prototype – with the evidence to back it up and a scoped path to full implementation.
Wondering what we'd build for your product? – Ask AI
Before you book a workshop, read how we think.
The framework is built on real product decisions – the ones that worked and the ones that almost didn't.
Here's what clients say about working with us.
Product Builders | AI-Native is a community for practitioners building digital products in the AI era – run by Boldare, powered by 20 years and 350+ products of hands-on experience.
We regularly go live with guests from product, design, and engineering for honest conversations about what building AI-native actually looks like in practice. Written recaps, articles, and show notes from every session live on Substack.
Modality chosen in a meeting, based on what competitors are shipping
Voice added because the roadmap had space, not because users needed it
Engineering starts before anyone asked where users will actually be using this
Fallback scenarios designed after launch – when it's already too late to change the interaction model
Team aligned on what to build, not on whether it's the right thing to build
First signal something is wrong: user feedback, three months post-launch
Modality chosen against a 5-test framework, not a trending feature list
Voice, image, or multimodal input validated against real user context before a line of design begins
Engineering briefed on a decision document, not a gut feeling
Failure modes mapped in the workshop
Team leaves the session with one agreed answer and the reasoning to defend it
First signal something is right: users engaging with the feature the way it was designed to be used
Tell us where your team is – what AI user interface you're building, what modality decisions are already on the table, and where you're stuck. We'll come back with a clear recommendation on where to start.
One partner from idea to launch
If you're still figuring out whether this is the right next step for your product, these answers should help you decide.
It’s a structured facilitated session that helps product teams decide which interface modality (voice, image, text, or a combination) fits their AI product before design begins. It runs on Anna Zarudzka's 5-test framework: Primary Task, Context-of-Use, Trust & Intimacy, Combination, and Failure Mode. The output is a decision document your designers and engineers can act on immediately.
Voice fits when the user's primary task is served by speaking – and when their hands, eyes, and environment support it. Voice fails when users came to your product for something else, when the context is screen-first, or when the product hasn't yet earned the level of intimacy that voice requires. The 5-test framework gives you a structured way to answer this for your specific product and user context.
Voice UI design focuses on a single input modality – what the interface says, how it listens, how it handles silence and misrecognition, and how it fails gracefully. Multimodal AI design combines two or more input types – voice, image, text, or audio – in ways that create something none of the individual modalities could achieve alone. The workshop identifies which approach fits – whether that's voice UX design for a single focused modality, conversational interface design across multiple touchpoints, or a fully multimodal experience combining inputs.
No formal brief required. It helps to come with a clear picture of who your users are, what they're trying to do, and what modality decisions are already on the table – but if those are still open questions, the workshop is designed to surface them. We'll send a short pre-session questionnaire once you book.
CPO or Head of Product is the core attendee. A product designer and a technical lead from your team make the session significantly more productive – the modality decision touches all three disciplines and alignment in the room means the decision document lands with the whole team behind it.
A clear written output covering: the chosen modality or combination of modalities, the rationale behind every choice scored against the 5 tests, the failure modes mapped for each modal decision, and a prioritised set of next steps for design and engineering. Built to be shared with stakeholders without additional explanation.
Yes. The facilitated Modality Strategy Workshop runs equally well in person or remote. The Modality Audit is primarily async with a structured review session at the end. For SCALE packages – Voice Interface Design Sprint and Multimodal AI Prototyping – we work in whatever setup fits your team's collaboration model.
If your team is ready to build, the SCALE packages take the decision document and turn it into a working prototype. If you need more time to align stakeholders internally before committing to a build, the decision document gives you exactly the material to do that. There's no obligation to continue with Boldare after the ASSESS or BUILD packages.
© 2026 Boldare. All rights reserved.
Boldare S.A. z siedzibą w Gliwicach, przy ul. Zwycięstwa 52, zarejestrowana w Sądzie Rejonowym w Gliwicach, X Wydział Gospodarczy Krajowego Rejestru Sądowego pod nr KRS 0000914518, NIP 6312698829, REGON 38958555. Wysokość kapitału zakładowego i wpłaconego 100 000,00 zł.