pencil Product Designer Interview: Questions, Experience & Prep (2026)

Pencil is an AI-powered creative platform that helps brands and agencies generate, test, and scale ad creatives automatically. Product Designers here sit at the intersection of generative AI and performance marketing, working on genuinely hard problems: making AI outputs feel controllable, helping non-designers produce on-brand work, and building interfaces that serve both agency power users and direct brand teams.

Candidates report the interview process typically runs two to four rounds. You can expect a portfolio review, a design exercise or take-home prompt, and one or two structured conversations covering your process, product thinking, and how you handle AI-specific design challenges. Rounds are typically conducted over video call.

Pencil currently has 11 open Product Designer roles, which signals active hiring across experience levels. The product is evolving fast, so interviewers are likely looking for designers who are comfortable with ambiguity and genuinely curious about how AI changes the creative process, not just designers who can produce polished Figma screens.

1. How would you design an AI-assisted creative workflow for a brand manager who has never used a design tool before?

1. Pencil's core promise is faster, better-performing ad creatives through AI. Walk us through a time you redesigned a complex creative or production workflow to remove friction.

1. How would you approach the onboarding experience for a new advertiser joining the platform for the very first time?

1. What metrics would you track to know whether an AI creative generation feature is genuinely helping users, and not just being clicked?

1. How do you decide where AI should make automatic decisions versus where users need manual control? Walk us through your reasoning with a real example.

1. Pencil serves both independent brands and larger agencies. How would you design a single feature that works well for both personas without building two separate products?

1. How would you design a feedback loop so users can 'teach' the AI to better match their brand guidelines over time?

1. Walk us through how you approach designing around AI outputs that are sometimes unpredictable or inconsistent in quality.

1. A performance marketer wants to test a large batch of ad variants quickly. How would you design the experience to make that fast without overwhelming them?

1. Describe a time you used quantitative data to challenge or revise a design decision you had already shipped.

1. How do you handle design handoffs when the content on screen is dynamically generated by AI and may look different every time?

1. How would you prioritise which new creative formats or channels to support next, given limited engineering capacity?

Q: How would you design an AI-assisted creative workflow for a brand manager who has never used a design tool?

*Situation:* At a previous company, we launched an AI image generator inside our marketing platform, aimed at small business owners and in-house marketing managers.

*Task:* I needed to design the end-to-end workflow so that someone with zero design background could produce an on-brand social post in a single session, without needing to write a prompt or understand design principles.

*Action:* I started with five moderated user interviews with marketing managers to understand their existing process. Most were copy-pasting references into other tools or emailing a designer for every small request. I mapped three major friction points: picking a starting template, writing an AI prompt, and feeling confident the output matched their brand. I then designed a 'brand kit' setup step that took under three minutes and automatically fed brand colours, fonts, and tone into every AI prompt. Instead of open-ended prompt fields, I gave users curated creative direction options like 'bold and energetic' or 'clean and minimal' that translated into structured prompts behind the scenes.

*Result:* In beta testing, users completed their first creative significantly faster than with the original open-ended flow. Qualitative feedback consistently cited feeling in control as a key theme. The brand kit pattern became the foundation for two additional AI features we shipped that year.

---

Q: Describe a time you used data to challenge a design decision you had already shipped.

*Situation:* Our team had shipped a redesigned campaign dashboard for performance marketers. It tested well in moderated usability sessions and the team was proud of it.

*Task:* Three months after launch, the data team flagged that a key action, editing a live campaign, was being completed far less often than with the previous design.

*Action:* I reviewed session recordings and found that users kept missing the 'Edit' button because I had placed it inside a contextual menu to reduce visual clutter. What felt tidy in testing felt hidden under real time pressure. I ran a quick unmoderated test comparing the current design against a version with an always-visible 'Edit' button. The results were clear. I worked with engineering to ship the fix in the next sprint, even though it meant revisiting a decision I had defended in the original design review.

*Result:* The target action rate improved noticeably in the following month. The episode also reinforced a team norm: ship design decisions as hypotheses, not conclusions.

---

Q: How do you handle designing for two very different user personas on the same platform?

*Situation:* At a B2B SaaS company, we had two distinct groups using our asset library: in-house brand teams who prioritised brand consistency, and agency account managers who prioritised speed and volume.

*Task:* I needed to redesign the library so both groups could work efficiently without requiring two separate products or interfaces.

*Action:* I ran a jobs-to-be-done workshop with members of both groups and found that the underlying goal was the same: find the right asset fast. But the paths differed. Brand teams searched by brand category; agencies searched by client or campaign. I designed a single library with a 'View by Brand' and 'View by Project' toggle, letting each persona set their preferred default. I also added a 'Quick access' rail surfacing each user's most recently used assets, which served both groups without any extra configuration.

*Result:* Adoption of the library feature increased across both segments in the quarter after launch. Support requests about 'I cannot find my assets' dropped to near zero based on ticket data, and both team leads reported the library had become a core part of their daily workflow.

For 'How would you design X' questions: Open with who you are designing for and what their core job to be done is. Then walk through four beats: understand the user, define success, explore options, and validate. Avoid jumping straight to a solution. Pencil's interviewers are likely more interested in your reasoning than your final wireframe.

For 'How do you handle AI in design' questions: Pencil is an AI-native product, so these carry real weight. A strong answer names the specific tension you are navigating (for example: user control versus AI automation, or predictability versus creative surprise), explains how you chose where to land on that spectrum, and describes how you planned to learn whether the decision was right.

For behavioural 'Tell me about a time' questions: Use a tight STAR structure. Keep Situation and Task together in two to three sentences. Spend most of your time on Action, with enough detail that the interviewer can picture exactly what you did versus what the wider team did. Close with a concrete Result. 'Support tickets dropped' is more credible than 'users loved it'.

For 'How would you prioritise' questions: Show that you weigh user impact, business value, and engineering effort together. Name your framework explicitly before applying it. Candidates who focus only on user needs without acknowledging business or technical constraints often get pushed back in product-company interviews.

Comfort with AI as a design material. Pencil is not a company where AI is a side feature. Interviewers will look for evidence that you have thought seriously about what it means to design with and around AI outputs: handling variability, setting user expectations, and building in the right amount of human control.

Product thinking beyond craft. Strong candidates connect every design decision to a user outcome or a business goal. A polished portfolio without a clear explanation of why you made each choice will not be enough at a product-led company.

Comfort with ambiguity. The product is evolving fast. Interviewers typically respond well to candidates who describe how they scoped unclear problems, not just how they executed clearly defined ones.

Clear communication and cross-functional collaboration. You will likely be asked how you work with engineers and product managers. Specific examples of navigating disagreement or simplifying a complex tradeoff land better than general statements about being a 'good collaborator'.

Ownership and initiative. Candidates who can point to moments where they spotted a problem that was not assigned to them and then did something about it tend to stand out at companies like pencil.

Week 1, Know the product deeply: Use pencil's platform or study their public demos, case studies, and blog content. Note at least three design decisions you find interesting and form a point of view on each. Being able to say 'I noticed you handle X this way, and I think that is because...' signals genuine preparation, not surface-level research.

Week 1, Audit your portfolio: Select two to three projects that show your process and product thinking, not just final screens. If you have any AI-related design work, lead with it. Prepare a five to seven minute walkthrough of your strongest project using a clear process narrative.

Week 2, Practise AI design questions: Work through the questions listed above. For each one, write your answer in bullet points before practising aloud. Focus especially on questions about user control versus AI automation, since that tension is central to pencil's product.

Week 2, Prepare your own questions: Good areas to ask about include how the design team measures success, how designers and ML engineers collaborate, and what the biggest open design challenges are right now. Thoughtful questions signal genuine curiosity.

Before each round: Re-read the job description and map your specific experience to the skills listed. Candidates report that referencing real product decisions pencil has made tends to land well, so stay current on any product updates or announcements. If you want to make sure you do not miss new openings while you prepare, knok checks 150+ job sites nightly, applies to jobs matching your resume, and messages HR for you.

Jumping straight to a solution. Many candidates hear 'how would you design X' and immediately describe a feature. Interviewers want to see you ask clarifying questions, define success, and consider options before landing on an answer. Slow down.

Treating AI as just a feature. Saying 'I would add an AI button' is a weak answer at a company where AI is the core product. Talk about specific UX challenges: output variability, user trust, latency, and how you design for error states.

Vague results in STAR answers. 'The users loved it' tells an interviewer nothing. Even qualitative results ('support tickets dropped', 'users stopped dropping off at step three') are more credible than adjectives.

Portfolio that only shows polished finals. Pencil is a product company, not an agency. Interviewers want to see sketches, decision points, and moments where you changed direction. Supplement high-fidelity screens with a brief process narrative.

Not asking substantive questions. Ending an interview without asking about real design challenges signals low curiosity. Skip questions about perks or standard process that a recruiter could answer. Ask about what the team is genuinely working through right now.