Product Space UIUC · Fall 2026

Four clients.
Twelve weeks.
Pick your team.

A speech AI startup out of a16z Speedrun, a Seattle AI-search company in public beta, a $1.5B partnerships platform, and a 49-year-old lighting manufacturer betting on 3D printing.

Oruk
Gumshoe AI
impact.com
Brownlee Lighting
End of September → mid-December 2026Fall 2026 Project Overview
Oruk

Speech AI that reads how something was said, not just what was said — emotion, sarcasm, and intent, exposed through a developer API.

a16z SpeedrunSpeech AISeed-stageDeveloper APIStanford · Berkeley · UCSB · Cambridge
Overview
  • Oruk builds speech models that understand nuanced human speech — emotion (15 labels), speaking style (16 labels), and sarcasm — returned alongside a transcript in a single API response.
  • The bet: as AI interaction shifts from typing to voice, demand grows for speech intelligence that captures tone, not just words. Backed by a16z Speedrun and built by researchers out of Stanford, Berkeley, UCSB, and Cambridge.
Product Areas
Developer Activation & Onboarding Funnel
Signup → First Successful API Call
Developer Research & Friction Analysis
Competitive Teardown (Deepgram, AssemblyAI)
Growth Experimentation & GTM Strategy

What your team actually does

  • Walk the signup-to-first-API-call flow yourself and document every point of friction, confusion, and delay
  • Map the activation funnel and set baselines: landing → signup → API key → first call, and time to first call
  • Interview developers who signed up but never made a call — find out what stopped them
  • Benchmark onboarding and docs against Deepgram and AssemblyAI
  • Design and prioritize activation fixes: quick-start, docs, messaging, lifecycle emails, in-product guidance

oruk.ai

Oruk homepage showing emotion labels detected across a speech waveform
77.8%
Speech emotion accuracy
4.44%
Word error rate
56.2%
Sarcasm macro-F1

Skills you'll build

Funnel analyticsDeveloper interviewsPLG growth experimentsCompetitive teardownsAPI & docs product sensePositioning
12 weeks · 1–2 Product Leads · 4–6 Product ConsultantsYou’d work with: Oruk's founding team — day to day with the CEO
03
Gumshoe AI

Tracks what ChatGPT, Claude, Gemini and Perplexity say about your brand — then tells you how to change it.

$2M pre-seedAI search / AEOSeattlePublic beta7,500+ brands & agencies
Overview
  • Gumshoe AI gives companies dashboards and research-backed reports on how they're represented across 11 AI models, so they can improve brand perception and visibility in AI-driven search.
  • Founded by Todd Sawicki (ex-CEO, Zemanta) and Patrick O'Donnell (co-founder, Urbanspoon and Mighty AI). Traction is strong — hundreds of companies in public beta — and the next problem is turning report insights into downstream products and recurring revenue.
Product Areas
Customer Discovery with Marketing Agencies
AI Search & Brand Monitoring Landscape
0→1 Product Opportunity Ideation
AI Workflow & Automation Design
Concept Validation & Prototyping

What your team actually does

  • Learn the platform and the AI search optimization landscape, then map the workflows Gumshoe's data already powers
  • Run customer discovery with marketing agencies — their workflows, priorities, and pain points
  • Brainstorm 5–10 downstream product opportunities and score them on customer value, business impact, feasibility, and AI automation fit
  • Narrow to the top 1–3 concepts with the Gumshoe team, then spend Weeks 4–10 building and validating them

gumshoe.ai

Gumshoe AI homepage showing a persona visibility spotlight card
11
AI models monitored
7,500+
Brands & agencies
5
Audits in the platform

Skills you'll build

Customer discoveryOpportunity sizing0→1 ideationAI workflow designPrototypingPrioritization frameworks
12 weeks · 1–2 Product Leads · 4–6 Product ConsultantsYou’d work with: Gumshoe's Head of Product
04
impact.com

The platform brands, agencies and publishers use to run affiliate, influencer, creator and referral partnerships at scale.

$1.5B valuationPartnership managementEnterprise B2B SaaS$100B+ commerce
Overview
  • impact.com is a leading partnership management platform — affiliate, influencer, creator and strategic partnerships in one place — powering over $100B in partnership commerce.
  • The ask is a "fresh eyes" analysis of features they recently shipped: Partner Marketplace, Creator Campaigns, and Recruitment Agent. Where is the friction, confusion, or unexpected behavior their team hasn't had the bandwidth to notice?
Product Areas
Behavioral Product Analytics (PostHog)
Voice of Customer Synthesis (Gong + NPS)
Fresh Eyes Assessment Framework Design
Competitive Context Research
Prioritization Against GTV & Contract Formation

What your team actually does

  • Dig into PostHog for rage clicks, dead clicks, drop-off points, and paths through the product nobody designed
  • Mine impact.com's internal Gong/NPS tool instead of manually reading 8,000+ customer calls and responses
  • Build the Fresh Eyes Assessment Framework on Partner Marketplace, then replicate it across Creator Campaigns and Recruitment Agent
  • Add competitive context per friction point — real gap, industry-wide limitation, or something impact.com already wins
  • Deliver short-, medium-, and long-term recommendations tied to contract formation and GTV

impact.com

impact.com affiliates page showing brand partners including Adidas, Allstate and Lenovo
$100B+
Partnership commerce
8 of 10
Top global brands
8,000+
Calls & NPS responses

Skills you'll build

PostHog analyticsVoC synthesisFramework designCompetitive researchEnterprise SaaSMetric-driven prioritization
12 weeks · 1–2 Product Leads · 4–6 Product ConsultantsYou’d work with: One of impact.com's product teams and their Director of Product
05
Brownlee Lighting

An Orlando lighting manufacturer since 1977, asking whether 3D printing can become a real production method for decorative shades.

Hardware & manufacturingFounded 1977Orlando, FLUS-manufactured
Overview
  • Brownlee Lighting manufactures high-performance decorative LED lighting — indoor and outdoor fixtures, acoustic pendants, cylinders, sconces — for commercial and residential spaces, built in the US.
  • This engagement evaluates additive manufacturing as a scalable way to design and produce decorative lighting shades. Product Space acts as an embedded innovation team delivering CAD designs, real printed prototypes, materials research, and a production feasibility case.
Product Areas
Additive Manufacturing & 3D Printing Research
CAD Modeling & Blender / G-code Customization
Physical Prototyping & Design Iteration
Materials Evaluation (UL, Flammability, Optics)
Production Feasibility & Unit Economics

What your team actually does

  • Research trends in decorative 3D-printed lighting and map the design constraints Brownlee actually works within
  • Build CAD models from Brownlee's real fixture geometries; experiment with textures, patterns, and perforations
  • Print and iterate physical prototypes on campus printers, keeping a design iteration log
  • Evaluate materials for lighting performance, UL considerations, and flammability
  • Estimate print time, material usage, throughput, and cost per shade — then recommend whether to scale

brownlee.com

Brownlee Lighting product catalog showing Gem Acoustic pendant fixtures
1977
Manufacturing since
4
Phases, CAD to roadmap
1
Hardware project this fall

Skills you'll build

CAD & Blender3D printing / DFMMaterials researchHardware prototypingUnit economicsDesign guidelines
12 weeks · 1–2 Product Leads · 4–6 Product ConsultantsYou’d work with: Brownlee's Head of Product Strategy & Design
06

Which project is for you

All four run the same twelve weeks with the same team shape. They differ in how much is already defined, what kind of evidence you'll work from, and what you'll be able to point to in an interview.

ProjectThe core questionYour main evidenceBest fit if you wantAmbiguity
OrukSpeech AI · a16z Speedrun Why do developers sign up and never make their first API call? Funnel metrics, developer interviews, competitor onboarding flows Growth and PLG work on a technical product you can poke at yourself Medium — Phase 1 sets the scope
Gumshoe AIAI search · $2M pre-seed What should we build next on top of the insights we already generate? Agency customer interviews, workflow analysis, market research 0→1 ideation and the chance to decide what actually gets built High — you write Phase 2
impact.comPartnerships · $1.5B Are our recently shipped features actually creating customer value? PostHog behavioral data, 8,000+ Gong calls and NPS responses Rigorous, data-heavy PM work inside a large enterprise SaaS org Low — scope is well defined
BrownleeLighting · since 1977 Can additive manufacturing scale for decorative lighting shades? CAD models, printed prototypes, materials and cost analysis To design and build physical things, not just decks Low — four defined phases
Think hard about where you'd excel and what you want to learn, and say why — we place people where they'll thrive, so we can only judge that if you convey it well.
07
1 / 7