User Research Summary — Design-Relevant Findings

Source: 26 discovery call interviews (November 2025, Polish market focus) Trimmed: 6 January 2026 for Zebra Design Sprint relevance

⚠️ Workshop 1 Warning: The Research Tech team explicitly stated this research should NOT be treated as definitive — it's "outdated," "one person's opinion," and "too heavily focused on Polish market." The ICP is NOT defined. Use this for directional UX guidance only.

Full archive: background/business-plan/archive/upfront-user-research-archive-2026-01-06.md

1. The Core Discovery

The "omission problem" is more critical than hallucinations.

While the market has focused on AI making things up, the deeper fear among investment professionals is AI missing critical signals — the buried red flag on page 99, the undisclosed mafia connection, the IP ownership clause hidden in footnotes.

UX Implication: The interface must communicate comprehensive coverage, not just accuracy. Show what was found AND signal confidence in coverage.

2. Key Pain Points (Design-Relevant)

Manual Research Time Burden

Prevalence: 69% of interviews

• VC: 50% of working week consumed by research
• M&A Advisory: 1-3 weeks per deal for due diligence
• Consulting: Multiple analyst-days per engagement

UX Implication: The interface must demonstrate time savings. Show the work being done (processing view, agent activity) to justify the wait and communicate value.

AI Quality Crisis & Trust Deficit

Prevalence: 58% of interviews

Users have already tried ChatGPT, Harmonic, Vocalo. Their complaints:

• Hallucinations: AI confidently presenting fabricated information
• Omissions: Critical signals missed entirely (the deeper fear)
• Source Opacity: No way to verify where information originated

UX Implication: Every claim needs visible citations. Evidence drawer is critical. Conflict badges must surface disagreements. The interface builds trust through transparency, not assertions.

"Unknown Unknowns" Risk Detection

Prevalence: 54% of interviews

The phrase: "I don't know what I don't know."

Real examples cited:

• Mafia connection discovered only after investment closed
• IP ownership clause buried on page 99
• Founder's previous failed venture hidden from standard searches

UX Implication: The cheat sheet format with "Top Risks" and "Questions to Ask" directly addresses this. Killer facts surfaced first. Review queue for flagged items.

Process Speed/Timing Pressure

Prevalence: 35% of interviews

• Lost deals when competitors move faster
• Founder frustration with slow VC response times
• Client expectations for rapid turnaround

UX Implication: Speed without sacrificing depth. The processing view should communicate progress and give users something to engage with during longer operations.

3. Competitive Differentiation

What the interface needs to communicate vs competitors:

Key differentiator to visualize: Multi-model verification catches what single models miss. The processing view should hint at this ("3 models cross-checking...").

4. What This Means for Interface Design

From the research, the interface must:

1. Build trust through transparency

   • Every claim has clickable citations
   • Evidence drawer shows source + extracted snippet
   • Conflicts are flagged, not hidden

2. Show value in complexity simplified

   • Processing view demonstrates "1,000 interactions" happening
   • But output is simple: cheat sheet with drill-down depth

3. Address the omission fear

   • "What we couldn't find" disclosure
   • Review queue for low-confidence items
   • Explicit coverage indicators

4. Enable speed without sacrifice

   • Quick first insights (cheat sheet)
   • Deeper report available on demand
   • Progress communication during longer operations

For business strategy, pricing, ICP prioritization, and full recommendations, see the archived version.