BackResearch Tech
Research Nodes — findings categories, source analysis, cross-referenced insights
Executive Brief — 12 key insights categorised by risk, opportunity, and conflict
The challenge

Pre-seed AI startup.
No product designer.
7 months to seed round.

From functional API to investor-demo-ready diligence platform.

The insight

User testing found AI-chat reduces trust.

That single finding saved months of wrong engineering.

Evidence Drawer — sources panel with verified sources, confidence analysis, and citation context
Validation

9 users navigated from summary to evidence without getting lost.

100% task-completion rate.

AI command palette — search, analyse risks, summarise findings
Audio Briefing on mobile — transcript with chapter navigation
Differentiation

Conflict detection — when sources disagree, show both sides.

No competitor has this.

Review Queue — conflicts, unverified claims, low confidence items flagged for attention
Research Planning — 27 categories, cross-verified, clarification questions before processing
The output

25+ screens. 173 components.
55,000 lines of TypeScript.

Shipped to the client's codebase in 4 weeks. No Figma.

Research Method Preview — 27-category DAG with dependency edges and colour-coded nodes
Edit Research Scope on mobile — node detail popup with dependencies
Evidence

See the shipped product.

Explore the actual shipped product and handover.

→ View live product
“I’m amazed. Unbelievable.
You are using AI so well,
and everything was done so fast.”

— Bartosz Barwikowski, Co-founder, Research Tech

Charlie Ellington

Research Tech

4 Sprints · 25+ screens

2026 Availability