▲ VerdictDesk/ Deal Screening Desk
Screening memorandum Evidence-gated · Nothing fabricated

Every deal deserves a verdict
it can't argue with.

Paste the deal — a listing, a pitch one-liner, a niche thesis. VerdictDesk runs real web research, scores it across twenty evidence-backed dimensions, and returns a cited screen in memo format with an honest proceed, wait, or reject. Built for acquirers, deal-flow teams, and e-commerce operators who make bets for a living.

Exhibit A — the screening engine
01 · The problem

Betting is expensive. Screening shouldn't be.

$25–70/hr what freelance market researchers bill per screen — re-bought deal after deal Upwork published rates (evidence E0010)
$1.3T private-equity dry powder waiting to be deployed — while deal flow stays flat Bain Global PE Report 2026
33,000 unsold portfolio companies — the cost of bets that cleared a weak screen Bain Global PE Report 2026

Every one of those numbers cites its source. That's the house rule: every claim in every screen links to real evidence or is labelled an assumption — mechanical audits fail the run otherwise.

02 · The output

Screens end in verdicts, not vibes.

Proceed

Evidence of real spending, urgency, and a reachable buyer. The screen says why, with sources.

Wait

Score or evidence confidence below the gate. You get the weakest dimensions and exactly what would change the call.

Reject

The evidence argues against the bet. Passing fast — with reasons you can defend — is the win.

Sample verdict states, illustrative. The gate is mechanical: a Venture Score across 20 dimensions AND a minimum evidence-confidence latch. It cannot be sweet-talked.

03 · The method

Opportunity Intelligence

Twenty evidence-scored dimensions, a mechanical Venture Score, and a gate that rejects weak deals with exact reasons — the market questions your analysts answer by hand today, cited and repeatable in minutes.

Exhibit B — one deal under the light

Validation before capital

ICP, competitive positioning, pricing intelligence, and a validation plan ranked by confidence gained per dollar — for the deals that pass.

Exhibit C — most deals stop here

A blueprint, not a hunch

A 21-section business blueprint, financial model, and operating plan — every number cites evidence or is a labelled assumption, ready to paste into an investment memo.

Exhibit D — the plan, built

Memory that compounds

Every screen is remembered with a permanent ID. Outcomes feed patterns and benchmarks — the more deals you screen, the sharper the next verdict. Built for continuous deal flow, not one-off ideas.

Exhibit E — every screen, remembered
04 · The e-commerce desk

Built for the people betting on e-commerce.

Buying a store?

Micro-PE firms, aggregators, and search funds screen dozens of FBA and Shopify listings per completed purchase. Paste the listing summary — get a cited market, competition, and moat screen before you spend real diligence money.

Betting on a niche?

Operators and DTC brands make inventory bets on niches and product lines. Screen the thesis first: who spends money here today, who serves it, at what price, and what the evidence says about entering now.

We don't scrape marketplace sales estimates — data tools do that well. We deliver cited judgment: every claim links to a source you can check, and the verdict includes an honest no.

Exhibit F — deal flow that compounds
05 · The terms

Priced under one hour of the old way.

Freelance researchers bill $25–70/hour and a manual screen takes hours. Every plan below costs less per screen than one of those hours.

VerdictDesk · runs on your infrastructure · your data never trains anyone else's model · every claim cited or labelled
▲ VerdictDesk

Runs locally — credentials never leave this server.

▲ VerdictDesk
Personal workspace
Desk New Screen Portfolio Opportunities Validation Product Launch Providers Settings
Personal / Home
Notifications

The desk

Screen store acquisitions and niche entries — evidence first, verdicts you can defend
New screen

Your screens

All screens →

Needs attention

Verdicts at a glance

Engine capabilities & providers
Full diagnostics →

New Screen

Paste a deal or describe a mission — the studio researches, scores, and answers honestly
Brief it like you'd brief an analyst.
A deal one-liner, a pitch summary, or a market thesis — who it's for, what problem, any constraints. The pipeline turns it into cited evidence, a Venture Score, and a defensible verdict.
Audience Problem space Market / constraints
Advanced options

Live pipeline

queued 0/32 phases
– Current phase
0Evidence sources
–Venture Score
–Launch readiness
–Est. API spend
Pipeline log › 

        

      


      

        
Portfolio

          
Every opportunity you've analyzed — and what
            the outcomes teach

        

        

          
Industries

            

          
Venture scores

            

          
Launch readiness

            

          
Outcomes

            

          
Recorded revenue

            

          
Portfolio growth

            

        

        
Recommendations

          pattern-based, only with evidence
        

        

      


      

        
Opportunities

          
Each run, its gate decision, and the full
            evidence trail

        

        

          
          

            

              

            

          

          

          

          

        

      


      

        
Validation

          
Launch readiness, validation plan, and revenue
            roadmap for the selected run

        

      


      

        
Product

          
MVP build verification for the selected
            run

        

      


      

        
Launch

          
Marketing assets and the investor
            package

        

      


      

        
Provider
          Diagnostics

          
AI provider health, model access, and honest
            failure reasons — separate from your app plan

          

        

        

      


      

        
Settings

          
Providers, gate threshold, plan, and
            session

        

          
Provider capabilities

            
Configured on the server via environment
              variables.

            

          

            

              
Venture gate threshold

              
Minimum Venture Score (0–100) before the
                pipeline builds anything.

              

                
                

            

              
Billing

            
Session

              

          

        

        

          
Provider setup guide

          
All keys live in the server-side
            .env file (project root) — never
            paste keys into this browser. After any change, restart the
            server. Verify with Provider
            Diagnostics → Test provider.

          
Anthropic
            
Create an API key in
            the Anthropic Console (console.anthropic.com → API keys).
            A Claude.ai subscription is separate from API billing
            — API credits are purchased in the Console under Plans &
            Billing. Set ANTHROPIC_API_KEY in
            .env, restart the server.

          
OpenAI
            
Create a key at
            platform.openai.com (API billing is separate from a ChatGPT
            subscription). Set OPENAI_API_KEY.
            Acts as the automatic fallback brain.

          
Search
            
One of
            TAVILY_API_KEY,
            SERPER_API_KEY, or
            BING_SEARCH_API_KEY is sufficient.
            Without one, DuckDuckGo remains the best-effort fallback.
            

          

            ElevenLabs

            ELEVENLABS_API_KEY enables real
            voiceover audio instead of render-ready packages.

          
HeyGen
            
Avatar video requires
            all three: HEYGEN_API_KEY,
            HEYGEN_AVATAR_ID,
            HEYGEN_VOICE_ID.

          
KIE
            
Optional
            image-generation fallback via
            KIE_AI_API_KEY.