Product Features
What VendorTruth actually does, what it doesn't, and when to use it.
Dialectical Verification Engine
How it works:
- Runs two parallel AI agents (prosecution + defense) investigating vendor claims from opposing perspectives
- Uses a small, fast language model optimized for follow-up question generation
- Recursively explores topics 1-5 levels deep (configurable)
- Streams live progress updates during 30-60 second research cycles
Strengths:
- Avoids confirmation bias by forcing exploration of contradictory perspectives
- Fast (2-5 minutes for comprehensive reports vs days for manual research)
- Transparent methodology—you see the questions asked and sources consulted
- Scales to investigate multiple vendors in parallel
Limitations:
- Limited by public web sources (Exa API)—can't access paywalls, private communities, confidential feedback
- Follow-up questions depend on initial search quality—weak initial results cascade into shallow research trees
- No human fact-checking layer before reports publish
- Limited effectiveness for new vendors (<6 months old) with sparse online footprint
- English-language bias—non-English vendor docs get shallow coverage
When to choose VendorTruth:
- Evaluating established B2B vendors with public documentation
- You need balanced research fast (pre-purchase evaluation, contract negotiation)
- Vendor marketing feels too good to be true and you want adversarial analysis
When to consider alternatives:
- New/stealth vendors with minimal public presence → use direct customer references
- Highly regulated industries requiring audit trails → use Gartner/Forrester analyst reports
- Custom enterprise software → hire consultants with domain expertise
Truth Reports
What's included:
- Claim Rating (True / Mostly True / Misleading / False / Unverified / Mixed)
- What's True - validated facts from defense perspective
- Limitations - caveats from prosecution perspective
- Strengths - positive aspects with evidence
- Weaknesses - concerns and gotchas
- When to choose [vendor] - use cases where vendor excels
- When to consider alternatives - scenarios where vendor struggles
- Sources - URLs from both prosecution and defense research
How it works:
- Reports cite specific sources with URLs
- Balanced structure forces inclusion of both supporting and refuting evidence
- Knowledge graph links enable deep-dive into related topics
- Markdown export supported for sharing with teams
Strengths:
- Snopes-style fact-check format is scannable and decision-focused
- Inline source citations let you verify claims yourself
- Dialectical structure surfaces gotchas vendors don't advertise
Limitations:
- Report quality depends on public data availability—new vendors get thin reports
- No ongoing updates to historical reports unless you manually re-run verification
- Synthesis (verdict) is AI-generated—you should validate conclusions against sources
- No market share data (not in our sources)
- No SLA on report accuracy
When to use truth reports:
- Pre-purchase vendor evaluation
- Validating vendor sales claims during demos
- Documenting decision rationale for stakeholders
When to skip:
- Time-sensitive decisions (<1 hour) where you can't afford 2-5 minute research latency
- Vendors with zero public footprint (stealth startups, internal tools)
Vendor Monitoring & Alerts
What we monitor:
- Vendor pricing pages, documentation, and policy pages
- Email alerts when changes detected
- Pro tier: hourly checks
- Free tier: daily checks
Strengths:
- Catch vendor price increases before renewal deadlines
- Track terms of service changes that shift liability
- Low-effort monitoring (set once, receive alerts passively)
Limitations:
- Alert significance is AI-judged—may flag trivial changes or miss critical ones
- Supports pricing and policy changes only (not security incidents or outages)
- Can't track changes behind login walls or in vendor dashboards
- No historical diff view (alerts show "changed" but you manually compare versions)
- Alert fatigue if you monitor high-churn vendors (frequent feature releases)
When to use monitoring:
- Tracking vendors you've already purchased from
- Evaluating multiple vendors over weeks (catch changes during evaluation period)
- Compliance requirements to document vendor policy updates
When to skip:
- One-time evaluations where you'll decide in days
- Vendors that change daily (monitoring noise outweighs signal)
Knowledge Graph
How it works:
- Reports auto-inject 3-8 inline links to related topics (like Wikipedia)
- Clicking pending links auto-generates new truth reports
- Full-text search across all published reports
- Visual graph interface shows vendor relationships
Strengths:
- Natural discovery of related vendors and topics
- SEO-optimized public pages (each report becomes searchable content)
- Low-friction deep dives (one click to explore tangent topics)
Limitations:
- Visual interface is basic (force-directed graph, not advanced filtering/clustering)
- Search is keyword-based (not semantic—may miss conceptually related reports)
- Graph grows organically based on user clicks (coverage has gaps)
- No manual curation to ensure high-value connections (relies on AI link injection)
- Visual graph gets cluttered with 100+ nodes
When to use knowledge graph:
- Exploratory research (you don't know what you don't know)
- Discovering vendor alternatives organically
- Building context before deep evaluation
When to skip:
- You know exactly what vendor you're evaluating (direct search is faster)
- Need comprehensive vendor comparison matrix (graph is discovery-focused, not systematic comparison)
Next Steps
- API Documentation - Programmatic access (rate limits apply, see pricing)
- Getting Started - Create your first truth report
- Technical Architecture - How the system works under the hood