Contentsquare
ComplementaryUX / behavioural analytics
STRENGTH
Best-in-class session + heatmap telemetry across 1000+ customer sites.
CEILING
Answers WHAT happened, not WHY. No causal framework, no decision engine.
Gong
ComplementaryConversation intelligence
STRENGTH
Call recording + deal intelligence. Strong for sales coaching.
CEILING
Tells you what reps said; not what GTM system change would fix the deal.
Clari
ComplementaryForecast + RevOps
STRENGTH
Pipeline forecasting with AI confidence scores.
CEILING
Great for the number; doesn't diagnose why the number is off.
HockeyStack
ComplementaryAttribution + journey analytics
STRENGTH
Multi-touch attribution across the B2B journey.
CEILING
Attributes; doesn't prescribe constraint-ordered action.
Dreamdata
ComplementaryB2B revenue attribution
STRENGTH
Deterministic attribution for B2B SaaS pipelines.
CEILING
Which channel worked; not what GTM system state caused the attribution skew.
ChatGPT / generic LLM
Substitute riskGeneral AI chat
STRENGTH
Broad reasoning, cheap to use.
CEILING
No framework, no workspace grounding, hallucinates on specific numbers, no audit trail.
SOPHIE · TRUST
How Sophie keeps herself honest
Every comparison row above only matters if Sophie's answers can be trusted. Five mechanisms run continuously so an answer is never just a confident-sounding guess.
- Citation discipline. Every claim cites a framework slug, a peer-reviewed paper DOI, or an operator-essay URL drawn from the Caugia knowledge corpus (360 GTM frameworks, 607 research papers, 636 operator essays, 4,364 framework-paper-essay citation links). A NEVER_MENTION list blocks confabulated vendor names.
- Native helpdesk corpus. 80 entries on Caugia itself across pricing, feature, how-to, connector, privacy/GDPR, public roadmap, sophie-self, and vision categories - source-controlled and synced to production at 04:00 UTC every day.
- Customer feedback loop. Thumbs-up/down on every Sophie reply writes to
sophie_feedback; thumbs-down fires a real-time email alert with the reason so the fix ships the same day. - Independent weekly grader. A Haiku 4.5 second-opinion runs every Wednesday 04:00 UTC against a 58-case rubric (archetype precision 25 + framework relevance 30 + coverage 20 + coherence 15 + honesty 10 = 100). Last run averaged 95.2/100.
- Confidence calibration. Relevance-weighted scoring with a hard cap at 0.85 when top-1 framework relevance is below 90. Sophie says "uncertain" rather than guessing - and Fair Billing means we do not bill turns below 0.5.
Native in 5 languages today: English, French, German, Spanish, Polish. Every question, answer, archetype, and helpdesk entry is written native-speaker - never machine-translated.
From Contentsquare? We're hiring you.
If you spent years at Contentsquare and you're ready to work on the layer above the dashboard, we want to talk. We're building the decision engine that takes Contentsquare-grade signal and turns it into shipped GTM work.
contact@caugia.com →