AI Claim Verification · Founder Intelligence · Asset Depth
Every claim.
Every founder.
Verified.
TechTruth cross-checks every technical and AI claim in a pitchdeck against real evidence — founder LinkedIn profiles, company websites, and actual AI asset depth. Is it a genuine AI product, or a wrapper dressed up as one?
Numbers update as decks come in. In beta, every submission refines the benchmark.
Every claim in the deck is matched against evidence. Architecture diagrams vs written assertions. Founder credentials vs LinkedIn and company websites. What they say vs what they can prove.
Beta note: claim matching rules are updated weekly based on human expert review.
Is there a genuine AI asset — proprietary models, training data, fine-tuning — or is it an API wrapper with a ChatGPT prompt? We classify AI heaviness and flag inflated AI claims before they cost you.
Every score is calibrated against 265 human-reviewed decks — and growing with every audit — across 4 dimensions: industry, deal size, maturity stage, and AI class. A 7/10 means something very different for a pre-seed wrapper vs a Series A deep-tech.
Beta note: benchmark pool expands with each deck reviewed. Early users directly shape what a calibrated score looks like.
Zero-knowledge GitHub scan — no access to private source code. Metadata extraction validates commit velocity, developer attribution, and dependency integrity against what founders claim to have built.
Get Full Code Scan or Full Tech DD via Enjins →
Unlimited audits.
Your feedback shapes the product.
We're onboarding a final group of VC funds as co-development partners. Max 6 funds total — 3 already confirmed. €799 trial fee, unlimited audits, direct line to the founding team. This is beta learn mode at scale: your deal flow trains the model, your feedback changes the scoring weights, and you get locked pricing before we go to market.
Upload & Analyse
Upload a pitchdeck PDF. We run claim matching, AI depth classification, founder background check, and stage benchmarking — 9-chapter report in minutes.
Auto-Weighted Average
TechTruth AI Generated Report
WCCAA Score
1. Executive Scorecard — Why Commit Capital Auto Weighted Average
| Dimension | Weight | Score | Notes |
|---|---|---|---|
| Founder Team | 70% | 9/10 | Commercial hustler with claimed exits + Ex-Google technical depth. Bus factor risk on CTO. |
| AI Asset Depth | 10% | 8/10 | Deep proprietary time-series stack elevates far beyond a standard wrapper. |
| Technical Moat | 8% | 8/10 | 12 years of R&D in time-series databases creates a massive barrier to entry. |
| Infrastructure & Scalability | 6% | 8/10 | Purpose-built for high-cardinality, high-velocity industrial data without downsampling. |
| Data Strategy | 6% | 7/10 | Strong integration strategy, though reliant on client-owned data. No proprietary global dataset. |
| WCCAA (Weighted Average) | 8.6/10 | Founder ×0.70 + AI Asset ×0.10 + Moat ×0.08 + Infra ×0.06 + Data ×0.06 | |
2. Executive Summary
NewCoEnergy is an industrial AI analytics platform built on a highly specialised, proprietary time-series database stack. Unlike generic AI wrappers that force LLMs to understand raw sensor data, the system uses AI as a natural language translation layer to query a deterministic, high-performance backend with 1,300+ custom sensor-data functions. Primary Reality Gap: temporal inconsistency — the roadmap presents Q1 2026 milestones as future "vision" despite today being March 2026. MLOps pipelines required to reach the promised "Autonomous Prediction" phase are entirely absent from documentation.
6. AI Asset Deep-Dive — Wrapper Check
3. Founder Check — Ghost Scan (preliminary)
5. The Main Pillars
7. Blue-Team vs Red-Team — Gemini vs Claude
NewCoEnergy has solved the hardest part of industrial AI: the data infrastructure. By owning a proprietary TSDB and high-cardinality storage stack, they have a massive unfair advantage over generic AI co-pilots trying to query slow, downsampled datalakes. A competitor with $1M and an OpenAI API key cannot replicate 12 years of R&D. If they position as the horizontal NLP query layer for all major IIoT platforms, they become the default intelligence standard for a $16bn market.
The GTM strategy severely underestimates industrial data integration complexity. Legacy SCADA systems are notorious for unstructured, poorly-labelled, broken sensor tags — an NLP agent is useless if the underlying data taxonomy is garbage. Scaling a highly proprietary database technology requires specialised engineering talent that is hard to recruit, potentially bottlenecking the aggressive 2027 roadmap. MRR: €18K — early traction confirmed.
9. The Why Commit Reality Check
| Element | Founder Claim | Reality | Match |
|---|---|---|---|
| Founder/Team | CEO (3 companies, 2 exits), CTO (Ex-Google), HW/IoT expert on team | LinkedIn: insufficient data. Exit/sale claim: not validated. | Gap |
| AI Architecture | "Instant raw data access through evolving time-series intelligence" | Relies on proprietary TSDB stack; AI acts as NLP-to-query translation layer only | Match |
| Performance | "Time to insight in seconds; cost per query under $5" | Plausible via no-downsampling architecture, but lacks independent benchmark data | Match |
| Roadmap | "1 platform, 10 clients in Q1 2026" | Today is March 2026; deck presents this as future vision — temporal inconsistency | Gap |
| Competitors | "No Time Series DNA, no way to compete with us" | Accurate — generic datalakes struggle with high-frequency industrial IoT data | Match |
| MLOps | "Ladder to Autonomous Prediction and automated reporting" | No MLOps pipeline, drift monitoring, or model retraining infrastructure shown | Gap |
8. Meeting Focus List — Actionable Questions for Founders
"Your roadmap lists Q1 2026 milestones as 'vision', but we are currently in March 2026. Have these targets been fully realised in production, or is there a delay in your execution timeline?"
"How does your AI agent handle the Cold Start problem when integrating with legacy industrial clients whose sensor data is poorly tagged, unstructured, or missing metadata?"
"What specific LLMs are handling the NLP-to-query translation, and what is the inference latency versus actual database execution time?"
"What is your current MLOps stack for training, monitoring, and updating predictive models across isolated client environments at scale?"
VC Launch Program
Unlimited audits.
Real feedback.
We're giving 6 VC funds early access to TechTruth before public launch. 3 spots are already confirmed. This is a working tool, not a demo — and we want partners who will actually use it and tell us what's broken.
Apply for a spot
€799 for 3 months of unlimited audits. We need real users who will push the product and give honest feedback — that's the deal.
We'll review your application and reach out within 48 hours. We're selective — max 6 partner funds, 3 spots remaining.