Role 1321
Multimodal AI-Generated Content Detection 
1. Executive Summary
AI Aware is a deep-tech company that has developed unique and proprietary advanced AI-generated content detection tools. These have been funded by four separate UK Government scientific grants over the past 18 months and are now being launched with 7 initial pilots in the UK. Our initial market focus is high-stakes text verification for legal, education, and publishing, where provenance, authorship, and trust are essential.
Once commercially established in text analysis, we will expand commercially into detection of AI-generated audio and video, targeting cybersecurity, enterprise security operations centres, media companies, and government agencies concerned with synthetic media, impersonation attacks, and synthetic fraud.
Our long-term vision is a unified “Content Authenticity Firewall”—a multimodal platform that verifies text, audio, video, and images, providing real-time authenticity scoring and risk alerts across digital ecosystems.
2. The Problems We Solve
Organisations increasingly rely on digital content, yet AI-generated text and media has created systemic challenges:
- In legal settings: AI-written documents pose risks to evidentiary integrity, contract validity, and due diligence accuracy, and there are now cases reaching court with hallucinated legal citations leading to lost court cases and individuals being referred to legal standards boards.
- In academia: AI-assisted writing undermines academic standards. Many institutions accept some use of AI but want to monitor it so as to create maximum permissible usage levels.
- In publishing: originality verification is essential for copyright, editorial integrity, and author brand protection.
Audio & Video
- Voice-cloned calls enable financial/social engineering attacks.
- Deepfake videos threaten elections, reputations, and media trust.
- Enterprises lack centralised tools to detect synthetic media entering their communication channels.
Current solutions are fragmented, unreliable, and limited to single modalities.
3. Solution & Product Overview
Phase 1 - Text Authenticity Suite (Year 1)
A SaaS and API-based platform offering:
- AI-generated text detection
- Document provenance scoring
- Bulk content scanning for publishers and legal firms
- Verified audit trails for academic institutions
Core technologies:
Only company that combines two different types of models in an ensemble approach: machine learning with zero shot. Other companies do one or the other. Machine learning is good at spotting types of content trained on but brittle when confronted with new types of content. Zero shot is good with new models (e.g. a new AI LLM) as it uses the same building but not as good with types of content seen before.
We ensemble both types of models, and by combining them, improve on both, which is why we have more accurate results. Once a company has an approach, training is on that approach which is why difficult for other companies to switch even if they could figure out how to do it. We have performed training on over 10 million pieces of content which also creates a barrier.
Phase 2 — Multimodal Detection (Year 2–3)
Add detection pipelines for:
- AI-generated speech (spectral artifact analysis, converting speech to text and using our text detector and voice-model fingerprinting)
- Deepfake video (GAN artifact detection, frame-level anomaly detection, facial inconsistency mapping)
Output:
- Real-time authenticity scoring
- Deepfake threat alerts
- Spoofing-prevention APIs for cybersecurity systems
- Forensic analysis dashboards with chain-of-custody exports
Phase 3 — Enterprise Cybersecurity Integration (Year 3–5)
Develop the Content Authenticity Firewall, integrating:
- Email gateways
- Messaging platforms
- Cloud storage scanning
- Identity verification systems
- SIEM/SOAR integrations (Splunk, CrowdStrike, Microsoft Sentinel)
4. Market Analysis
Initial Markets (Text Detection)
- Legal industry
- Global legal tech market: ~£25B
- Immediate need for contract verification and evidence authenticity
- Education
- Global EdTech market: ~£120B
- Driven by post-GPT academic integrity requirements
- Publishing
- ~£100B market
- Publishers need to detect AI-written manuscripts and plagiarism-adjacent content
Expansion Markets (Video/Audio Detection)
- Cybersecurity (primary growth driver)
- ~£140 B market by 2026
- Growth in voice deepfake fraud and synthetic identity attacks
- Government & Public Sector
- Disinformation defence
- Media & Social Platforms
- Verification of user-generated content
- Financial Services
- Fraud prevention in communications
Competitive Advantage: A unified multimodal detection system where most competitors specialize in only one modality (text OR deepfake video). By combining models, we are more accurate.
5. Business Model
Primary Revenue Streams
1. SaaS subscriptions
- Tiered per-seat and per-volume text scanning subscriptions – aiming at averaging £15k per client
2. API-based usage billing
- Docs, audio seconds, video minutes processed
3. Enterprise contracts
- On-premise or private cloud deployment
4. Cybersecurity integration fees
- SIEM connectors, advanced analytics, SLA-backed monitoring
5. Professional forensics services
- Expert reports, legal certification, incident investigations
6. Technology & R&D Roadmap
Year 1
- Text detection engine MVP has been built and launched
- APIs for legal and educational platforms
- Continuing research work on audio and video
Enterprise admin dashboard 
Year 2
- Speech detection models
- Video deepfake detection platform
- Forensic reporting tools
Year 3–4
- Real-time detection agents for communication platforms
- Enterprise cybersecurity integrations
- Multimodal authenticity graph and risk scoring engine
7. Go-to-Market Strategy
Phase 1: Text Detection Launch
- Direct sales to legal firms, universities, publishers
- Partnerships:
- Learning management systems (Canvas, Blackboard, Moodle)
- Legal tech platforms and consultancies (Clio, LexisNexis alliances, E&Y, Microsoft)
- Editorial tools (Submittable, manuscript systems)
- Publish whitepapers demonstrating detection accuracy
Phase 2: Enterprise Deepfake Detection
- Target:
- Cybersecurity vendors
- SOC teams
- Financial institutions
- Publish quarterly “State of Deepfake Threats” to build authority
- Integration partnerships with:
- Security platforms (CrowdStrike, Okta, Sentine)
- Communications providers (Zoom, Teams, Slack)
Phase 3: Public Sector & Media
- Certification with government standards
- Engage with journalism and fact-checking organisations
8. Core Team
Stephen Harmston, CEO (Successfully built B2B Software businesses) 
Dr Tillman Weyde CTO (Research lead – Tillman is Director of the AI Research Centre at City St Georges 
Daryl Ramdin, Head of Engineering – (5 years’ experience of multimodal deep learning) A person smiling for a picture
Norrie Hernon, Head of Sales – (10 years’ experience of selling B2Ba at high price points). 
Jaine Green, Head of Marketing (5 years’ experience of marketing and product outreach) 
9. Financials
Full financial forecast is available on request
10. Funding Strategy
- Seed round (for 18 months): Raising £250k
- Year 2,3,4: R& D from grant funding and cash from sales to fund commercial expansion
- Full financial forecast available on request.
11. Investing NEDs
AI Aware is seeking 3 or 4 investing NED’s with broad sector experience and good networks who appreciate the value that the identification of AI generated content can bring to society.
12. Long-Term Vision
AI Aware will provide the foundational infrastructure for trust in digital content. As AI-generated media proliferates, every institution from courts to universities to large corporations will need robust AI authenticity verification.
Our mission:
To make the digital world verifiably authentic.
Our future:
A world where all content—text, images, videos, audio—carries trustworthy, verifiable provenance in real time.
How to apply:
Please reach out to This email address is being protected from spambots. You need JavaScript enabled to view it. to express your interest.
Closing Date: 4 weeks from posting.
