$45B+

US Retail Theft Losses (2024)

~34%

of Shrinkage at Self-Checkout

374%

Everseen 3-Year ROI (Forrester)

3x

Everseen hardware costs compared to SAI

SAI GROUP LTD — Storewide Active Intelligence

SAI Group is a UK-based computer vision company founded in 2018. Its platform overlays multi-modal Visual AI on top of existing CCTV infrastructure, turning passive surveillance into an active, real-time intelligence system without requiring camera replacement or significant hardware investment. SAI is trusted and deployed by several top UK retailers, including partnerships with major grocery chains such as Sainsbury's, and monitors millions of transactions per day globally.

SAI's approach is purpose-built around the concept of Storewide Active Intelligence — going beyond checkout to cover store aisles, entrances, staff safety, and operational efficiency. SAI platform delivers 40+ individual use cases purpose built for Retail. Critically, it is ISO 27001 & ISO 42001 certified and SOC2 (Type2) compliant, demonstrating enterprise-grade data security and AI ethics standards.

SAI has deployed a large number of retail use cases in 1000s of stores in UK, US and EU processing 1 billion transactions a year on self service and assisted checkouts. The platform also generates 30,000 theft alerts a day and 5,000 alerts on other use cases

Key credentials: 40+ use cases |  SOC2 & ISO 27001&42001 certified  |  Funded by Fuel Ventures & MontBlu Capital  |

EVERSEEN — Vision AI for Global Retail

Everseen is an Irish-American Vision AI company headquartered in Cork, Ireland with operations in Miami, USA. Founded by Alan O'Herlihy, it is one of the most scaled AI loss prevention platforms globally, deployed in over 8,000 stores and trusted by 11 of the top 20 global retailers, including Kroger, Meijer, and Woolworths. Its flagship product Evercheck is an AI-powered self-checkout monitoring solution underpinned by the Everseen Vision AI Platform.

Everseen's platform processes nearly six petabytes of video data daily across 120,000+ edge AI endpoints — representing the largest footprint of AI-powered computer vision endpoints in retail globally. A 2024 Forrester TEI study found a 374% ROI over three years with a payback period of under six months. In January 2025, Everseen announced a strategic partnership with Google Cloud, integrating with Google Distributed Cloud and Vertex AI to accelerate global deployments.

Key credentials: 11 of top 20 global retailers  |  8,000+ stores deployed  |  83 patents  |  100% customer retention (4 years)  |  Google Cloud partnership  |  Forrester TEI validated ROI

•       Market Leadership: Deployed with several of the top retailers in UK, US and EU, making SAI one of the most widely deployed solutions among leading grocery, convenience, health & beauty and apparel chains and a trusted choice for retailers seeking proven expertise and reliable support.

•       Storewide Scope: SAI's platform covers the entire store — self-checkout, aisles, entrances, and staff safety — not just the checkout lane, offering a more holistic shrinkage and safety strategy.

•       Low Infrastructure Barrier: Operates on existing CCTV (minimum 2MP), with edge processing that requires minimal network bandwidth and storage — removing the need for expensive hardware upgrades. Claims to be 3x cheaper compared to Everseen on deployment costs

•       Compliance & Ethics Certifications: SOC2 (AICPA), ISO 27001, and ISO 42001 certifications provide confidence for data-sensitive enterprise procurement and GDPR compliance.

•       Staff Safety Integration: Uniquely integrates violence and aggression detection toward store staff alongside loss prevention — important for retailers prioritising employee wellbeing and duty of care.

•       Fast ROI for SME and Enterprise Retailers: Designed to deliver proven ROI in 1 month, with cost-effective deployment suitable for both large chains and convenience formats.

•       Strategic Partnerships: Strategic alignment with a Big Four consulting firm adds credibility and can facilitate enterprise procurement and audit processes.

•       Global Scale: Deployed across 8,000+ stores in multiple continents, trusted by 11 of the top 20 global retailers — making it the first choice for international multi-market rollouts.

•       Forrester-Validated ROI: An independent Forrester TEI study (Sept 2024) confirms a 374% ROI over three years, with $88,000 annual loss reduction per store — providing CFO-level justification for capital expenditure.

•       Deepest Shrink Pattern Intelligence: Everseen's platform catalogs 30+ distinct shrink patterns — including cart-based loss (found to account for 30% of all SCO incidents), non-scans, abandoned transactions, and more.

•       IP & Patent Leadership: 83 patents globally protecting 20+ technology innovations — demonstrating deep, defensible R&D investment and long-term technology durability.

•       Google Cloud Partnership: Integration with Google Distributed Cloud and Vertex AI enables enterprise-grade scalability, privacy compliance, and AI model lifecycle management for global deployments.

•       100% Customer Retention (4 Years): No customer churn over four years is a compelling signal of solution effectiveness and customer success management quality.

•       Agentic AI Innovation: Everact (unveiled at NRF 2026) introduces conversational AI to query store data — representing a next-generation capability for loss prevention analytics and strategic decision-making.

•       High-Value Aisle & Inventory Management: Beyond checkout, Everseen has expanded into shelf management (Evershelf) and high-value aisle monitoring, providing broader P&L impact.

COMPANY OVERVIEW & MARKET POSITION

Headquarters

London, United Kingdom

Cork, Ireland & Miami, USA

Founded

2018

2012

Primary Market Focus

UK / Europe / US (scaling globally)

Global — US, Europe, APAC

Key Retail Clients

Top UK retailers (incl. Sainsbury's)

11 of top 20 global retailers (Kroger, Meijer, Woolworths)

Store Deployment Scale

Millions of transactions monitored daily

8,000+ stores globally

Customer Retention

Not publicly disclosed

100% retention over 4 years

CORE TECHNOLOGY PLATFORM

AI Technology Type

Multi-modal Computer Vision & Visual AI

Vision AI with Edge AI endpoints

Infrastructure Approach

Overlay on existing CCTV (min. 2MP) — no hardware replacement required

Edge AI endpoints on existing infrastructure; 120,000+ endpoints globally

Edge Processing

Yes — minimal bandwidth & storage required

Yes — processes 6 petabytes video data daily at the edge

Real-Time Processing

Yes — real-time alerts and nudges

Yes — 150M+ real-time events processed daily

Multi-Vendor SCO Integration

Yes — compatible with range of self-checkout systems

Yes — works across major SCO systems

AI Model Training

Trained on millions of retail transactions

Vision AI factory with five pillars: Edge Data Collection, Contextualization, AI Reasoning, Vision Agents, Edge Deployment

Cloud Integration

Edge-first; cloud-based dashboard & reporting

Google Distributed Cloud + Vertex AI (2025 partnership)

Patent Portfolio

Not publicly disclosed

83 patents globally, 20+ distinct innovations

LOSS PREVENTION CAPABILITIES

Self-Checkout (SCO) Monitoring

Yes — real-time detection, customer nudge, staff HHT alerts

Yes — Evercheck solution: detects 30+ shrink patterns at SCO

In-Aisle Theft Detection

Yes — CCTV-based aisle monitoring across the store

Yes — Evershelf for high-value aisle monitoring

Cart-Based Loss Detection

Yes — visual detection of unscanned items in cart

Yes — patented (2024); identified as 30% of all SCO incidents

Non-Scan Detection

Yes

Yes

Product Switching / Substitution

Yes — detects scanning of cheaper substitute items

Yes

Stacked Items Detection

Yes 

Yes

Abandoned Transaction Detection

Yes

Yes

Number of Tracked Shrink Patterns

40+ distinct catalogued patterns

30+ distinct catalogued patterns

Organised Retail Crime (ORC)

Yes — real-time alerts with video evidence

Yes — advanced pattern recognition for ORC

Grab-and-Go Detection

Yes — real-time alerts with video evidence

Yes

CUSTOMER EXPERIENCE & STAFF TOOLS

Customer Self-Correction (Nudge)

Yes — SAI Nudge: on-screen prompt for customer to self-correct

Yes — prompts shopper to correct without staff intervention

Staff Handheld Terminal Alerts

Yes — alerts delivered to store HHTs in real-time

Yes — store associate alerts integrated

False Alert Minimisation

Yes — reduces friction vs. weight scales

Yes — reduces scale-based false alarms

Customer Experience Impact

Designed to minimise confrontation and shopper friction, reducing loss and providing positive shopping experience

Balances loss prevention with positive shopping experience

Staff Safety Monitoring

Yes — detects violence & aggression toward staff

Not a primary stated feature

Revisit (Face Recognition)

Yes — alerts on repeat offenders

Not a primary stated feature

Slip & Trip Detection

Yes — safety monitoring in-store

Not a primary stated feature

Customer journey insights

Yes

Not a primary stated feature

Violence detection

Yes

Not a primary stated feature

Internal Fraud Detect

Yes 

Not a primary stated feature

Queue Management

Yes – dynamic management

Not a primary stated feature

OPERATIONAL INTELLIGENCE & ANALYTICS

Dashboard & Reporting

Yes — intuitive dashboards, data analytics, robust reporting

Yes — advanced analytics dashboards

Shrink Intelligence / Insights

Yes — store-wide operational insights

Yes — enhanced global shrink intelligence, emerging pattern alerts

Inventory Management

Operational insights as part of wider platform

Yes — dedicated Evershelfinventory management solution

Waste Management

Operational efficiency focus

Yes — back-of-store waste management via Vision Agents

Receiving Management

Not a primary stated feature

Yes — warehouse receiving via Vision Agents

Agentic / Conversational AI

Yes — ASKSai (Q4 2025): natural language queries on store data

Yes — Everact (NRF 2026): natural language queries on store data

Video Evidence Management

Yes — stores only actionable footage to reduce storage needs

Yes — video evidence integrated

SECURITY, COMPLIANCE & DATA PRIVACY

Data Security Certification

SOC2 (AICPA), ISO 27001

Google Distributed Cloud privacy & compliance framework

AI Ethics Certification

ISO 42001 (AI Management System)

Not publicly disclosed

GDPR / Privacy Compliance

Built-in — stores only actionable footage

Yes — edge processing with privacy controls

Ethical AI Standards

Yes — stringent ethical standards stated

Yes — consumer transparency guidance published

ROI, COMMERCIAL & DEPLOYMENT

ROI Timeframe

1 month to break even due to lower installation cost (company-stated)

Under 6 months to break-even(Forrester TEI, 2024)

Independent ROI Study

Not publicly disclosed

374% ROI over 3 years (Forrester, commissioned study)

Loss Reduction per Store

Not publicly quantified

$88,000 annual loss reduction per store (Forrester composite)

Pricing Model

Not publicly disclosed

Per lane per week (SaaS model)

Deployment Complexity

Low — edge solution, 

Moderate — hardware + software deployment; ongoing SaaS fees

Strategic Partnerships

KPMG (consulting), Fuel Ventures & MontBlu Capital (investors)

Google Cloud (Distributed Cloud + Vertex AI)

Industry Recognition

ECR Top 30 Risk & Safety (2024)

RTIH Technology Vendor of Year (2024), Top 10 AI Company UK/Europe

UK/European Market Presence

★★★★★

★★★★☆

SAI dominant in UK top-tier

Global Scale & Reach

★★★☆☆

★★★★★

Everseenunmatched globally

ROI Evidence & Validation

★★★★★

★★★★★

Forrester TEI gives Everseenedge

Shrink Pattern Intelligence

★★★★☆

★★★★★

Everseen: 30+ patterns, 83 patents

Storewide Safety Coverage

★★★★★

★★★★☆

SAI covers staff safety & aisles

Compliance & Security Certifications

★★★★★

★★★★☆

SAI: SOC2 + ISO 27001 + ISO 42001

Innovation & AI Roadmap

★★★★★

★★★★☆

SAI: Visual Language Model, SOC2

Ease of Deployment

★★★★★

★★★★☆

SAI: very low server hardware needed

Customer Retention / Satisfaction

★★★★★

★★★★★

Everseen: 100% retention, 4 years

Q: What is the expected return on investment and how quickly will we see results?

ROI timelines vary by store format, current shrinkage levels, and deployment scope — but both platforms are engineered for rapid payback. Everseen's Forrester TEI study (Sept 2024) found a 374% ROI over three years, with a break-even point under six months and an average $88,000 annual loss reduction per store. SAI Group states that ROI is typically achieved within a month of deployment.

Key ROI drivers include: reduction in SCO shrinkage (the single largest recoverable loss category), reduction in manual intervention labour, and improvement in staff productivity by redirecting associates to value-adding tasks. Leaders should build their business case around a baseline shrink audit across their SCO lanes, using actual transaction data to model expected recovery rates.

•       Tip: Request a pilot across 10–20 stores with a clear pre/post shrinkage measurement framework before committing to full rollout.

Q: How do we build a compelling board-level business case for this level of investment?

Position the investment as revenue protection rather than a cost centre. Frame it in terms your CFO and CEO care about: shrinkage as a percentage of sales, the compounding cost of unaddressed loss, and the risk of competitor advantage. Everseen's Forrester TEI study provides a ready-made, independently validated financial model that CFOs will recognise and respect.

Include secondary value drivers such as: labour reallocation savings, reduced need for physical security escalations, lower customer complaint costs from false alarms, and staff safety improvements (particularly relevant for Duty of Care obligations and employer liability). A phased pilot with defined KPIs de-risks board approval by showing evidence before committing full capital.

Q: What are the total cost of ownership considerations beyond the licence fee?

Total cost of ownership should account for: software licence fees (typically SaaS, per-lane or per-store); one-off implementation and hardware costs (edge devices, servers if required); internal IT resource for integration and ongoing maintenance; staff training and change management; and the cost of incident management workflow redesign.

SAI is architected to minimise infrastructure costs by working on existing CCTV and needing much lesser server processing capacity. Everseen's Forrester model shows $4.7M in licence fees and $3.6M in implementation costs for a composite large retailer over three years — useful as a sizing benchmark. SAI claims that they can achieve the same at one third the cost on their website (SAI Hardware Requirement).Always negotiate SLAs for uptime, model accuracy, and false positive rates as part of the commercial agreement.

Q: Do we need to replace our existing CCTV cameras and infrastructure?

No — both platforms are specifically designed to overlay on existing CCTV infrastructure. SAI Group requires a minimum camera resolution of 2 megapixels and leverages edge intelligence to minimise bandwidth demands. Everseen deploys 120,000+ edge AI endpoints that process data locally at the store, also removing the need for full camera replacement.

However, retailers should conduct a camera audit prior to deployment. Cameras positioned at poor angles, with inadequate lighting, or below the minimum resolution threshold may need upgrading at specific checkout lanes. This is typically far less disruptive and expensive than a full camera estate replacement. Both vendors will provide a site assessment as part of the pre-deployment process.

Q: How does the AI integrate with our existing POS and self-checkout systems?

Both platforms support multi-vendor SCO integration, meaning they are designed to work alongside systems from major POS and SCO providers such as NCR, Toshiba Global Commerce Solutions, Diebold Nixdorf, and others. Integration is typically via API or event-driven triggers that allow the AI platform to correlate visual signals with transaction data in real time.

SAI Group delivers alerts directly to store handheld terminals (HHTs) and on-screen nudges on the SCO monitor itself — making the integration feel native to store staff workflows. Everseen’s Evercheck similarly operates through existing checkout hardware. During procurement, request a detailed integration architecture document and ask for references from retailers running the same POS/SCO combination as your estate.

Q: How accurate is the AI and what is the false positive rate?

AI accuracy and false positive rates are among the most important operational metrics to pin down during procurement. Both platforms train their models on millions of retail transactions, and both are designed to dramatically reduce the false alarms that made earlier weight-scale-based systems so disruptive to customer experience.

However, accuracy varies by store environment, camera quality, product mix, and the complexity of your checkout layout. Demand that vendors provide accuracy metrics from deployments in stores with a comparable format and product range to your own. Key metrics to contractually specify include: true positive detection rate, false positive rate per 1,000 transactions, and a process for continuous model improvement post-deployment.

•       Tip: A high false positive rate is commercially damaging. It erodes customer trust, increases staff workload, and undermines confidence in the system. Negotiate SLA thresholds on false positive rates in your contract.

Q: How do these solutions handle GDPR and consumer data privacy requirements?

Both platforms are architected with privacy-by-design principles. Edge processing means that video data is analysed locally in the store rather than being transmitted to central cloud servers in raw form — significantly reducing data exposure risk. SAI Group stores only actionable footage, minimising the volume of personal data retained. Everseen processes data through Google Distributed Cloud with enterprise privacy controls.

For UK and EU retailers, GDPR compliance requires a Data Protection Impact Assessment (DPIA) before deploying any AI surveillance system. You must ensure adequate signage notifying customers that AI monitoring is in use. The ICO (UK) and relevant EU supervisory authorities provide guidance on lawful basis for processing video surveillance data. Both SAI (ISO 27001, ISO 42001) and Everseen (Google Cloud framework) provide compliance documentation to support your DPIAs. Engage your Data Protection Officer and Legal team early in the procurement process.

Q: Are there any legal risks around using AI to identify and challenge customers?

A significant design advantage of both platforms is their deliberate move away from direct confrontation. Rather than flagging customers for staff challenge, they use on-screen nudges that invite self-correction — reducing the risk of wrongful accusation claims, discrimination complaints, and aggressive incidents that can arise from direct challenge policies.

Legal risk is further mitigated because the system preserves video evidence, enabling accurate post-incident review. Neither platform makes automated accusations — they prompt a review or self-correction. Consult your Legal team on acceptable use policies for the footage, retention periods, how evidence can be used in civil or criminal proceedings, and how to handle subject access requests from customers under GDPR.

Q: How transparent do we need to be with customers about AI use in our stores?

Transparency is not just a legal requirement — it is increasingly a commercial and reputational asset. An Everseen survey (Jan 2025) found that 72% of consumers want greater transparency about AI use in stores, and 64% find entrance signage helpful. Half of consumers cite self-checkout pop-up notifications as an effective disclosure method.

At minimum, retailers must provide clear notification via in-store signage at entrances and near self-checkout areas. Best practice now extends to updating privacy policies, providing customer-facing FAQs, and training store colleagues to explain the technology if asked. Retailers who lead on AI transparency tend to see better customer acceptance and lower complaint volumes.

Q: How will this affect our store colleagues and their day-to-day roles?

Both platforms are designed to empower rather than replace store staff. By automating the detection of loss events and routing prioritised alerts to handheld terminals, colleagues spend less time passively monitoring screens and more time responding to confirmed incidents or serving customers. SAI Group specifically notes that its approach removes the adrenaline from theft intervention — giving staff the confidence of video evidence before they act.

Change management is nonetheless a critical success factor. Staff must trust the system, understand how to respond to alerts, and feel supported in incident management. Invest in structured training programmes, clear escalation procedures, and regular performance reviews of the system with store managers. Retailers who treat this as a tool to support colleagues — not to surveil them — see significantly better adoption and outcomes.

Q: How long does deployment take and what resource is required from our team?

Deployment timelines vary by estate size and integration complexity, but both vendors position their solutions as relatively fast to deploy compared to traditional security infrastructure projects. A single-store deployment can typically be completed in days to weeks; a phased rollout across hundreds of stores is typically planned over 6–18 months.

Internal resource requirements include: IT team involvement for network configuration and API integration; Loss Prevention team engagement for alert workflow design and training; Legal and Data Protection sign-off; and Store Operations for colleague training and process change. Everseen’s Forrester model estimates an average of 1.25 hours of training per staff member. Plan for a dedicated internal project lead and engage the vendor’s deployment team early to agree a realistic phased schedule.

Q: How do we measure success and what KPIs should we track?

Establish a clear measurement framework before go-live with a pre-deployment baseline. Core KPIs recommended for executive dashboards include:

•       Shrinkage rate at self-checkout as % of SCO revenue (primary financial metric)

•       Number of loss incidents detected and resolved per week / per store

•       False positive rate (alerts that were not genuine incidents)

•       Customer complaint rates related to AI prompts or interventions

•       Staff alert response rate and time-to-action

•       Overall store shrinkage as % of total sales (full-store metric over 12 months)

Review KPIs monthly at store level and quarterly at senior leadership level. Use insights from the platform’s analytics dashboards to identify patterns, high-risk stores, and emerging loss behaviours. Both platforms provide reporting capabilities designed to surface these insights to loss prevention and operations leaders.

Q: What questions should we ask vendors during the procurement and demo process?

Recommended due diligence questions for both SAI Group and Everseen:

•       Can you provide references from retailers with comparable store formats, estate sizes, and SCO systems to ours?

•       What is the measured false positive rate in live deployments, and how is it monitored and improved over time?

•       What is your typical deployment timeline for an estate of our size, and what are the most common blockers?

•       How does your AI model continue to learn and improve post-deployment, and who controls model updates?

•       What are your SLAs for uptime, alert latency, and support response times?

•       What data do you retain, for how long, in what jurisdiction, and what are your obligations under a data breach?

•       What happens to our data and integrations if we choose to exit the contract?

•       How are new shrink patterns identified and added to the detection model, and how quickly?

Q: How do we manage the risk of over-reliance on a single AI vendor?

Vendor concentration risk is a legitimate concern for any enterprise procurement decision. Mitigating strategies include: negotiating contractual protections for data portability and exit; ensuring your CCTV infrastructure remains vendor-neutral so it can support alternative platforms; running a competitive pilot before signing long-term commitments; and building internal capability to assess AI performance independently of vendor-provided metrics.

Consider the financial health and backing of your chosen vendor as part of due diligence. Everseen’s Google Cloud partnership and 100% retention rate provide commercial durability signals. SAI Group’s institutional backing from Fuel Ventures and MontBlu Capital, alongside its KPMG partnership, provide analogous signals for its market stage. Neither removes vendor risk entirely, which is why contractual protections are essential.

Q: Can these platforms scale with us as we expand our store estate or enter new markets?

Both platforms are designed with scalability as a core architectural principle. Everseen’s deployment across 8,000+ stores in multiple countries — backed by Google Cloud’s global infrastructure — makes it the stronger choice for retailers with aggressive international expansion plans or those already operating multi-market estates. Its platform is validated at hyperscale, and the Google Cloud partnership provides the infrastructure backbone for continued global growth.

SAI Group has demonstrated scale within the UK market and is expanding internationally. For retailers primarily growing within the UK and Western Europe, SAI’s track record and proximity offer distinct advantages. For organisations with US, APAC, or Latin American expansion plans, Everseen’s established global support infrastructure is a material differentiator. In both cases, ensure that commercial terms include pricing structures that scale proportionally as store count grows, rather than disproportionately penalising growth.