PF & Co Site Intelligence — AI Supplier Due Diligence Pack

For RICS-Regulated Professionals
Document Version: 1.1 · Date: 5 May 2026
Classification: Public supplier due-diligence summary. For client and regulator review.

1. Company Overview

Company: PF & Co Holdings Ltd, trading as PF & Co Site Intelligence
Registered: England & Wales · Companies House no. 16649319
Registered office: 2 Queens Drive, Guildford, GU2 9PP
Services: AI-assisted site intelligence reports for English development sites, covering Planning, Feasibility, Viability and Constraints
Contact: info@pfandco.co.uk · 01483 363 020
Web: pfandco.co.uk

PF & Co Site Intelligence is not RICS-regulated. This document is provided to assist RICS-regulated professionals in fulfilling their supplier due diligence obligations under the RICS "Responsible Use of AI in Surveying Practice" standard (mandatory from 9 March 2026).

2. AI Systems Summary

2.1 What "AI-assisted" means

Our site intelligence reports are produced using an AI-assisted system:

Automated data acquisition (querying authoritative public data sources)
Automated analysis (interpreting data into risk assessments)
Automated initial report drafting (assembling findings into structured documents)
All outputs pass through a multi-stage quality assurance pipeline
All outputs are reviewed before issue by a named technical reviewer identified on the report issue sheet

AI produces the analysis and draft conclusions; it does not finalise them autonomously. A named technical reviewer checks and is accountable for every report before it is issued.

2.2 System scale

Metric	Count	Detail
Data Sources	130	Across 11 categories
QA Categories	24	Across 5 quality domains
Report Types	24	Covering planning, environmental, geological, heritage, transport, ecology, and construction

3. Data Source Categories

Our system draws on 130 authoritative datasets across 11 categories. Individual source counts per category, source names, and per-report mappings are proprietary.

Category	Type of data
Geology & Ground Conditions	Bedrock and superficial geology, borehole records, geohazard data, soil classification
Flood & Water	Flood maps, monitoring stations, defence assets, water company records
Heritage & Conservation	Listings, Heritage at Risk Register, conservation area boundaries
Ecology & Environment	Statutory designations, protected species records, priority habitats, ancient woodland
Planning & Land Use	Planning portals, local plan policies, Land Registry, Green Belt, Article 4 data
Mapping & Spatial	Base mapping, LiDAR elevation data
Climate & Energy	Climate projections, EPC register, distribution network capacity
Safety & Risk	UXO risk mapping, COMAH major hazard installations
Market & Infrastructure	Transactions, price indices, broadband, NSIP register, renewable energy data
Transport & Accessibility	Stops, timetables, rail frequencies, traffic data, cycle infrastructure
Amenity & Services	Schools, healthcare, retail, food hygiene data

All sources are authoritative government or regulated-industry datasets. We do not use unverified web scraping, social media data, or crowd-sourced information as primary inputs. Named providers include (inter alia) the Environment Agency, British Geological Survey, Historic England, Natural England, DEFRA, Ordnance Survey, HM Land Registry, and the Department for Transport.

4. How AI is used and controlled

AI assists with gathering source material from authoritative datasets, analysis, and preparing initial drafts. Every output then passes an automated quality-assurance pass and is checked by a named technical reviewer before issue. The detailed technical design of the system is proprietary and is not published.

5. Quality Assurance Pipeline

Every report passes through a 24-category QA pipeline before delivery. The pipeline covers five quality domains:

Accuracy — factual verification against source data, numerical precision, citation validity.
Regulatory compliance — planning policy cross-referencing (NPPF, local plan, SPDs), building regulations, environmental regulatory alignment.
Consistency — cross-report data agreement, terminology consistency, internal logic validation.
Formatting — document structure, visual asset validation (maps, charts, tables), table integrity.
Citation integrity — source attribution verification, data freshness confirmation, reference link validity.

Individual QA check names, pass/fail criteria, and scoring algorithms are proprietary.

6. Human Oversight Model

6.1 Technical review

Every report is reviewed before issue by a named technical reviewer. The reviewer is identified on the report issue sheet and checks for accuracy, regulatory relevance, and professional appropriateness.

6.2 Accountability

The named technical reviewer is accountable for the final report content. AI produces provisional conclusions, recommendations and risk ratings; these are reviewed and owned by the named reviewer, not finalised autonomously by the system.

6.3 No autonomous delivery

No AI-generated report is delivered to a client without human review. There is no automated delivery pathway that bypasses human oversight.

6.4 Feedback loop

Quality issues identified in delivered reports are fed back into the QA pipeline as new checks, ensuring continuous improvement.

7. Data Governance

7.1 Client data processing

Client data (site addresses, project details) is processed solely for the purpose of producing the ordered report.
Client data is not used to train, fine-tune, or improve any AI models.
Data is processed on secure infrastructure.

7.2 Data retention

Contact form data: retained for up to 12 months.
Order data: retained for 6 years (UK tax/accounting requirements).
Report processing data: retained only for the duration of report production.

7.3 Regulatory compliance

All data processing complies with UK GDPR and the Data Protection Act 2018.
See our published Privacy Policy for full details: /legal/privacy-policy.html.

7.4 Third-party AI services

Where we use third-party AI models or services, we assess them against data processing policies, security standards, and alignment with our governance framework.

8. Mapping to the RICS AI standard

The table below maps our documented practices to the four pillars of the RICS "Responsible Use of AI in Surveying Practice" standard. PF & Co Site Intelligence is not RICS-regulated; this mapping is provided to assist RICS-regulated clients in recording their supplier due diligence.

RICS Pillar	Our Practice
Transparency	Public AI Standards & Compliance page. AI disclosure at point of order. AI methodology section in every report. Client information rights in Terms of Service (Section 7.3).
Accountability	Named technical reviewer for every report, identified on the report issue sheet. No autonomous AI decisions. Human-in-the-loop at every stage.
Fairness	Consistent methodology is applied across all sites. Constraint thresholds, data sources and risk-scoring logic are documented, version-controlled and periodically reviewed. Outputs remain subject to human technical review before issue.
Data Governance	No model training on client data. Secure processing. GDPR-compliant data handling. Published Privacy Policy and AI Governance Policy.

9. Public Resources

Resource	URL
AI Standards & Compliance	pfandco.co.uk/ai-compliance
Trust & Data Sources	pfandco.co.uk/trust
Terms of Service (Section 7)	pfandco.co.uk/terms-of-service
Privacy Policy (Section 3A)	pfandco.co.uk/legal/privacy-policy.html

10. Contact

For questions about this document or our AI governance practices:

Email: info@pfandco.co.uk
Phone: 01483 363 020