The credit paper is the deal
In development finance, the credit paper is not just documentation — it is the deal, as far as the lender is concerned. A credit committee does not see your site, meet your contractor, or walk through your planning documents. They see the credit paper. Its quality determines their confidence, speed, and terms.
The question of how credit papers are produced — manually by brokers and analysts, or automatically by AI platforms — has become one of the most significant shifts in UK development finance. Here is how the two approaches compare.
Speed
Manual: A skilled broker or credit analyst typically takes 2–5 working days to produce a comprehensive credit paper, assuming all source documents are available. In practice, gathering missing information often extends this to 2–4 weeks.
AI: Platforms like Assesr generate a complete credit paper in hours from uploaded documents. The bottleneck shifts from writing time to document upload time.
Verdict: AI is 10–20x faster. For time-sensitive deals (site acquisitions with exchange deadlines, competitive processes), this speed advantage is decisive.
Accuracy
Manual: Depends on the individual. Experienced analysts rarely make calculation errors, but data transcription mistakes (copying the wrong figure from a QS report, misreading a planning condition) are common. These errors erode lender confidence and cause delays.
AI: Extracts data directly from source documents, eliminating transcription errors. Calculations are deterministic — sensitivity analysis, ratio calculations, and cost aggregations are always mathematically correct. The main risk is extraction errors from poorly formatted documents.
Verdict: AI is more consistently accurate for data extraction and calculation. Manual analysis is better for subjective judgements (market sentiment, borrower credibility). The best approach combines both.
Consistency
Manual: Every broker writes credit papers differently. Formatting varies, section ordering varies, depth of analysis varies. A lender receiving 10 manual submissions from 10 brokers gets 10 different formats to parse.
AI: Every credit paper follows the same structure, formatting, and analytical framework. Lenders who receive multiple AI-generated papers can compare deals at a glance — which is exactly what credit committees want.
Verdict: AI produces more consistent output, which benefits lenders and speeds up credit decisions.
Cost
Manual (via broker): The broker's time is embedded in their fee — typically 0.5–1.5% of the facility. On a £3m loan, that is £15,000–£45,000. On a £10m loan, £50,000–£150,000.
AI (via platform): Assesr is free for borrowers. The platform is compensated by the lender on successful drawdown. Other platforms may charge subscription or per-deal fees, but typically far less than broker fees.
Verdict: AI platforms are significantly cheaper, especially on larger deals where percentage-based broker fees become substantial.
Lender reception
This is the metric that matters most. Early scepticism about AI-generated credit papers has given way to preference in many cases. Lenders report that AI-generated papers are typically more complete (fewer follow-up questions), more consistently structured (easier to compare), and more accurate on numbers (fewer inconsistencies to investigate).
The lender does not care whether the credit paper was written by a human or a machine. They care whether it is complete, accurate, well-structured, and makes the deal easy to assess.
When manual still wins
- Highly bespoke deal structures — unusual legal arrangements, complex multi-party transactions, or deals with significant narrative complexity.
- Relationship-driven lending — situations where the broker's personal endorsement carries weight with a specific lender.
- Post-submission negotiation — a broker can negotiate terms, manage the relationship, and handle difficult conversations in a way that technology cannot.