ComplyLens Accelerates RegTech Product Development with SourceFuse
How a fast-moving startup partnered with SourceFuse to design, build, and launch a cloud-native compliance product with speed, flexibility, and execution discipline.
Overview
ComplyLens is a UK-based regtech startup focused on helping financial institutions manage regulatory risk. The company built its product to address a major gap in the market, that is, risk assessment processes in financial services remain highly manual, spreadsheet-driven, and difficult to scale or audit.
When ComplyLens decided to turn its concept into a software platform, it faced a foundational choice of spending months hiring and assembling an internal technical team or partnering with an experienced product engineering firm that could accelerate execution.
The company chose SourceFuse. That decision gave ComplyLens immediate access to product architecture, business analysis, UI/UX, development, QA, and cloud engineering skills, allowing the startup to move quickly from idea to product build and position itself for early client adoption.
The Challenge
A legal requirement stuck in spreadsheets
Financial institutions are required to run regulatory risk assessments. In most firms, that still happens in fragmented, spreadsheet-based workflows.
No real auditability
Spreadsheet processes are hard to trace and hard to defend when an auditor or regulator asks how a risk decision was made.
Too much key-person dependency
The process leaned on a few individuals who held the knowledge. That is fragile, and it does not scale as a firm grows.
A market opportunity with a clock on it
ComplyLens had early, positive signals from prospective clients. Turning that momentum into a working product meant moving fast, not losing quarters to setup.
The cost of building a team in-house
Doing it internally meant first hiring an architect, business analyst, full-stack developers, QA, and DevOps: months before a single feature got built.
Speed versus specialisation, at the same time
The build needed architecture, design, engineering, QA, and cloud expertise together from day one. Assembling all of that internally, quickly, was the real bottleneck.
The Solution
SourceFuse partnered with ComplyLens to design and build AssessLens, a cloud-native regtech product for financial services. This was a net-new build, not a legacy modernization. The job was to take a startup from concept to product readiness with the technical foundation and delivery pace to execute fast.
A team stood up, and a product was defined
The work started with rapid team formation and early product definition, so ComplyLens moved into building without the usual hiring lag.
Architecture and cloud foundation on AWS
SourceFuse set up architecture planning, database design, application foundations, and the AWS environment: the base a scalable SaaS platform needs from day one.
Design early enough to sell with
UI/UX work produced real product designs early, giving ComplyLens something concrete to put in front of prospects and validate while the platform was still being built.
An iterative, startup-friendly build
Development ran in an iterative model suited to a startup, moving without waiting for perfect upfront specs and refining features as it went.
AI-assisted development for speed
AI-assisted development practices were introduced during the engagement to increase development speed and productivity, enabling faster iteration.
A structured handover to their own team
As ComplyLens began building its internal product team, SourceFuse led transition planning, handing over documentation, source code, processes, and product knowledge in a structured, collaborative way.
The Impact
Cross-functional team building within two weeks
MVP kept on a 4 to 6 month track
Net-new SaaS platform, built cloud-first on AWS
Early designs used to validate with prospects
Secure, enterprise-ready foundation from day one
Delivered readiness for first live client deployment
Smooth handover to ComplyLens’s own team
Hybrid model: internal team plus on-demand expertise
AWS Services Used
- Amazon EC2 (Elastic Compute Cloud)
- Amazon ECS (Elastic Container Service)
- Amazon ECR (Elastic Container Registry)
- Amazon VPC (Virtual Private Cloud)
- Elastic Load Balancer (ELB / ALB)
- Amazon CloudFront
- Amazon RDS (PostgreSQL / Aurora)
- Amazon ElastiCache (Redis)
- Amazon S3 (Simple Storage Service)
- AWS KMS (Key Management Service)
- AWS WAF (Web Application Firewall)
- AWS GuardDuty
- AWS Security Hub
- AWS Secrets Manager / Parameter Store
- Amazon CloudWatch
- Amazon SES (Simple Email Service)
- Amazon SNS (Simple Notification Service)
- AWS CodePipeline (or Jenkins-integrated pipelines)
Technology Highlights
Cloud-Native SaaS Architecture:
Multi-tenant AWS-based design enabling scalability, efficiency, and tenant-level data isolation.
Scalable Infrastructure:
Auto-scaling, load balancing, and caching ensure high availability and performance under varying workloads.
Modern Full-Stack Stack:
A react frontend and LoopBack 4 backend enable rapid, modular, and maintainable development.
API-First Design:
REST APIs provide flexibility for integrations and future extensibility.
Enterprise-Grade Security:
RBAC, JWT authentication, encryption, and AWS security services ensure robust protection.
High-Performance Data Layer:
PostgreSQL and Redis caching deliver reliable storage with optimised response times.
CI/CD Automation:
Automated pipelines enable faster releases, consistent deployments, and improved code quality.
Observability & Monitoring:
Real-time logging and monitoring ensure system reliability and proactive issue resolution.
AI-Assisted Development:
AI-assisted development practices were introduced during the engagement to enhance development speed and productivity, enabling faster iteration in a startup environment.
Compliance-Ready Design:
Built to support audit trails, risk modelling, and regulatory reporting requirements.
The Next Phase: Building Regulatory Intelligence with Claude
As ComplyLens evolved from a digital risk assessment platform into a broader regulatory intelligence platform, a new challenge emerged: how to scale regulatory coverage globally.
Financial institutions operate across multiple jurisdictions and regulatory regimes. To help compliance teams understand changing risks, ComplyLens needed to continuously ingest and analyse regulatory enforcement actions and guidance from regulators around the world.
The original approach could not scale. Each regulator required engineers to manually understand how its website was structured, how content was published, how pagination worked and the terminology used. Scrapers were built individually around regulator-specific page templates. New regulators could take hours or days to onboard, while website changes could break existing extraction logic.
ComplyLens turned to SourceFuse to help redesign the regulatory intelligence pipeline using Claude, Anthropic’s AI models running on Amazon Bedrock. The objective was not simply to add AI classification. It was to use Claude to reduce the engineering effort required to discover, onboard, maintain and understand regulatory information at scale.
From Regulator-by-Regulator Engineering to AI-Assisted Onboarding
SourceFuse introduced Claude across the regulatory intelligence pipeline, rebuilding one workflow from the ground up and enhancing three existing workloads.
Automated regulator onboarding
Previously, onboarding a new regulator required an engineer to manually analyse the regulator’s website and create regulator-specific scraping logic.
The new approach uses Claude to accelerate this discovery process. Given a regulator’s seed URL, Claude can analyse the site’s structure, identify relevant content patterns and pagination, and help generate the extraction logic required to onboard the source.
The result: new regulators can be onboarded in minutes rather than hours or days of manual engineering.
This significantly reduces one of the biggest scaling bottlenecks in building global regulatory intelligence coverage.
AI-assisted recovery from website changes
Regulatory websites change frequently.
Under the original approach, a change in page structure could cause an existing scraper to stop extracting content correctly. In some cases, the pipeline could continue running while producing incomplete or blank records.
SourceFuse enhanced the pipeline using Claude. Existing extraction rules are cached and used for normal processing. When the system detects that a regulator’s page structure has changed or extraction is no longer working as expected, Claude can analyse the updated page and help identify the relevant article links and revised extraction rules. The updated rules can then be stored and reused.
Instead of every website change automatically becoming an engineering problem, the system can adapt its extraction approach to changes in the source.
Expanding regulatory document coverage
Regulatory intelligence is not published in one consistent format.
Important information may appear in:
- Web pages
- PDFs
- DOC and DOCX files
- Scanned documents
SourceFuse enhanced the extraction pipeline to process multiple document formats, including OCR where required. This brought regulatory publications that were previously outside the scope of the pipeline into coverage, including PDF-only content.
Turning unstructured regulatory content into structured compliance intelligence
Once regulatory content is ingested, Claude is used to classify and structure it against the ComplyLens regulatory risk taxonomy. For each enforcement action or piece of regulatory guidance, the system can identify relevant attributes across multiple compliance dimensions, helping connect unstructured regulatory content to the risks faced by financial institutions. Classification outputs are subject to confidence thresholds. Cases where confidence is insufficient can be routed for human review rather than automatically accepted.
Classification results, supporting metadata and rationale are retained to support review and auditability.
The resulting intelligence can then be connected to:
Regulator → Jurisdiction → Regulation → Compliance Domain → Risk Factor → Enforcement or Guidance
This structured intelligence forms part of the foundation for more context-aware regulatory risk assessments.
The Results
The new pipeline was measured against the legacy approach over the same date window.
122% increase in regulatory coverage
495 → 1,100 records
More regulatory content identified and brought into the ComplyLens intelligence platform.
100% of legacy records maintained
The migration increased coverage without losing records already captured by the existing pipeline.
~99% reduction in blank record descriptions
161 → 2
Significantly improving the quality and usability of extracted regulatory data.
Hours or days → minutes
New-regulator onboarding was dramatically accelerated.
From Eight Regulators to a Global Regulatory Intelligence Platform
The engine is running in production today across 20+ global regulators, with an architecture designed to scale significantly further. Its output is already surfaced within AssessLens, providing regulatory intelligence and risk factor insights that can support customer risk assessments.
For ComplyLens, the use of Claude represents an important shift in the product’s evolution.
The goal is not simply to collect more regulatory data.
It is to build an intelligence layer capable of connecting:
Financial institutions → jurisdictions → regulators → regulations → risks → enforcement actions and regulatory guidance
By using Claude to help automate the discovery, extraction and classification of regulatory information, ComplyLens is building towards a platform where regulatory intelligence can continuously inform how financial institutions understand and assess their regulatory risks.
Looking Ahead
ComplyLens now has its own core team in place and continues to work with SourceFuse as an extension of that team. This hybrid model gives the company the flexibility to scale support up or down based on product priorities, client demands, and specialised skill requirements.
As the company continues to grow in the regulatory technology space, it is positioned with a live product in the market, a scalable cloud-based foundation, a more mature internal operating model, and a trusted partner that continues to support future development.
For startups building ambitious B2B products in regulated industries, the ComplyLens story shows the value of choosing the right product engineering partner early, especially when speed, specialisation, and execution quality all matter at once.
