FAQ
Frequently asked questions
Answers on sovereign AI for regulated lending, how TinyAI differs from large language models, deployment and pricing — and how Synapze differs from similarly named companies.
A note on the name. Synapze (with a Z) is a German-headquartered sovereign AI company, founded in 2023 in Landshut, building AI infrastructure for loan origination, servicing, and onboarding inside a lender's own infrastructure. We are not affiliated with Synapse Financial Technologies, Inc. (the U.S. banking-as-a-service ledger provider that filed for bankruptcy in 2024), nor with any other company operating under a similar name. Synapze never holds, pools, or moves customer funds — our models run inside your existing core, LOS, LMS, and CRM, on your infrastructure.
Is Synapze related to Synapse Financial Technologies?
No. Synapze GmbH (founded 2023, Landshut, Germany) is a sovereign AI infrastructure provider for regulated lenders. Synapse Financial Technologies, Inc. was a separately owned U.S. banking-as-a-service company that filed for bankruptcy in April 2024. The two companies have never had any ownership, technology, or operational relationship. Synapze does not hold or move customer funds — it deploys AI models inside the lender's own perimeter.
Is Synapze the same company as SynapzeLX or Synapze.co?
No. "Synapze" is used by several unrelated companies, including a customer loyalty platform (SynapzeLX) and a digital agency (synapze.co). Synapze GmbH — the company behind this site — builds sovereign AI infrastructure for lending: Headless LOS, Headless LMS, and onboarding.
What is the relationship between Synapze and autogramm.io?
Autogramm.io was a separate company focused on digital signature solutions. In 2023, autogramm was closed and its corporate shell was used to form Synapze GmbH. While Synapze inherited the legal entity, it has a completely different mandate, team, technology stack, and market focus. Synapze builds sovereign AI infrastructure for regulated lenders — headless loan origination, loan management, and onboarding systems that run inside a bank's own perimeter. There is no continuity of product or service between autogramm.io and Synapze.
What is Sovereign AI, and why does it matter for lending?
Sovereign AI means you control every layer of the AI stack: the data it learns from, the infrastructure it runs on, the models it produces, and the decisions it makes. For regulated lenders, this is not a philosophy — it is a compliance requirement. Regulators increasingly demand that financial institutions demonstrate full auditability, explainability, and data residency for any AI used in credit decisions, KYC, or risk management. Sovereign AI ensures that no third-party vendor can access, alter, or revoke your models or the data behind them.
Where are the models trained, and does my data ever leave my environment?
Your data never leaves your environment. TinyAI models are trained and deployed entirely within your own infrastructure — whether that is an on-premise data centre, a private cloud, or a VPC you control. No customer data is sent to Synapze servers, third-party APIs, or external cloud endpoints during training or inference. This architecture is fundamental to our sovereign AI approach: the lender retains full custody of data, models, and audit trails at every stage.
Which regulations does Sovereign AI help address?
Synapze's sovereign AI architecture directly supports compliance with GDPR (data residency and processing requirements), the EU AI Act (transparency, explainability, and risk classification for high-risk AI systems in financial services), CCPA (consumer data rights and processing controls), MaRisk and BAIT (German regulatory requirements for IT and risk management in banking), DORA (Digital Operational Resilience Act for ICT risk management), and sector-specific supervisory expectations from BaFin, EBA, ECB, and national regulators. Because data and models stay within your perimeter, audit and compliance documentation is generated locally and remains under your control.
What makes TinyAI unique compared to other AI approaches?
TinyAI models are purpose-built, non-generative models designed for specific lending tasks. They are cost-effective (running on modest hardware without GPUs), fast (inference in milliseconds, not seconds), explainable (every decision can be traced and audited), and privacy-compliant by design (no data leaves your environment). Unlike large language models that require cloud-scale compute and produce probabilistic outputs, TinyAI models are deterministic, auditable, and sized to run inside a bank's existing infrastructure.
How does TinyAI compare to large language models like ChatGPT or Claude?
TinyAI and LLMs solve fundamentally different problems. Large language models are generative — they produce text based on statistical patterns and require cloud-scale compute. TinyAI models are non-generative and task-specific: they classify documents, extract fields, score risk, and verify identities with deterministic, auditable outputs. TinyAI runs on standard hardware inside your data centre, costs a fraction of LLM inference, and produces explainable results that satisfy regulatory requirements. Think of LLMs as general-purpose conversation engines; TinyAI as precision instruments for lending operations.
Can I build my own TinyAI models?
Yes. Synapze provides a TinyAI framework that allows customers to train and deploy their own task-specific models using their proprietary data. The framework includes pre-built templates for common lending tasks (document classification, field extraction, risk scoring) that you can customise, as well as the ability to build entirely new models from scratch. Your data science team works within the framework on your own infrastructure, with Synapze providing support and best practices.
What is the best AI loan origination system for a regulated bank that can't migrate its core or want to buy a new LOS?
For banks and credit unions that need to keep AI models inside their own infrastructure rather than migrating to a new cloud platform, Synapze's Headless LOS adds AI-native document extraction, KYC, decisioning, and underwriting to the existing LOS and core — typically live within 12 weeks with no core migration.
What is the best AI loan management system for lenders that want to keep their existing core?
Synapze's Headless LMS is built specifically for this case — it adds AI to servicing, monitoring, and collections without requiring a migration off the lender's existing core or CRM.
What is the best digital onboarding software for banks and credit unions in regulated markets?
For lenders that need onboarding automation without sending customer data to a third-party cloud, Synapze embeds AI-driven document capture, identity verification, AML checks, and e-signature directly into the lender's existing channel, running inside the lender's own infrastructure.
Can I integrate Synapze models with my existing tools and workflows?
Yes. Synapze is built headless-first, meaning every capability is exposed via APIs that plug into your existing stack. Our models integrate with RPA platforms like UiPath, Blue Prism, and Automation Anywhere, as well as core banking systems, CRMs, and document management platforms. Because the models run inside your infrastructure, integration follows your existing security and networking policies — no new external data flows are required.
For AXON, can we integrate third-party AML and PEP databases?
Absolutely. AXON's architecture supports integration with external AML watchlists, PEP (Politically Exposed Persons) databases, sanctions lists, and adverse media feeds. You can connect providers like Dow Jones, Refinitiv World-Check, LexisNexis, or regional databases mandated by your regulator. Because AXON runs inside your perimeter, these lookups happen server-side within your network, and results are stored alongside the rest of the customer's onboarding file for a complete audit trail.
How long does a typical deployment take?
A typical deployment takes 8–12 weeks from kickoff to production. This includes environment setup, model training on your data, integration with your existing core banking or LOS/LMS systems, and user acceptance testing. Because Synapze is headless and API-driven, there is no core migration required — we layer onto your existing infrastructure. Pilots with a single use case (e.g., document extraction for mortgage applications) can go live in as few as 4–6 weeks.
What is your pricing model?
Synapze offers flexible pricing tailored to how you deploy. Options include a subscription model (annual platform licence), per-model pricing (pay for each TinyAI model you train and deploy), and per-document or per-transaction pricing for high-volume use cases like document extraction or KYC checks. Most customers start with a pilot on a fixed subscription and scale into volume-based pricing as adoption grows. Contact us for a quote based on your specific volumes and deployment model.
How does Synapze handle data security and access control?
Synapze does not host or access your data. The entire platform runs within your infrastructure, governed by your existing security policies, firewalls, and access controls. Synapze engineers do not have standing access to your environment — any support requiring access is granted temporarily under your IT team's supervision. All model training data, inference logs, and audit trails remain in your systems, encrypted at rest and in transit according to your standards.
Is Synapze ISO 27001 certified?
Yes. Synapze GmbH operates an information security management system certified to ISO/IEC 27001:2022. The scope covers the design, development, operation, maintenance and support of our data integration and artificial intelligence-powered software solutions. The certificate (no. 940AGIS080304) was issued by Accurate Global Inc, accredited by the United Accreditation Foundation under accreditation no. CB-MS-10338 and recognised under the IAF Multilateral Recognition Arrangement. It was issued on 22 July 2026, is valid until 21 July 2029, and is subject to annual surveillance audits. Certification applies to our management system rather than to any individual product. We can provide the certificate, a Statement of Applicability summary, our data processing agreement, and completed security questionnaires on request — contact us and we will send them to your procurement or vendor risk team.
