Subsidy Document Retrieval

Texterous developed a bespoke subsidy retrieval tool for a public sector client in the Netherlands, designed to streamline the subsidy search process and enhance proposal accuracy. By integrating Large Language Models with a vector-based database system, our solution provides instant access to relevant subsidies, simplifying grant applications and fostering alignment with subsidy requirements. This tool, designed for offline use and capable of self-hosting, represents our commitment to practical innovation and user-centric design.

Subsidy Document Retrieval

Problem Statement:

Approach and solution 

Impact achieved 

Internal capabilities and scope 

Anonymous; Netherlands; Public Sector

Client

We get asked this a lot

Contact us for a free consultation for us to answer all your open questions.
What if my organization has no experience with artificial intelligence?

You bring the industry knowledge, we bring the AI expertise. Together, we identify automation use-cases. Texterous takes over strategy, development, training and deployment. In addition, our workshops are a great way for organizational learning in AI adoption.

How do you work?

Fantastic question! As AI projects tend to be exploratory in nature, we follow an agile philosophy: Working in iterative sprints resulting in testable outcomes every two weeks. This is how we make sure the project is aligned with your business goals.

Can't I just employ someone inhouse?

You can, in fact you should! Will that be sufficient to adopt to the depths and breadths in a rapidly changing AI environment? Likely not. We know it is hard finding the right people in the space of large language models right now. There are a lot of self-proclaimed "AI specialists" out there - that just jumped the ChatGPT wave in 2023. Our people have extensive professional and academic expertise in the field - in fact, we have been among the first 6000 people globally to get access to GPT-3 in 2020. If you are looking to develop your in-house AI team and need assistance in doing so, we offer a CIO as a Service.

Do I need training data?

Large language models have been pre-trained to understand human language. In most cases, no additional training data is needed. However, one or few examples afford "in-context learning" and if you happen to possess training data, we are happy to fine-tune our models to your use-cases.

Are you GDPR compliant?

Yes! We are fully compliant with local AI and data protection requirements. Our server centers live within the European Union and your data is not shared with third countries.

Do you train your own models?

There are new models released every few weeks by major players like OpenAI, Mistral, Google, Meta, Aleph Alpha and others. We are using these models as they are released and - if necessary - fine-tune them to automate your business processes. However, many of those models are open-source and can be hosted on your own servers, if desired.