DAPHY®
Hybrid AI Technology from CONTEXTSUITE
DAPHY® (Data-driven Prompting with Hybrid AI) is an innovative approach to minimizing AI hallucinations that differs fundamentally from conventional Large Language Models (LLMs). The patent-pending technology combines analytical and generative AI approaches to overcome the inherent weaknesses of classic LLMs.
Unlike traditional AI solutions, DAPHY® combines the analytical precision of specialized AI with the expressive power of generative models to create controllable, comprehensible and powerful answers – to numerical and analytical questions alike.
Hybrid AI: The Most Beautiful Connection of All
DAPHY® at a Glance
DAPHY® works with a hybrid AI architecture. This first uses analytical AI to prepare, cleanse and classify relevant company data – structured and unstructured.
The results are then provided to generative AI models (LLMs) for natural, comprehensible synthesis and explanation.
How Does DAPHY® Work?
01
Capture context
What happens?
Request in natural language is understood semantically.
Result
The data domains required for the response are determined.
02
Select data
What happens?
ContextSuite already provides structured data records.
Result
Only relevant fields and documents are loaded.
03
Analytical processing
What happens?
Numbers are calculated, patterns recognized, entities linked.
Result
Precise quantitative information.
04
Data-driven prompting
What happens?
DAPHY® uses this to create a prompt including a fact snippet (data).
Result
The generative model is given a secure context.
05
Generative response
What happens?
The LLM prepares a comprehensible, reasoned response.
Result
Precise answer combines figures, graphic notes, evaluation & recommendation depending on the task.
06
Source documents
What happens?
Reference to data records or documents.
Result
Complete traceability and auditability.
DAPHY® Prompts for You:
Data-Driven Prompting –
The Revolution in AI Control
Instead of relying on static prompt templates, DAPHY® technology generates prompts in real time based on the relevant company data and adapts the prompts precisely to the respective context.
What Are the Advantages Over Conventional Approaches?
- Elimination of time-consuming manual prompt engineering
- Consistent, reproducible results
- Automatic adaptation to different data types and questions
- Precise contextualization through structured data bases
What Does "Structured Data Basis" Mean?
DAPHY® works with structured data obtained from a company’s raw data and information.
These are intelligently processed and used for the targeted control of generative AI models, which enables precise contextualization and minimization of hallucinations.
Ensures Minimization of
AI Hallucinations Through Data Grounding
What Does "Data Grounding" Mean?
In contrast to conventional Large Language Models (LLMs), which often suffer from hallucinations, DAPHY® uses live data grounding.
Only current company data is used as context for answers, minimizing the risk of hallucinations due to outdated or missing information.
Each answer is also backed up with precise source information from the original data sets, making all statements immediately verifiable. This creates the transparency and traceability that is essential for companies.
How Can This Reduce Hallucinations?
Since DAPHY® works exclusively with structured, up-to-date company data and not on potentially outdated training data, it significantly minimizes the risk of AI-generated misinformation.
The results can be checked and audited by referring back to the sources.
Enables a Unique Combination of Quantitative and Qualitative Analyses
Complex Analyses: Quantitative and Qualitative Data Combined
Question: "Where will the most contracts expire in the next 6 months?" This seemingly simple question requires a complex analysis, which DAPHY® handles in several steps:
01
Understanding of the facts
The AI interprets the context and identifies relevant contract types (employee, customer, rental, supplier or service contracts, etc.)
02
Quantitative data search
a. Existing contracts with expiration dates
b. Terminations made including deadlines
c. Geographical and organizational areas
d. Time parameters (6-month period)
03
Qualitative assessment
Contextualization of the results with background information on:
- Types of contract and their significance
- Business impact
- Recommendations for action
04
Combined answer
Precise figures combined with explanatory context and strategic assessments
Intelligent Distribution of Tasks
GenAI can’t do everything. The DAPHY® technology therefore automatically decides which components are used for which tasks.
It uses analytical models for calculations and trend analyses and generative AI for explanations and summaries. This optimizes both the use of resources and the quality of results.
Resource-Saving and Sustainable AI Architecture
Efficient Model Utilization
DAPHY® uses specialized analytical models for specific tasks and uses generative AI only where it is actually needed.
This leads to a significant reduction in resource consumption compared to approaches based exclusively on resource-intensive generative models.
Energy-Efficient Processing
Through intelligent data filtering and targeted AI model selection, DAPHY® minimizes energy consumption and contributes to the sustainability of AI use.
This is particularly relevant for companies that value environmentally conscious technologies.