deepsense.ai

Advanced

Applied AI experts delivering tailored AI solutions

Overview Locations Service lines Industries Technologies Case studies Rates Certifications Leadership Insights Events Social networks

Overview

120

employees

2014

founded

Warsaw

Poland

We are applied AI experts delivering tailored AI solutions. We stay at the forefront of applied AI by partnering with AI leaders such as OpenAI, NVIDIA, Anyscale, and LangChain and applying our own in

We are applied AI experts delivering tailored AI solutions, offering both guidance and implementation to help our clients unlock the full potential of AI. With 10+ years of AI experience, we have completed 200+ commercial projects with clients spanning both global brands and innovative scale-ups such as Johnson & Johnson, Sky, Zebra, Danone, Hexagon, Docplanner, Google, Volkswagen, Nvidia, L’Oreal, Nielsen, Whirlpool, Intel, Brainly, WWF, European Commission, United Nations, Santander, BNP Paribas. We specialize in applying LLMs, MLOps, computer vision, edge solutions, and predictive analytics to enhance our clients’ products and operations. Being official partners with AI leaders such as OpenAI, NVIDIA, Anyscale, LangChain, and Vespa, and collaborating closely with their teams ensures that we stay at the forefront of AI innovation and can effectively apply their technologies to solutions we build for our clients. We leverage our in-house developed solutions like db-ally (a database i

Contact details

Business info

Publicly traded: No
Majority shareholders: Management & Employees

Locations

Warsaw Office, Poland

Headquarters

Founded2014

Employees120

Al. Jerozolimskie 44, 00-024 Warszawa, Poland

Service lines

Artificial Intelligence (AI)100%

Industries

Technology30%

Healthcare20%

Technologies

Python

Case studies

View all1 / 2

5x Boost in In-Silico Drug Discovery with a Multimodal LLM

Working with the client’s in-silico drug discovery team, we developed a pipeline to train multi-modal LLMs with a deep understanding of chemistry. We integrated Llama 3.1 and Graphormer, incorporating molecular structures as graphs into the LLM, creating a unique model that could analyze textual, SMILES, and molecular data.

GenAI-Powered Frontline Worker Assistant

Collaborating closely with the client’s development teams, we created a LLM-powered assistant capable of real-time information retrieval from both structured and unstructured data sources. To ensure scalability and safety in deployment, we implemented a robust input and output safeguarding mechanism to protect against errors.

130% Improvement with AI Appointment Scheduling

The client aimed to automate doctor-patient appointment scheduling, increasing self-scheduled bookings to let doctors focus on key tasks and allow patients to book anytime. We created a natural conversation flow that allowed the system to gather and provide all necessary information without relying on rigid scripts.

Indicative rates

Rates not shared by provider.

Certifications⁰

No certifications data available.

Leadership

Michał Iwanowski

VP of AI Advisory | Board Memb

Insights

View all1 / 6

AI agents for reliable workflows, data access, and automation

AI agents are moving from “cool demos” to practical systems that connect LLMs with tools, data, and rules in a controlled way. The article outlines a common agent setup (LLM controller + planner, memory, and tool-calling) and why guardrails matter when models hallucinate or struggle with precise reasoning. It then grounds this in three operational patterns: a Text2SQL database agent that constrains actions to reduce unsafe queries/SQL injection risk; an HR automation agent that converts existing UI/instruction flows into validated click/API scripts; and a conversational agent that books, reschedules, and cancels appointments via scheduling APIs (text + voice). Tech leaders should view agentic work as integration-heavy software engineering, not “just prompting.”

Hardening GenAI: From Prototype RAG to Production Systems

As GenAI pilots move into customer-facing and internal production use, security, extensibility, and operational control become primary concerns—not model quality alone. Deepsense.ai’s RAGbits 1.4 release focuses on production enablers: OAuth2-based access control for APIs, a more extensible schema layer to adapt to evolving business needs, and structured file handling to safely ingest and manage documents in RAG workflows. The emphasis shifts from experimentation to governed integration. For technology leaders, this signals the next maturity step: treating GenAI as an enterprise system with identity, compliance, and lifecycle controls.

Custom MCP Servers in Enterprise AI Architecture

As enterprises operationalize generative AI, managing how models access data, tools, and internal systems becomes an architectural concern rather than an experimentation detail. The content outlines how custom MCP (Model Context Protocol) servers can act as controlled integration layers between LLMs and enterprise environments, structuring context delivery, enforcing governance, and standardizing tool access. It explains how this approach supports secure orchestration, observability, and reuse across AI use cases. For technology leaders, MCP-based design offers a way to scale AI adoption without bypassing security, compliance, and platform standards.

Events

No events data available.