§ 01Agentic System · rev 1.0

Building trustworthy, auditable and sovereign agents.

AgentFormers

We are a Business Finland research-to-business project at the University of Helsinki, exploring how agentic solutions could solve the problem of complex continuos workflows.

We are seeking potential seed-stage investors, new strategic partners, and early adopters for our system.

§ 02The problem with agents today

Agent development is slow, expensive and risky.

In today's data-driven world, organisations face significant inefficiencies with developing their own agents. Organisations typically offload these tasks to IT departments who may not even have the resources or expertise. AgentFormers addresses these challenges by providing a seamless, scalable solution that enables users to securely create agents to work with enterprise data and automate routine tasks, freeing up organisations to focus on more valuable work and improving the service experience for all users while saving time.

FIG. 02A — IT PROJECT vs. PLATFORM AGENTDETAIL
TRADITIONAL IT PROJECTAGENTFORMERS PLATFORMCAPABILITY 01SPECBUILDTESTSHIP~5 MO. · €€€CAPABILITY 02SPECBUILDTESTSHIP~6 MO. · €€€CAPABILITY 03SPECBUILDTESTSHIP~4 MO. · €€€CUMULATIVE15+ MO. · NO REUSE · €€€€€CAPABILITY 01 · CONFIGURE AGENT~2 WK.CAPABILITY 02 · CONFIGURE AGENT~1 WK.CAPABILITY 03 · CONFIGURE AGENT~2 WK.AGENTFORMERS RUNTIMEGENERIC AGENTS · SHARED DATA LAYERGOVERNANCE · SANDBOXINGOBSERVABILITY · AUDIT TRAILCUMULATIVEWEEKS · REUSABLE · €
§ 3 · PROJECT MOTIVATION
"During our research, we saw a clear, recurring need in every organization: the immense time wasted on manual information retrieval. We realized our technology wasn't just an academic exercise; it was a practical solution to give people back their most valuable resource: time. "
Sasu Tarkoma
Sasu Tarkoma
Professor of Computer Science · University of Helsinki
§ 04Layer 01 · Control

Controlling the chaos.

Without proper structure and grounding, AI assistants can go off-track, hallucinate facts, and produce unreliable outputs. AgentFormers addresses this by providing users full control to define agents and their behaviour, connect agents to retrieve data from enterprise systems in a secure manner, and govern agents to operate in safe sandboxed environments.

FIG. 04A — EXECUTION PIPELINEDETAIL
POLICY RAILPROMPTscope · intentGROUNDfacts · contextPLANcost · stepsACTtools · permsVALIDATEformat · safetyRESPONDSTRUCTURED · GROUNDED · SELF-CORRECTING
§ 05The AgentFormers value prop

Our offer, to you.

01

Autonomous Engine

The system intelligently decides which tool to use in each situation and knows when the task has been successfully completed.

02

Programmable and Modular Agents

Build, configure, and manage your own custom agents and tools with a modular, granular, role-based access control.

03

Transparent Architecture

An open and modular architecture ensures high-quality answers without vendor lock-in, giving you full control.

04

Trust and Safety by Design

Built-in guardrails, monitoring, and governance features ensure your agents operate safely and reliably.

05

Ethical and Compliant AI

Built-in compliance with the EU AI Act (2024/1689), ISO/IEC 42001, and GDPR ensures your operations are responsible and future-proof.

06

Secure Data Integration

Securely connect your documents, databases, and applications using vetted and reliable integration methods.

§ 06For Investors

Market opportunity.

There is a growing demand for enterprise AI, and a platform built to meet it.

AgentFormers is intended to be a B2B SaaS platform targeting large, knowledge-intensive organisations. Our solution enables these organisations to easily build and deploy their own secure and trustworthy AI agents, significantly improving internal efficiency.

SaaS for knowledge-intensive sectors

This technology is scalable and has potential for widespread adoption. Our agents are built to solve complex workflows and integrate with enterprise data, making them ideal for knowledge-intensive sectors such as healthcare, legal, finance, and academia.

Impact and benefits

AgentFormers’ broader impact is the democratisation of AI for organisations lacking dedicated AI teams. Our solution enables companies to develop robust custom solutions, improve service quality and boost productivity by automating knowledge work.

Future and path to commercialisation

The immediate next step involves completing production pilots with partners. Validation with real users and data enables a successful transition to market.

§ 7The team

Researchers who want better agentic systems.

We are a team of experienced researchers from the University of Helsinki with a wide range of expertise in agentic systems, software development, the Internet of Things, publish/subcribe, and data science.

MEMB 001• ACTIVE
ST
Sasu Tarkoma
PROJECT LEAD
Professor of Computer Science and Dean of the Faculty of Science at the University of Helsinki.
UNIVERSITY OF HELSINKI
MEMB 002• ACTIVE
JW
Jussi Wright
COMMERCIAL CHAMPION
Commercial Champion for the AgentFormers project.
UNIVERSITY OF HELSINKI
MEMB 003• ACTIVE
EK
Emil Kelhälä
RESEARCH ENGINEER
Researching and developing agentic systems. Masters Student at the University of Oulu.
UNIVERSITY OF HELSINKIUNIVERSITY OF OULU
MEMB 004• ACTIVE
AE
Alexander Engelhardt
RESEARCH ENGINEER
Researching and developing agentic systems. Doctoral Researcher at the University of Helsinki.
UNIVERSITY OF HELSINKI
MEMB 005• ACTIVE
BU
Burhan Uddin
RESEARCH ENGINEER
Researching and developing agentic systems.
UNIVERSITY OF HELSINKIUNIVERSITY OF OULU
MEMB 006• ACTIVE
JC
Jacky Cao
RESEARCH ENGINEER
Researching and developing agentic systems. Postdoctoral Researcher at the University of Helsinki.
UNIVERSITY OF HELSINKIUNIVERSITY OF OULU
MEMB 007• ACTIVE
AK
Abhishek Kumar
RESEARCH ENGINEER
Researching and developing agentic systems. Senior Lecturer at the University of Jyväskylä.
UNIVERSITY OF JYVÄSKYLÄUNIVERSITY OF OULUUNIVERSITY OF HELSINKI
§ 8Questions

Everything else, answered.

On our cloud (dedicated VPC) or on yours. We are designing the system stack to be portable to on-premises and cloud/edge environments.
§ 9 · NEXT STEP

Pilot your ideas with us.
Test trustworthy, auditable and sovereign agents.

We'll discuss how we can bring our agentic systems to your enterprise, or how you can invest in our technology.