aihumin GmbH
aihuman GmbH builds pagent, an AI agent and experimentation platform for A/B testing, personalization and conversion-rate optimization. Its mission is to make every website personal by automating the experimentation loop: finding hypotheses, building on-brand variants, testing them and learning from results.
At a Glance
- E-commerce and D2C brands
- CRO and experimentation teams
- Marketing, product and engineering teams
- CRO and digital marketing agencies
- +6 more
AI Tools by aihumin GmbH
(1)pagent
AI Agent for A/B Testing
Discussions
No discussions yet
Be the first to start a discussion about aihumin GmbH
Latest News
Ship Your Winners, Keep What You Learn: winning tests now include shipping guidance and holdout support; knowledge docs, self-written reports and named statistics levels were added.
From Winning Test to Pull Request: pagent added GitHub repository connection and pull-request creation from finished tests.
pagent joined Claude's connector directory, allowing users to connect and ask about tests without manually managing tokens.
Connect Your AI Tools in One Click: Claude Code, Claude.ai, Codex, Cursor, Windsurf and Zed received browser-based connection support.
Products & Services
A complete platform that finds hypotheses, builds and checks real-code website variations, runs statistically rigorous A/B tests, supports personalization and analytics, and lets teams review and approve changes.
An autonomous agent for CRO and experimentation teams that identifies visitor drop-offs and test opportunities, creates on-brand variants in a site's design, runs tests and learns what to try next.
Creates a landing-page version for each advertising angle from an existing page and tests it against the original across ad platforms.
An MCP server that lets Claude, ChatGPT, Codex, Cursor and other MCP clients read test results, capture hypotheses, create and start tests, and manage audiences and goals through OAuth or a personal token.
Market Position
pagent positions itself as an autonomous experimentation and personalization layer that does the work usually left to a CRO team or developer: it proposes hypotheses, writes real-code variants, checks them, runs rigorous tests and learns from the outcomes. It differentiates from Optimizely, VWO, Kameleoon and AB Tasty by automating test creation and execution rather than only supplying testing infrastructure; it overlaps with AI optimization platforms such as Coframe and Evolv AI, while emphasizing approval controls, design-system/on-brand code and compatibility with an existing testing stack.
Leadership
Founders
Constantin Dißelkamp
Founder and CEO; machine-learning engineer and physicist. Before pagent he was a Machine Learning Engineer and Team Lead at Trality in Vienna (2020–2023), working on machine-learning systems for algorithmic trading; he was also a guest scientist at Forschungszentrum Jülich and studied physics at RWTH Aachen University.
Christopher Helf
Founder and CTO. His profile identifies previous roles at Trality, Robimo – Robots in Motion, Spil.ly and the University of Vienna; he studied computer science at the University of Vienna.
Executive Team
Constantin Dißelkamp
Founder & CEO
Machine-learning engineer and physicist; former Machine Learning Engineer and Team Lead at Trality, guest scientist at Forschungszentrum Jülich, and RWTH Aachen physics graduate.
Christopher Helf
Founder & CTO
Computer-science-trained technology leader with previous roles at Trality, Robimo – Robots in Motion, Spil.ly and the University of Vienna.
Board of Directors
Founding Story
The founders started pagent because most websites showed every visitor the same experience and conventional testing and personalization required more people, time and budget than many teams had. Their initial vision was to use generative AI, marketing expertise and Bayesian statistics to automate website experimentation and ultimately build a leading AI model for personalized communication.
Business Model
Revenue Model
B2B platform licensing priced by a customer's monthly unique website users rather than seats or separate modules. The license includes experimentation, personalization, analytics, brand guidelines, approval workflows and a monthly allowance of agentic work; customers can buy optional top-up packs and scope paid pilots.
Pricing Tiers
One license includes all platform features; there is no public list price, no seat fee and no separate module pricing.
Optional capacity packs are available beyond the monthly task allowance included in a license.
Target Markets
- E-commerce and D2C brands
- CRO and experimentation teams
- Marketing, product and engineering teams
- CRO and digital marketing agencies
- Telecommunications
- Fashion e-commerce
- Conversion-rate optimization for e-commerce and D2C websites
- Increasing sales, sign-ups, newsletter registrations and video clicks
- Personalized landing pages for advertising campaigns
- Continuous experimentation on long-tail pages and audience segments when developer capacity is limited
- Marketing, product and engineering teams that need on-brand website variants
- CRO agencies running experiments across multiple client websites
- Pixum – reported +16% conversion uplift within three weeks
- NEOH – reported +25.74% revenue uplift and +16.81% performance increase in seven days
- Brandnamic – click-through rate increased from 6.35% to 11.65%
- Zahnklinik Wien Döbling – reported +15% conversion-rate uplift