Blog: Can an AI assistant become part of a research team instead of living in a browser tab?
| Fig.1. From OpenClaw to Your Favorite Messaging App |
At the Interdisciplinary Institute for Societal Computing (I2SC), we recently explored OpenClaw—an open-source, self-hosted AI assistant. It connects to platforms such as Telegram, WhatsApp, Microsoft Teams, Slack, and Discord. Unlike most AI chatbots, it also lets you choose the AI model that best fits your task.
One of OpenClaw’s biggest advantages is control. Conversations, memory, API keys, and configuration stay on infrastructure that you manage. The assistant can also run continuously on a server, even when your laptop is turned off.
One assistant, multiple channels
OpenClaw uses a central gateway between messaging platforms and AI models. The gateway keeps conversation history, runs scheduled tasks, uses browser tools, and executes reusable skills.
It also supports multiple AI providers. You can use Claude for complex reasoning, GPT for general tasks, Gemini for Google integrations, or switch models depending on cost and performance.
| Fig. 2: Our Openclaw agent sent a daily update to a teams channel by suggesting talking points to meetings on a calendar or general news related to my research. |
For our pilot, we connected OpenClaw to Telegram and Microsoft Teams. Through Telegram, it answered questions, remembered previous conversations, and triggered scheduled jobs. Through Teams, it posted automated updates based on calendars and selected blogs and sources. The MS team's connection was set up using the workflow API module.
Although we encountered some Microsoft Teams permission restrictions, the overall experience showed how AI assistants can work directly inside communication tools that research groups already use.
What can it do?
OpenClaw is more than a chatbot. It can automate recurring tasks that many researchers perform every day.
Examples include:
- sending daily meeting briefings
- monitoring webpages and notifying users about changes
- summarizing papers or online resources
- organizing messages and documents
- translating information between collaborators
- scheduling recurring research updates
These workflows have the potential to reduce repetitive administrative work and make research coordination easier.
What we learned
Getting OpenClaw running was straightforward. Making it reliable required more work.
We encountered challenges with browser automation, authentication, permissions, and external APIs. Autonomous agents also need clear stopping conditions. Without them, they may repeatedly retry failed tasks and increase token costs.
Prompt design also mattered. Simple, well-defined tasks worked much better than broad instructions.
For deployment, we used Hostinger, which offers a one-click OpenClaw template. Their plans also include credits for Nexos AI (a unified LLM API) and Oxylabs AI (web scraping), making it easy to experiment with different models and tools. During setup, we relied on both ChatGPT and Hostinger’s AI support assistant, Kodee, to troubleshoot configuration issues.
Looking ahead
This was an exploratory deployment, but it highlighted interesting opportunities for research and teaching. We see potential applications in research in the classroom.
If you’re interested in experimenting with OpenClaw or have ideas for collaborative projects, we’d be happy to hear from you.
Learn more
- OpenClaw on Hostinger: https://www.hostinger.com/applications/openclaw
- OpenClaw community use cases: https://www.reddit.com/r/OpenClawUseCases/comments/1rxey2s/use\_case\_multiagent\_voice\_assistant\_on\_a/
- Awesome OpenClaw Agents: https://github.com/mergisi/awesome-openclaw-agents