Albert: The AI Assistant Inside MyRapidi
Last week we introduced AI in MyRapidi at a high level. This week we go one layer deeper into Albert, the AI assistant that lives inside the platform. Albert is not a bolt-on chatbot. He has direct access to your service, your documentation, and your transfer history, and he is built to help you actually get work done, from troubleshooting an error to creating a new connection or pulling a list of transfers in a specific status.
Below is a full walkthrough of what Albert can do today, how to use him in practice, and what is coming next.
A Quick Summary
- Albert is the AI assistant embedded in MyRapidi. Anyone who logs in has access.
- He is specialized in MyRapidi: connections, transfers, schedules, logs, and the Rapidi wiki.
- You talk to him in any language. He replies in whatever language you write in.
- He has persistent memory, so you do not have to re-explain your context every time.
- Always verify the actions Albert performs. AI is a great accelerator, but human review is still the rule.
- The next session covers the Rapidi MCP Server, which lets external tools like Claude or ChatGPT reach the same context Albert has.
Watch the Full Session
What Albert Actually Does
Albert is the AI assistant layer inside MyRapidi. He is not the transfer engine, and he does not decide what data to sync. Those are still deterministic and configured by you.
What Albert is good at is acting as a knowledgeable teammate inside the platform. When you ask him in the chat what he can do, he answers with a list that maps to the things integration teams do every day:
- Manage and troubleshoot connections to external systems
- Create and configure transfers
- Handle schedules
- Investigate errors and explain what they mean
- Provide documentation and instructions for MyRapidi features
- Manage AI memories so he keeps your context across conversations
- Send error reports and notifications
And many more day-to-day tasks tied to your service. The point is that Albert is not a generic AI. He is grounded in your MyRapidi setup and the Rapidi documentation.
How To Use Albert
You do not need to set anything up. Open the chat in MyRapidi, start typing, and have a conversation. A few practical patterns we showed in the session:
- Paste an error message. Albert reads it, explains what it means, and suggests fixes.
- Ask him to do something. "Can you test this connection?" Albert performs the action and confirms whether it succeeded. The action is also recorded in the logs, so you can verify it.
- Ask for a list. "List all my schedules." "List transfers with status testing." Albert returns the matching items with codes and descriptions so you have a quick overview.
- Continue the conversation. Every chat is saved with a generated title that summarizes what was discussed. You can revisit it any time, and so can your team.
Albert responds in real time. If he needs more information to act, he will ask for it. If he cannot do something, he will say so clearly.
A Real Example: Decoding A Duplicate Account Error
One of the live demos in the session was a typical CRM-side duplicate error. The pattern is something every integration team has hit:
- Paste the raw error message into the chat.
- Albert reads it and explains, "This is a duplicate error. You cannot create a duplicate account; you have to use the existing one.
- Ask the natural follow-up: "Please suggest ways to handle duplicates."
- Albert returns several options: use external IDs, implement duplicate detection logic before sync, or run a data cleanup.
None of those suggestions is automatically the right one for your business. They are not meant to be. The value is that Albert gives you a clear direction within seconds, instead of you spending 20 minutes formulating the same options yourself. From there, you decide which path fits your setup and apply it.
Always Verify
When you let any AI create a connection, configure a transfer, or change a schedule, go back and check the result yourself. Open the connection. Open the transfer. Look at the configuration that was actually written. Albert is a great accelerator, but a small misunderstanding of your business rules can produce a result that looks correct and is not. If something is missing or wrong, you can ask Albert to update it. The verification habit is what keeps AI safe in production.
What's Coming Next
Two pieces follow Albert in the AI roadmap. Both are about extending what you can automate.
1. Custom Agents
You will be able to build your own agents that run alongside Albert. Each agent can have a specific job: send a daily email of new data errors, generate a report on transfer activity, or perform a specific change on your integration on a schedule. Agents are equipped with skills and memory, the same way Albert is, so they stay focused on the task they are designed for.
2. Agent Cloning
Once you have an agent that works, you will be able to clone it as the basis for a variant. A troubleshooting-focused agent, a reporting-focused agent, a notifications-focused agent. This makes it practical to have a small team of specialized agents instead of one general-purpose one.
3. Rapidi MCP Server (Session 15, May 14)
The MCP Server is what allows external AI tools, like Claude or ChatGPT, to access the same context Albert has in MyRapidi. You will be able to ask your preferred AI tool to create a transfer, list connections, or change a configuration in MyRapidi, and the request runs against the real platform, not a mock. We cover this in detail in Session 15.
Getting Started Today
Three concrete things you can do this week without waiting for anything new:
- Open Albert and paste a real error from a recent failed run. Read his explanation. Compare it to what you would have written yourself.
- Ask Albert to list something useful. Schedules in a particular state, transfers in testing, and connections you have not touched in a while. The conversational query is often faster than navigating the UI.
- Tell us what works and what does not. If Albert gave you a partial answer or missed the right fix, send feedback. That is how we improve the responses.
Frequently Asked Questions
Does Albert make changes to my configuration without asking?
Albert can perform actions like running a connection test or creating a configuration when you ask him to. The platform records every action in the logs, and you should always verify the result. Treat any AI-driven change like any other change in production: review it before you trust it.
Can I talk to Albert in my own language?
Yes. Albert handles many languages, including English, German, and Spanish. He will answer in whichever language you start the conversation in. This is useful for teams that prefer to work in their native language, especially when reading or explaining error messages.
Does Albert remember previous conversations?
Albert has a persistent memory for context, like your name, time zone, and preferences, so he gets more precise the more you use him. Every chat is also saved with a summary title, so you and your teammates can revisit any past conversation.
Which connected systems does Albert understand?
Albert understands errors and configurations from the systems Rapidi connects to, including Microsoft Dynamics 365 Business Central, Dynamics 365 Finance and Operations, Salesforce, and HubSpot. Accuracy is highest for the systems and error patterns Albert has seen most.
Can I use Albert from outside MyRapidi?
Albert lives inside the MyRapidi interface today. The upcoming Rapidi MCP Server (Session 15, May 14) is the bridge that lets external LLM tools query MyRapidi for context, so you can effectively reach Albert-style help from your preferred AI tool.
What if Albert's suggestion is wrong?
It happens. Treat suggestions as a fast first pass, not a final answer. If a suggestion does not fit your context, ignore it. If you spot a recurring miss, share it with us through support or the next Open Office Hours session so we can improve Albert.
Register for session 15: MCP Server
