"I have been experimenting with a new way to use Form 5500 data, and I am curious whether retirement plan professionals would find it useful -- or consider it a bridge too far. I built a connection between a Form 5500 databaase and AI agents such as Claude Code and Codex. The technical name for the connection is an 'MCP,' but the simplest explanation is that it works like a secure data source. A normal AI chatbot can talk and write using the open internet, but it cannot automatically reach into a specialized database. The MCP gives it permission to ask PlanProvider.Pro (like Judy Diamond) specific questions and return source-linked answers.

"You still ask the question in plain English. For example: • 'Look up the retirement plan for these five companies and summarize the latest filing year, participants, and assets.' • 'Who are the listed advisor, auditor, recordkeeper, and administrator for this plan?' • 'Find retirement plan auditors in Texas and show me their profiles.' • 'Turn this named list of prospective employers into a research brief and identify possible reasons for a conversation.'

"So the real question may not be, 'Would you let AI do your client acquisition research?' It may be, 'Which parts of prospect research would you trust it to prepare before a human reviews the work?' I would especially like to hear from the skeptics: • What research task would you try first? • Where would you require a human checkpoint? • What mistake or privacy concern would make you refuse to use it? Here is the plain-English overview and setup guide: https://planprovider.pro/form-5500-ai-mcp Full disclosure: I built this, and I am looking for candid feedback -- not just compliments."

"Someone mentioned to me that he was on a conference call with an actuary that works for the PBGC. The actuary suggested, as a remedy for overfunding (since the 417(e) rates have gone up to the 5% range and lump sums have gone down) to amend the plan's equivalence to the 2% range and this would help raise the liabilities and reduce the overfunding. Is this reasonable to use an interest rate that is quite low, for the plan equivalence, and is considerably lower than the current 417(e) rates? Can it be justified? This could help solve many overfunded plans, and especially owner only plans (that don't mind raising the liabilities for themselves)."

"About half a generation ago, someone might face a pressure to help a plan’s sponsor date a document falsely. (See prior discussion here.) Now, many service providers use software and internet delivery to present a ready-to-sign document, and expect a plan sponsor’s adoption or approval through DocuSign or another e-sign tool. Is an ink-on-paper signature such a disruption that it’s too hard to pretend a document was signed sooner than when the plan sponsor signed it? Are there still situations in which a temptation to date a document falsely persists?"