“Can I have two health insurance plans?” How a benefits chatbot answers the questions HR shouldn’t guess at.

What do employees ask about their benefits?

Employees ask the same benefits questions every year. The top ones I hear: Can I change my plan this year? Can I add dependents? And the one that sounds absurd: Can I have two health insurance plans?

The answer to that one is yes. People can be covered by two plans. Coordination of benefits rules decide which plan pays first, and the second plan generally picks up some of what the first didn’t. The details depend on the plans, so the right answer always ends with “check your plan documents.”

Why do wrong answers cause so much trouble?

Wrong benefits answers cause trouble because they surface at open enrollment, when the agency is already scrambling for stragglers. HR giving the wrong kind of answers causes multiple hitches in the giddyup. If people get the wrong information, we have to double back and correct their assumptions, and that risks calling the carrier for an extension.

HR isn’t careless. They’re answering questions about plans they didn’t design, from memory, between everything else on their desk.

What do most agencies try first?

Most agencies try a benefits guide and an FAQ sheet, handed out before enrollment. A good guide answers most questions ahead of time. My proposal generator builds one straight from a carrier quote. The trouble is nobody opens the guide at the moment they have a question. They ask HR, or the coworker who enrolled last year.

Some agencies try a generic HR chatbot or insurance chatbot. Those know general rules. They don’t know this group’s plan, this year’s deductible, or what this carrier requires to add a spouse.

Wanna fix it?

Wrong benefits answers get fixed with a benefits chatbot that answers only from the agency’s email history and the group’s plan designs, shows where each answer came from, and hands anything uncertain to an account manager who approves the reply. Finn proved the first part works. Here’s how I’d build the next one into a client portal.

1. Feed it your real answers. Past emails where account managers answered these questions, plus the plan designs for the health plan and the ancillary benefits. Finn started with about 300 emails’ worth.
2. Keep it to one group’s plans. An employee gets answers about their own plan, at this year’s numbers.
3. Make it show its source. Every answer points to the plan document or email it came from.
4. Hand off anything uncertain. Eligibility edge cases, life events, specific claims, and every health question go to an account manager.
5. Approve the handoffs. The chatbot drafts, and the account manager reviews and sends. My CRM already handles my own email this way: it drafts the reply, and nothing goes out until I approve it.

What would it look like once it’s running?

Once it’s running, “Can I add my kid?” gets answered at 9pm without waiting on HR. The account manager’s queue holds the questions that need a person, each with a drafted answer to review. HR stops being the help desk for a plan somebody else designed.

On the next one, I’d count from day one: questions answered, questions handed off, and wrong answers caught before enrollment closed.

Can you do this yourself?

Yes, partly. Take the last year of account manager email, pull out every question that came up more than twice, and write one clear answer for each. Give that sheet to HR before enrollment starts. That alone cuts the guessing.

It stops working when the sheet gets too long for HR to find the answer fast, or when every group needs its own version. One group, a good FAQ sheet. A book of groups on different plans, a chatbot that knows which plan it’s talking about.

Quick answers

What is a benefits chatbot?

An assistant that answers employees’ plan questions from the group’s plan documents and the agency’s past answers.

Is an HR chatbot the same thing?

A general HR chatbot covers company policy. A benefits chatbot has to know one group’s specific plans.

Can you have two health insurance plans?

Yes. Coordination of benefits decides which plan pays first. Check your plan documents for the details.

Should a benefits chatbot answer health questions?

No. Anything medical, or anything it isn’t sure about, goes to a person.

When do wrong answers hurt most?

At open enrollment, when a correction can mean asking the carrier for an extension.

Has DoBetter built this?

Yes. Chris Cordon built Finn, a benefits chatbot that answered from about 300 emails’ worth of real questions and answers plus all of the agency’s plans. It still runs on his last agency’s website.