How a Retirement Services Provider Put Every Fund-Filing Decision on the Record
Thread AI Editorial · June 11, 2026
A retirement services provider runs employers' retirement plans, which offer investment funds from firms like Vanguard or BlackRock. When one of those firms merges, closes, or renames a fund, the provider has to catch it and tell the employers with affected plans.
A Lemma Worker now reviews the daily flood of announcements, records a decision on every one, and sends nothing until two people sign off.
What Lemma delivers
Before
~50%
With Lemma
100%
Relevant fund changes on record, including the ones ruled out
Before
78%
With Lemma
100%
Entries labeled with the type of fund change
The Challenge
Hundreds of fund changes can be announced in public filings in a single day. Most don't matter to the provider, since it doesn't offer the affected funds, but the operations team's job is to find the few that do. The busiest day on record brought 682 filings, and the team was down to two analysts.
For each filing, an analyst checks whether any employer's plan offers the fund, which can involve other teams and slow turnaround. If a plan is affected, the analyst alerts the relationship manager, who then works with that employer.
In addition to the manual review, when an analyst decided a filing didn't apply, that decision wasn't recorded anywhere. The tracker only logged filings that led to an alert, so a filing dismissed by mistake couldn't be found later, and an auditor had no way to check the call.
The Solution
Lemma Workers now read every filing and record a decision on each one, every day, including the ones that don't apply to the provider's plans. For each relevant filing, an AI model pulls out the key details, including which fund is affected, what's changing, and when it takes effect. A separate Worker sends each alert to relationship managers only after two people approve it, adding a safeguard before anything goes out.
The Workers were tested using the team's own records, along with test filings added to check each safeguard. They flagged the same changes the analysts had, and they caught every test filing.
Every filing now ends with a recorded decision and a reason, including the ones ruled out. The team is now able to show an auditor exactly what was reviewed and why, and a filing dismissed by mistake can be found and corrected.
Every entry is also labeled with a standard type of change, which determines what should happen next with the plans. Before, more than one in five entries had no label, making it harder to tell which changes needed follow-up.
Lemma Capabilities
Logs every filing, then narrows the day’s list.
The Worker reads everything that arrives each day and keeps only the announcements of fund changes. Everything else is set aside with a recorded reason, so a dismissed filing becomes a logged decision instead of a missing one. Even on days when nothing needs action, the review is still logged, so a quiet day can’t be mistaken for a missed one.
Follows the team’s own procedure.
The Worker checks filings in the same order as the team’s written procedure and sorts them into the team’s own categories of change. It automates the steps the team already follows.
AI never guesses.
If a filing’s two fund codes (its ticker and CUSIP) point to different funds, or a key detail like the effective date is missing, the filing goes to an analyst. Changes still awaiting a shareholder vote are held back too.
Keeps the review and send alert Workers separate.
The Worker that reviews filings can’t send anything, and the one that sends waits for two different people to approve. If the same name is entered twice, nothing goes out.
Does this sound familiar?
More documents keep coming in, and bandwidth is stretched.
Your team reviews a high volume of documents but only keeps a record of the ones that lead to action.
If an auditor asks why something was ruled out, you can’t point to a record to explain.
Your tracker doubles as your audit trail, but parts of it are incomplete.