Product update · June 2026

Expert-guided scripts, finish estimates, and answers you approve

Here is what we shipped in Convershake in June. Three improvements your team will feel first — collection scripts that carry the fundamentals, a finish estimate before you launch, and safe answers to off-script questions — plus everything else, listed at the end.

Vitālijs Silins-Ozerson LinkedIn

CTO & co-founder · 30 June 2026

major

Build collection scripts like your best agent

When you build a collection or payment-reminder script, the AI Assistant now guides the fundamentals the way your market expects — so scripts follow proven collection practice even when the person building them isn't a specialist.

From a plain description, the AI Assistant builds a full debt-collection flow — reminder, payment date, and live-agent transfer.

Not everyone who builds a call script is a collections veteran — the person setting up a campaign often isn't the one who knows, line by line, what a compliant, effective collection call sounds like. The build itself should carry that expertise.

Confirm you're speaking to the right person before revealing anything about the account.
Verify identity before sharing account details — a date-of-birth check, the pattern used across the EU.
Read the promise-to-pay back to lock in the amount and date, and meet "I already paid" or a hardship with the right tone.
Guidance follows the borrower's language and your region's rules, and only loads for lending scripts — other flows are untouched.

What shipped

Right-party confirmation before any disclosure
Identity verification before account details (date of birth)
Promise-to-pay read-back — amount and date
"Already paid" and hardship handling
Region-aware guidance: EU, Mexico, US
Undo any change and see exactly what changes before it goes live

major

Know when your campaign will finish before you launch

The campaign builder now shows a grounded finish estimate before you launch, and paces redials across days within your calling rules so more of your list is actually reached.

The campaign builder's Timing step — weekday pills, calling window, and concurrent lines on the left, with a 'What to expect' panel on the right showing dispatch rate, time to finish, contacts reached per day, and total reached, all before launch. — Set your lines and calling days, and the builder projects when the campaign will finish — before you launch.

Before a campaign goes live, you want a reliable read on how long it will take to work through the list. Concurrent lines, redials, and calling windows all shape it — and it helps to see the finish line up front.

Set how many lines a campaign uses and see a finish estimate that accounts for redials and calling days.
See how many lines your other campaigns are already using, so a campaign can't promise capacity it doesn't have.
Choose which weekdays a campaign calls; existing campaigns keep calling every day unless you change them.
Redials pace across days within your calling rules, so more contacts are actually reached.

What shipped

Finish estimate before launch (lines, redials, calling days)
Live line usage across your campaigns
Weekday calling windows
Redials paced across days, not stopped at 24 hours

meaningful

Off-script questions, answered safely

Your AI voice agent now handles those questions from a set of answers you approve — the same way on phone, web, and text — and can hold a sensitive answer until the caller's identity is verified.

The flow editor's Questions and answers panel, enabled, with two approved Q&A pairs. The first — a request for a few more days to pay — is marked available only after identity is verified. — Approve the answers your agent may give off-script, and hold sensitive ones until identity is verified.

Borrowers ask things that sit outside the script — "how much do I owe?", "can I pay next week?" — and each one deserves a clear, approved answer, with the sensitive ones held until you know who's asking.

Add approved question-and-answer pairs to a script; the agent uses them when a caller goes off the main flow.
Answers work the same across phone, web, and text.
Gate a sensitive answer behind conditions — for example, only answer "how much do I owe?" after identity is verified.
Answers that aren't cleared are never shown to the agent, so it can't share something the caller isn't entitled to.

What shipped