All updates

Call transfers are now guaranteed

When Alli transfers a call or wraps up a conversation, those actions now execute every single time — even when multiple things happen at once. Previously, a transfer and a closing message happening together could cause the transfer to silently get dropped.

Duplicate summary emails eliminated

There was a scenario where the same call could trigger two summary emails to the business. That is now impossible — the system checks before sending and discards any duplicate.

Summary emails survive server restarts

If a summary email was being processed at the exact moment of a server restart or deploy, it would be silently lost. The system now recovers and resends any missed emails automatically on startup.

Post-call webhooks now retry on failure

If a business's configured webhook URL fails or times out after a call, the system now automatically retries up to 3 times before giving up. No more missed notifications from a temporary connection hiccup.

Long calls no longer drop mid-conversation

Calls over 8 minutes used to hit a hard connection limit, causing a 1–3 second gap in the conversation. Alli now anticipates this, briefly lets the caller know, and picks up exactly where she left off — no abrupt interruptions.

Smoother reconnection after dropped calls

When a call drops and Alli reconnects, the handoff is now cleaner. There are no ghost actions firing after the session has already ended.

Increased call capacity

Alli can now handle significantly more simultaneous calls by automatically routing to whichever region has available capacity. Effective capacity has doubled — high-volume periods won't cause delays or missed calls.

Alli's behavior is fully consistent across all calls

The core rules that govern how Alli behaves on every call — how she handles being asked if she's an AI, how she responds to silence, what she never says — are now centralized in one place. Previously these could drift out of sync across different parts of the system.

Caller manipulation attempts are now blocked

A security gap has been closed where a caller could craft a specific phrase that might influence how Alli behaves after a reconnection. Caller messages are now sanitized before being replayed into a new session.

Greeting latency tracked on every call

The exact time between a call connecting and Alli's first word is now logged for every call. This gives us a clear, ongoing picture of how responsive Alli feels to callers and flags any slowdowns immediately.

Alli now automatically researches your leads

When a lead has a business email, Alli now automatically looks up their company, pulls key info from their website, and uses industry-specific talking points in follow-up conversations. Better context means more relevant outreach.

154 stuck leads recovered and back in nurture

A group of leads who came in through event keyword flows were stuck with no status and no follow-up — they had fallen completely through the cracks. Root cause identified and fixed, all affected leads are now active in their nurture sequences.

Daily leads report rebuilt — now always accurate

The previous daily report was based on snapshots that would reset every time the server restarted, causing all 46,000+ leads to show as "new" after a deploy. It now pulls live from the database every time — accurate regardless of server state.

API rate limit handling — no more dropped campaigns

When the AI Leads API hits its rate limit, the system now automatically retries up to 3 times before giving up. Previously a rate limit hit during onboarding would silently fail and leave the client without a campaign.

Daily AI Lead Detection summary email

Every lead flowing through the system is now logged and delivered as a daily summary email grouped by campaign — so you always have a clear picture of what came in and where it went.