Ship Log

What we've
been shipping

A dated record of what goes live at Big Sister, newest first.

2026 so far: 47 updates shipped and 280+ releases to production.

October 2026

Product

Connectors catalogue

Browse connectors by category and search, plus a setup card that shows what's left before your first score.

Trust

Every quote checked

Every quote in a score's reasoning is checked against the transcript, and any that don't match are flagged.

How we build

Mission Control HQ

The engine behind this page. Our AI agents and our team work from one control room, on the same map.

Method

Rubric v2.8

Fairer on short cold-call openings and inbound calls. Calls that switch languages mid-call are transcribed in both.

How we build

Every change reviewed twice

Every change to the product gets an AI code review, then an engineer's approval, before it ships. It is how a small team ships most days.

Integrations

Sybill and transcript files

Sybill calls and uploaded transcript files can now be scored.

Product

Team plan from one rep

No seat minimum on Team: $50 per scored user.

Read more →
Integrations

Big Sister in Claude

Ask Claude about your team's scores and calls. Read-only, signed in with your Big Sister account.

Read more →
Integrations

Zoho CRM v2

Each rep's call skills sit next to what happened to the deal in Zoho.

Crew

Inbound agent on our site

Our inbound agent answers visitors on big-sister.ai by text or voice and hands off to a person.

Crew

Rematch

Replay a real sales call with an AI coach, by voice.

Product

Monthly report

Managers get a monthly report in the Inbox, with a full page that prints to PDF.

Trust

Not a sales call

A rep can flag a call that isn't sales. A referee decides, and excluded calls drop out of every metric.

September 2026

Method

Rubric v2.7

Calibrated against blind human labels. Error against those labels dropped by about a fifth.

Product

Conversion next to skills

Win and conversion numbers from your CRM sit next to each rep's Sales Score.

Trust

DPA, Terms and subprocessors

Click-through DPA, new Terms and Privacy, and a public subprocessor list. No NDA needed to start.

Read more →
Trust

AI judge, blind by default

An AI judge gives a second opinion on disputed scores. Referees see it only after their own call, because we measured that seeing it first anchors them.

Product

Inbox and weekly report

An Inbox in the app with a weekly report for each rep and the team.

Method

The Manifesto

What we believe about sales skill, in five beliefs.

Read more →
Product

CRM hygiene

Each rep's CRM hygiene sits next to their call skills.

Crew

Prep agent

Before a sales call, our prep agent writes a brief from the CRM deal and past conversations, with the rep's Sales Score alongside. We run it on our own pipeline first.

Integrations

Bring your own key

Paste a notetaker's API key and we start pulling calls. Granola first, others next.

August 2026

Product

Self-serve signup

Create an account, connect a recorder, pick your reps and pay, without talking to us.

Product

Ukrainian app and playbook

The app and the playbook in Ukrainian, with a language switch in the user menu.

Product

Hear the moment

Every score links to the part of the call that earned it. Press play.

Method

Rationale in the call's language

Each score's reasoning comes in the language of the call, with the evidence quoted word for word.

Trust

Scoring health alerts

If a team's calls stop being scored, we get an alert the same day.

Integrations

Fireflies for the whole team

A manager connects Fireflies once and every rep's calls are covered.

Integrations

Connect Ringostat yourself

Paste your Ringostat key, we verify it and start scoring.

July 2026

Product

Appeal a score

A rep who disagrees with a score can appeal. A trained human referee decides, with the call and transcript in front of them.

Product

Plans and billing in the app

Public pricing, monthly or annual, seats and invoices in Settings.

Trust

Weekly random spot-check

Each week a referee re-checks a random 10% of scores, not only the appealed ones.

Trust

Security review

Private API docs and tighter cloud permissions across production.

Product

Team management

Managers invite people, set roles and deactivate users in Settings.

Method

A gold set from three humans

Scores calibrate against calls labelled by three human raters. Five skills got sharper rules.

Trust

Reps see their own calls

Sellers see only their own calls and scores. Managers see the team.

Trust

Our own scoring pipeline

Scoring runs on our own cloud pipeline. Every score event is written to an append-only log.

Product

The new app

The redesigned app on real data: Home, Team, Interactions and Playbook, with search across every call.

Product

Shareable Sales Score card

Every rep can export their Sales Score card as an image.

Method

Better Ukrainian transcripts

Ukrainian calls are transcribed with speaker roles, and names stay exactly as spoken.

Integrations

Phone calls

We score phone calls, not only video meetings. Recordings are pulled, transcribed and scored.

Method

All 18 skills on every call

The scorer rates every conversation on all 18 skills of the method.

June 2026

Integrations

Zoho CRM

Connect your CRM in one click. Full deal history loads first, then it updates daily.

May 2026

Integrations

Self-serve connectors

Settings has a Connectors page. Link Fireflies yourself, no support ticket.

April 2026

Integrations

One framework for connectors

We rebuilt how integrations plug in, so every new notetaker and CRM connects through one framework.

January 2026

Integrations

Fathom and more notetakers

Calls come in from several notetakers, Fathom included. Teams without a CRM can start too.

Product

Start with your past calls

When a team joins, we score their past calls too, so the first report already has history.

Where this is going

One method for
people and agents.

Read why we build this way, then try it on your own calls.

V2 is live. Start with Solo →