Case study · SHAI · 2026

SHAI

An in-house AI agent that reads the whole business and drafts the work.

SHAI is the AI agent we built to run Twin-Ex itself. It is not a bolted-on chatbot but part of the back office — wired directly into the platform, seeing the pipeline, the finances, the inbox, the calendar and the knowledge base, and acting from inside them.

Under the hood: 13 specialist agents — pipeline, finance, inbox and calendar, meetings, knowledge base and research — reaching the business through 63 tools, on Claude Haiku 4.5.

An AI agent that touches a real business is only useful if it is both deep and safe: it has to see everything to help, yet it must never send an email, move money or change a record on its own.

Most "copilots" are a shallow chat box glued onto an app — able to talk, not really able to touch the business.

  1. 01 / 04

    Ask — in plain language

    Open SHAI over any screen in the platform and ask the way you would ask a partner — "what is on my morning?", "what happened to the stalled deal?", "draft a follow-up for yesterday's lead".

    Opens over any screen — with quick actions
  2. 02 / 04

    The agent reads the system

    The question routes to the right specialists among the 13 — pipeline, finance, inbox and calendar, meetings, knowledge base and research — and they pull real data through 63 tools. Read actions run freely, so the answer comes from the system itself, not from a guess.

    The SHAI app full-screen — the daily briefing, the agent roster and the console.
  3. 03 / 04

    Action — drafted, never sent

    Anything that writes — an email, a Slack message, a calendar event, a Google Doc or a deck — is drafted first and waits for one-click human approval. Destructive actions sit behind a double confirmation. We read, correct if needed, and approve — instead of typing from scratch.

    The Twin-Ex platform SHAI lives inside — the dashboard it reads and acts from.
  4. 04 / 04

    Control over every call and every shekel

    Every call is logged, every shekel is metered against a monthly cap, and every automation keeps a manual fallback. That is how you let AI see everything — without letting it send anything on its own.

The embedded assistant over the platform — day and night, side by side.
Fig. 1The embedded assistant over the platform — day and night, side by side.
SHAI as an app — the same agent, full screen.
Fig. 2SHAI as an app — the same agent, full screen.

It writes the daily briefing, drafts the follow-ups and surfaces the stuck lead — and we approve, instead of typing.

Two people who operate like twenty. And the exact same design — an agent that knows the business deeply, with the same safety brakes — is what we build for other businesses too.

13AI specialists
63tools · actions
100%audited · reversible
Next case studyTiger Sport
WhatsApp