Sitruna AI Skills Platform

Platform Overview

Sitruna Consulting manages 111 Amazon seller accounts across a team of brand managers and client success managers. Each client call required manually checking five separate tools — CRM records, task history, marketplace performance, team email, and calendar notes.

The AI Skills Platform replaces that manual assembly with purpose-built AI skills that synthesise all five data sources into structured briefings, targeting under two minutes per call.

Key outcomes

Faster prep

Seamless transitions

Proactive coverage

Better conversations

Skills

Delivered v1.6

CSM Pre-Call Briefing

Full client context assembled in 2 minutes, ready before every call.

Before every call

Delivered

Next Build

Account Handover

Complete client history packaged for incoming Business Managers.

Each staff transition

Full build

Planned

Holiday Handover

30-day coverage brief so cover teams stay on top of every account.

Monthly +

Light derivative

Planned

Pre-BM Call Briefing

All clients per BM surfaced in a single view for weekly reviews.

Each review cycle

Full build

Shared infrastructure

Integration Layer

111 Clients mapped

5 Data sources

Error handling Client–team mapping Structured prompts

Data sources

HubSpot CRM

Client records

ClickUp Tasks

Task history

Merchant Spring

Platform data

Drive / Calendar

Docs & schedule

Front Email

Comms history

How Each Skill Gets Built

Every skill follows a six-phase lifecycle designed around two principles: verify everything against real data, and reduce effort as the platform grows. The shared integration layer built for the first skill carries forward to every subsequent one.

How we build

Skill Creation Process

Each skill follows a proven six-phase lifecycle. The shared integration layer means subsequent skills build faster and cost less.

Scope

Define what to build and why. Capture requirements through a structured interview; lock the MVP before any tool work begins.

Your input Half day

Explore

Verify every data source returns what we need. Test against a known client; document gaps and quirks before writing a single instruction.

You verify 1 day

Build

Write the AI skill instructions and reference files. Every data point traces to a verified tool call; no guesswork, no black boxes.

1–2 days

Test

Clean-room testing against multiple client profiles. We compare AI output to real data, document failures, and fix in a separate revision pass.

You validate 1 day

Deliver

Packaged as a portable skill file you upload to Claude. No software to install; works with your existing setup from day one.

Half day

Evolve

Real-world use surfaces gotchas and edge cases. Each failure captured feeds the next version; the skill gets smarter with every run.

You flag issues Ongoing

The architecture repeats

Once the integration layer and first skill are built, subsequent skills share the same data connections, error handling, and client mappings. A full build becomes a light derivative when the foundation already exists.

~5 days First skill

~2 days Derivative skill

How It Works

Each skill follows the same pattern: pull data from the shared integration layer, apply domain-specific logic, and produce a structured briefing optimised for the specific use case.

The integration layer maps all 111 client accounts to their data across HubSpot, ClickUp, Merchant Spring, Google Drive/Calendar, and Front — handling authentication, error recovery, and client-team relationships.

Technical Approach

Built on Claude’s Skills architecture with MCP (Model Context Protocol) connections to each data source. Skills are packaged as portable instruction sets — no custom software to install, no new tools to learn. The AI reads from Sitruna’s existing systems and produces briefings in natural language.