SGA Dental Partners · Internal Reference

AI Enablement
Workstream

A shared reference for the tools and systems the AI Enablement team uses together, so our work stays connected and builds on itself.

Owner: Dakota Milner Audience: SGA AI workstream (internal) Cadence: Reviewed in daily AI standup Version: v0.3 · 2026-04-21
Why this page exists

A shared foundation for the team.

When we use the same core tools, our dashboards, data pipelines, and creative assets connect naturally. This page captures the tools we've chosen together and the decisions still being worked through.

Our shared approach

Wherever possible, the team's work connects to the same database, the same DAM, and the same task system. Confirmed tools are listed on the following slides. Open decisions are being finalized collaboratively in standup.

The goal is simply to make it easier for each of us to build on one another's work.

The stack at a glance

What's confirmed, what's in progress.

AI Assistant
Confirmed
Claude

Claude Code and the API power our internal tooling and agents.

Source Control
Confirmed
GitHub

Home for our code, skills, configurations, and shared references.

Hosting
Confirmed
Railway

Hosts our backend and primary database (app.sga3p.com).

Operational Database
Confirmed
PostgreSQL (Railway)

System of record for operational data.

Analytics Warehouse
Confirmed
ClickHouse

Where reporting and analytics data is served from.

Orchestration
Confirmed
Temporal

Runs our scheduled and long-running workflows.

BI & Reporting
Confirmed
Power BI

Our reporting layer, which reads from ClickHouse.

Communication
Confirmed
Microsoft Teams

Standups and day-to-day team communication.

Task Management
In Progress
To be selected

Decision lead: Sarah

DAM
In Progress
To be selected

Decision lead: Mike

Knowledge Base
In Progress
To be selected

Decision lead: Amy

Database Stewards
Assigned
Scott & Dakota

Support and maintain the Postgres setup.

Confirmed · Data & Backend

PostgreSQL on Railway is our primary database.

Stewards Scott & Dakota
Status: In Production

How it works

Railway hosts our backend and our Postgres database. Applications, ingestion workflows, and integrations all read and write here for operational data.

Analytics data flows onward into ClickHouse, and Power BI dashboards read from there rather than from Postgres directly.

Before adding a new datastore

Please check in with Scott or Dakota first. If Postgres doesn't fit a particular use case, we'll look at extending the shared setup together rather than running a parallel one.

This includes SQLite files, Supabase projects, Airtable bases, and similar temporary datastores.

In progress

Task management.

Decision lead Sarah
Status: In Progress

We're selecting a single place to track work across the AI workstream. Once chosen, it will replace the mix of spreadsheets, Teams threads, and email chains currently in use.

What we're looking for

Important: clear task ownership, status visibility, and the ability to link tasks to GitHub PRs and documents.

Helpful: Teams integration, support for recurring work, and a reasonable cost for around ten seats.

Options under consideration: ClickUp, Asana, Linear, Monday, Jira, GitHub Projects, and Microsoft Planner.

In progress

Digital Asset Management (DAM).

Decision lead Mike
Status: In Progress

A single home for creative — brand materials, campaign files, social posts, logos, and templates. Having one source makes assets easier to find, reuse, and keep current.

What we're looking for

Important: version history on assets, a clear folder structure by practice and brand, and permissions for internal versus agency access.

Helpful: previews for creative review, tagging and search, and integration with Canva or Adobe.

Options under consideration: Brandfolder, Bynder, Canva (team library), Frontify, SharePoint, Google Drive (shared drives), and Notion databases.

In progress

Knowledge base.

Decision lead Amy
Status: In Progress · options below

The central home for what the team needs to reference — SOPs, runbooks, onboarding materials, decisions, playbooks, and meeting notes.

Notion
Popular with AI-forward teams. Strong search, flexible databases, and clean templates for SOPs and playbooks.
Modern workflows
Microsoft SharePoint
Integrates directly with Teams and Microsoft 365, which the team already uses. Low adoption friction.
Microsoft alignment
Confluence
Structured page trees and strong permissioning. Pairs well with Jira if that becomes the task tool.
Structured documentation
GitHub Wiki / Pages
Keeps documentation alongside the code. Well-suited for technical content; less accessible for non-technical contributors.
Technical content
Coda
A docs-and-database hybrid. Flexible for interactive documents, with a slightly steeper learning curve.
Flexible power users
Google Drive (current state)
Lowest friction to start with, though search and structure are limited. Useful as a baseline reference.
Baseline reference
How we use GitHub

Working in GitHub together.

GitHub is where our code, configurations, and shared references live. A few light practices help the team collaborate smoothly.

Structure
One repository per scopeEach workstream, product, or skill set has its own repository so things stay easy to find and maintain.
Structure
README as the front doorEach repository's README explains what it is, how to run it, and who to ask. Keeping it current helps everyone onboard faster.
Workflow
Branch, PR, mergeWork happens on a branch, gets reviewed through a pull request, and then merges into main.
Workflow
Small, frequent commitsCommitting regularly with messages that explain the reasoning behind each change makes history easier to follow.
Safety
Keep secrets out of commitsAPI keys, passwords, and .env files belong in Railway environment variables or GitHub secrets. If something gets committed by mistake, rotate it right away.
Safety
Use .gitignoreKeep node_modules, build outputs, and local configuration files out of the repository. Every new repo starts with a .gitignore.
Collaboration
Use Issues for trackingUntil a task tool is selected, GitHub Issues is a good place for bugs, feature requests, and follow-ups. Link them in related PRs.
Collaboration
Request reviewers on PRsTag the person closest to that area when opening a PR. The team aims to review within one business day.
Discovery
Pin key repositoriesPinning the AI workstream repos on the SGA GitHub organization page makes them easy for everyone to find.
Discovery
Share references via PagesDocuments like this deck can be published through GitHub Pages for a single, always-current link.
Shared agreements

Five light agreements we work from.

  1. Build on the shared stack.If a tool you'd like to use isn't listed here, bringing it to standup first helps us keep things aligned.
  2. Data flows in one direction.Source systems feed Postgres and ClickHouse, and Power BI reads from there. Temporal handles ingestion.
  3. Creative lives in the DAM.Once the DAM is selected, it becomes the shared home for campaign files and brand assets.
  4. Keep this page current.When an open decision is resolved, updating the deck promptly keeps everyone working from the same reference.
  5. Share work where the team can find it.Keeping work in GitHub, the task tool, the database, or the DAM makes it accessible to everyone.
Next steps

A shared foundation.

Sarah, Mike, and Amy will bring recommendations to standup this week. In the meantime, the rest of the team can build against the confirmed tools. As the open decisions close, this deck will be updated so we're all working from the same reference.

Questions: Dakota Deck location: pdakotamilner.github.io/sga-ai-enablement-reference Updated: 2026-04-21
1 / 10
Click or use ← → keys