Platform Learning

Stage 0.5 — Platform Waifinder

Stark Industries PMO · how I learned a platform I'd never usedI'd never used monday.com before. After mapping the problem in Stage 0, the next step wasn't clicking around — it was building a Waifinder. I scraped every relevant help doc and developer reference from the monday.com site, fed them into an AI copilot, and used it to guide every build decision. This wasn't a chatbot — it was a platform expert that knew column types, formula syntax, automation triggers, Connect Board mechanics, and API limitations. Every "how do I…" question got an answer backed by the actual docs.

What Went In — 30+ Platform Docs

~7,000 lines of reference.

Column types & values

Every column type, settings, constraints

Formulas & functions

Full function reference, WORKDAY arithmetic, IF/AND patterns

Automation triggers & actions

Every trigger type, every action, condition syntax

Connect Boards & mirror columns

Two-way linking, roll-up mechanics, limitations

Dashboards & widgets

Widget types, data sources, filtering

AI features & MCP tools

AI column types, Open Block prompts, API mutations

API reference & rate limits

GraphQL mutations, authentication, error codes, versioning

Forms, sprints, integrations, bulk import

Full platform surface area

What Came Out — A Copilot That Knew the Platform Cold

Connect two boards

→ Connect Boards column type, mirror mechanics, link-back setup

Formulas trigger automations?

→ No — use 'When item created' trigger + condition check on formula value. This limitation became a demo talking point.

WORKDAY arithmetic

→ WORKDAY function syntax, date column dependency, SLA deadline pattern

AI column for completeness

→ Open Block with custom prompt, {pulse.column_id} references, API-Version: 2026-10 header

Two boards structured differently

→ Portfolio: Status-based pipeline. Execution: Sprint-based groups. Structural contrast is a deliberate choice, not template default.

Widget for SLA compliance

→ Number widget with formula-backed source, or Chart widget for status breakdown

Docs to Guided Build

flowchart TD
  DOCS["30+ Platform Docs\nColumn types · Formulas · Automations\nConnect Boards · AI Features · API\nDashboards · Forms · Rate Limits\n~7,000 lines of reference"]:::source

  DOCS --> INGEST["Ingested into\nAI context window"]:::process

  INGEST --> WAYFINDER["Waifinder Copilot\nknows every column type, formula function,\nautomation trigger, API mutation,\nand platform limitation"]:::core

  WAYFINDER --> Q1["How do I\nconnect boards?"]:::query
  WAYFINDER --> Q2["Can formulas\ntrigger automations?"]:::query
  WAYFINDER --> Q3["What AI column\nfor completeness?"]:::query
  WAYFINDER --> Q4["How do I mirror\nstatus back?"]:::query
  WAYFINDER --> Q5["What widget\nfor SLA %?"]:::query

  Q1 --> A1["Connect Boards column\n+ mirror mechanics"]:::answer
  Q2 --> A2["No — use item created\ntrigger + condition"]:::answer
  Q3 --> A3["Open Block AI\n+ custom prompt"]:::answer
  Q4 --> A4["Mirror columns on\nExecution Board"]:::answer
  Q5 --> A5["Number widget\nformula-backed"]:::answer

  A1 --> BUILD
  A2 --> BUILD
  A3 --> BUILD
  A4 --> BUILD
  A5 --> BUILD

  BUILD["Guided Build\nEvery architectural decision backed by\nplatform-specific knowledge, not guessing"]:::outcome

  BUILD --> S1["Stage 1\nBoard Structure"]:::stage
  BUILD --> S2["Stage 2\nIntake Flow"]:::stage
  BUILD --> S3["Stage 3\nAutomations"]:::stage
  BUILD --> S4["Stage 4\nDashboard"]:::stage
  BUILD --> S5["Stage 5\nAI/Formula"]:::stage
  BUILD --> S6["Stage 6\nConnected Arch"]:::stage

  classDef source fill:#EFE7FB,stroke:#7C3AED,stroke-width:2px
  classDef process fill:#E6EEF6,stroke:#2F6FA8,stroke-width:2px
  classDef core fill:#ECEAFE,stroke:#4F46E5,stroke-width:3px
  classDef query fill:#FBEFD8,stroke:#A5650A,stroke-width:1px
  classDef answer fill:#DDF2EE,stroke:#0E7C6E,stroke-width:1px
  classDef outcome fill:#FBE9E7,stroke:#C23A30,stroke-width:2px
  classDef stage fill:#FFFFFF,stroke:#CDBBA8,stroke-width:1px

Why This Matters for the Interview

I didn't learn monday.com by clicking around and hoping. I built a tool that made the platform teachable. Every build decision — from the two-gate intake filter to the AI/Formula separation pattern to the Connect Boards mirror mechanics — came from the Waifinder, not from guessing. When the panel asks "how did you know that?", the answer is: I built a copilot that knew it first.

Platform Limitations Caught Before I Hit Them

Formulas

Can't trigger automations directly. Workaround: "When item created" trigger + condition check on formula value.

AI Columns

Require API-Version: 2026-10 header. Open Block type uses {pulse.column_id} references, not source_type/source_column_id.

Date Columns

WORKDAY arithmetic in formulas doesn't fire automations. Need a formula → date column → automation chain.

Mirror Columns

Only Status and Date columns support mirroring for roll-up. Sprint assignments and task details stay local to Execution Board.

Rate Limits

Complexity-based limits per minute. Batch mutations and query batching needed for mock data injection.