Skip to content
Michelle McGovern StewartStrategy / Story / Systems

05 / Information Organization & Retrieval

Making project information searchable.

An architecture firm’s project history lived inside its old proposals, described differently from one document to the next. I turned that fragmented institutional knowledge into a searchable binder, a structured project database, and an AI agent that finds and checks the details.

Client
Ruhnau Clarke Architects
My role
Marketing Director
Status
Implemented at Ruhnau Clarke
Draws on
  • Strategy (not a focus)
  • Story
  • Systems
Capabilities
Knowledge managementInformation architectureAirtableRecord matchingAI-assisted retrievalSortable AI agent

Illustration

From scattered documents to answers you can check

  1. The starting point

    Historical proposals

    Project information spread across years of documents, described inconsistently.

  2. Layer 1 · Source

    Searchable PDF binder

    Historical proposal material consolidated into one place that can be searched.

  3. Layer 2 · Structure

    Airtable project database

    One record per project, connected by DSA application number.

  4. Layer 3 · Retrieval

    Sortable AI agent

    Searches the project information and validates it against the records.

How the pieces relate. An illustration of the workflow, not a screenshot.

The challenge

The tension

A firm with a long history of school projects should be able to answer a simple question quickly: what have we done that’s like this?

At Ruhnau Clarke, the answer lived inside proposal documents. The same project could be described differently from one proposal to the next, details didn’t always agree, and nothing was organized for retrieval. Finding project information was slow, and confirming it was right was slower.

That’s not a writing problem. It’s an information problem, and it quietly taxes every proposal, interview, and conversation that depends on the firm’s experience.

The strategy

Three layers, each with a job

Rather than one big fix, I built three connected layers. Each one solves a different part of the problem, and each one makes the next one possible.

  1. 01

    Gather it

    I consolidated historical proposal material into a single searchable PDF binder. The original language stays intact and in context, so there is always a source to go back to.

  2. 02

    Structure it

    I developed an Airtable project database that gives each project one record, instead of a different description in every document that mentions it.

  3. 03

    Retrieve and check it

    I developed a Sortable AI agent to search the project information and validate it, so answers come back faster and can be checked against the records.

The insight

Names change. Numbers don’t.

The hardest part wasn’t storing the information. It was knowing which records described the same project. Names shifted between proposals, phases were described differently, and similar projects at the same campus were easy to confuse.

California public school projects are reviewed by the Division of the State Architect, and each one carries a DSA application number. I used those numbers as the connecting key, matching inconsistent project records to the project they actually describe.

Illustration

How inconsistent records become one project

  • In one proposal

    The project’s full name

  • In another

    A shortened or informal name

  • In a third

    A phase or building described separately

The connecting key

DSA application number

Assigned by the state, consistent across every document.

The result

One project record

In the Airtable database, traceable back to the binder.

An illustration of the matching logic, not actual project data.

My contribution

What I built

I identified the problem, designed the approach, and built each layer: the consolidated PDF binder, the Airtable project database, the record matching by DSA application number, and the Sortable AI agent. All of it was implemented at Ruhnau Clarke.

The point

Why it matters

Every proposal, interview, and project story depends on getting the firm’s experience right. Structured, searchable project information means that experience stops depending on who happens to remember where something was written down.

Good storytelling needs good source material. This is the part of the work that makes the stories accurate.

The work

Explore more work.

Back to the work

Brand, story & strategy

  1. 01Brand Strategy & ExperienceFirefighters Community Credit UnionNot just for firefighters.
  2. 02Strategic Marketing & DifferentiationRuhnau Clarke ArchitectsA different kind of proposal.
  3. 03Brand Voice & Demand GenerationNBS Government Finance GroupLess labor, more wins.
  4. 04Content Strategy & StorytellingRuhnau Clarke ArchitectsSeeing it from their side.

Systems building

  1. 05Information Organization & RetrievalRuhnau Clarke ArchitectsYou’re reading this one
  2. 06Research & Opportunity IntelligenceRuhnau Clarke ArchitectsResearching the opportunity before writing the proposal.
  3. 07Marketing Workflow & AutomationNBS Government Finance Group, Ruhnau Clarke Architects, Firefighters Community Credit UnionBuilding better marketing workflows.

Start a conversation

Ready to see it differently?

Let’s talk