Athletes University already owned the hard-won knowledge: course material, the audience's objections, organic content frameworks, case studies and client-success context. The bottleneck was making that body of work available at the moment a department needed it.
Evidence at publication: hundreds of source chunks were loaded into gbrain; four agents, six bindings and active Caddy, Command Center, Ollama, billing-proxy and OpenClaw services were verified on 30 May 2026. Google Workspace remained authorization-ready, and routine adoption was not quantified.
The company
Will Straka runs Athletes University and The Scholarship System. The operating knowledge spans a course, organic content, ideal-avatar detail, objections, offer material, classroom context, sales and client success.
That corpus has value precisely because it is specific. A generic model can suggest a marketing post or sales response, but it cannot know Will's offer, teaching framework or audience objections unless that information is structured and retrievable.
The constraint
The knowledge existed, but it was too large and scattered for normal chat. Each useful request risked beginning with another manual paste of course context or another explanation of the avatar.
One assistant would create a second compression problem. Marketing, sales and fulfillment use overlapping knowledge, but they apply different standards. A content angle and a client-success answer may draw from the same course, yet they should not be produced through the same operating instructions.
Why existing tools were not enough
Documents preserve information; they do not route work. Slack and Discord carry requests; they do not guarantee that the right specialist retrieves the right portion of a course. Google Workspace can eventually support email, calendars and files; an authorization scaffold alone is not a live business workflow.
The missing layer was a searchable knowledge foundation paired with departmental roles and an honest boundary between retrieved intelligence and provider actions that still required client consent.
Transformation map
Before
Course files + avatar notes + SOPs -> Will finds and pastes context -> general assistant -> draft
After - knowledge layer live, adoption pending
Department request -> specialist agent -> gbrain retrieval -> source-grounded draft -> human review -> future approved provider action
The source corpus becomes shared infrastructure without flattening every department into one voice.
One workflow, end to end
- Input: A request for an organic content asset enters the marketing channel.
- Routing: The
oc-marketingbinding directs it to the marketing specialist rather than command, sales or client success. - Retrieval: The agent searches the ingested Scholarship System course, organic ecosystem, avatar and offer chunks for relevant source context.
- Production: It prepares a draft grounded in those materials and keeps unsupported assumptions visible.
- Human gate: Will or the team reviews the positioning and any claim before publication.
- Provider boundary: A future Drive, Docs, Calendar, Gmail or Analytics action waits for completion of the single Google Workspace OAuth consent.
- Record: Approved material and lessons can return to the structured workspace for later sales or client-success retrieval where appropriate.
What changed
- Built: Four workspaces and agents for main command, marketing, sales and client success, backed by structured knowledge retrieval.
- Connected: The live runtime includes Slack, Discord, PostgreSQL, Ollama, gbrain, Command Center, Caddy and model/billing routing.
- Tested: Hundreds of chunks from course, organic-content, avatar, classroom and offer materials were ingested and made available to the specialist structure.
- Verified: Four agents, six bindings and the core services were active at the 30 May 2026 verification point.
- Actively used: The source does not retain routine task volume, published content volume or a measured change in sales or fulfillment. Those outcomes are not inferred.
- Remaining dependencies: Will must complete Google Workspace OAuth before Gmail, Drive, Calendar, Sheets, Docs and Analytics can be treated as connected action lanes; sustained department adoption also needs evidence.
- Financial result reported by client: None retained.
Proof
- Knowledge receipt: Hundreds of chunks loaded from the operating workspace.
- Corpus receipt: Scholarship System course material, organic ecosystem, avatar/training, classroom and offer context are named in the retained record.
- Agent receipt: Four department roles and six bindings.
- Runtime receipt: Caddy, Command Center, Ollama, billing proxy and OpenClaw services active on 30 May 2026.
- Claim boundary: Google Workspace is described as scaffolded for one consent, not live; ingestion is not represented as adoption.
Technology - revealed last
The environment runs OpenClaw 2026.5.2 on a VPS with Slack, Discord, model routing, PostgreSQL, Ollama, gbrain, Command Center and Caddy. SmartLead and Fathom support are included in the operating context. The Google Workspace skill uses gws-auth, with one planned OAuth consent for Gmail, Drive, Calendar, Sheets, Docs and Analytics rather than a second competing provider route.
Client quote
No source-backed verbatim client quote is published for this case. The public evidence is the documented knowledge corpus, agent structure and runtime state; RECAST does not convert project notes into praise.
Next transformation
See how LEED Agency turned a large client file base into persistent reporting memory. Both systems make proprietary context retrievable, but one serves education departments and the other recurring agency analysis.
Get your AI blueprint
If your best material is trapped in folders and repeatedly pasted into chats, your first agent may be a knowledge operation. Book a free AI audit to map the corpus, department boundaries and approval path before connecting external actions.
Source record
OPENCLAW-DEEP-CLIENT-CASE-STUDIES.pdf, page 16.RECAST_CASE_STUDIES_2026_UPDATED.pdf, page 8, documentRA-CASES-001, version 2.0.RECAST_WEBSITE_MASTER_BLUEPRINT.md, sections 10.4, 11 and 58.