# Source Pack: June Cohort Fireside Chats (Sara Brown)
## Source Boundary
This pack separates verified session evidence from outside context. It uses:
1. Portal event 59: https://portal.raidguild.org/events/59
2. Summary artifact: source request `ee0ee806-e3bb-4622-b484-8c9346492df4`, artifact `0aa17b1f-206c-4de1-a03a-8b451db80dfb`
3. Transcript artifact: source request `ee0ee806-e3bb-4622-b484-8c9346492df4`, artifact `3fb54689-8bf1-49f6-91a0-2b6c6bab65a6`
4. Prior intake report on this request: artifact `8fb8bb38-6703-4026-85e3-784daa9546a4`
No outside industry research was used to expand or validate the advertising/AI claims. Any external framing in later steps should be labeled separately.
## Portal Data
- Portal event id: 59
- Title: June Cohort Fireside Chats (Sara Brown)
- Session type: brownbag
- Starts: 2026-06-25T19:00:00.000Z
- Ends: 2026-06-25T19:30:00.000Z
- Location: RaidGuild Discord
- Discord scheduled event id: 1514405041228877856
- Discord sync status: synced
- Source status: summarized
- Recording URL: not present on the Portal event
- Portal resources: none attached at source-pack time
- Related Portal thread: How to RaidGuild - Field Experience from the Edge, thread id 5
## Direct Session Evidence
### Participant / Perspective
- Sara Brown introduced herself as a friend of RaidGuild working in digital advertising.
- Her work centers on programmatic campaigns across smart TV, audio platforms, mobile web, desktop/mobile web traffic, and video.
- She stated that she does not specifically run search engine or social media ad placements as her core platform work.
- She described using multiple DSPs, including The Trade Desk, Amazon, and DV3 as named examples in the transcript/summary. Verify the exact spelling of `DV3` before public use.
### Current AI Use
- Sara described AI as already having a heavy impact on programmatic advertising, especially targeting, cost efficiency, reporting, and platform-level campaign tooling.
- Organizational AI adoption was described as grassroots rather than top-down: employees are encouraged to learn and find use cases, with access to Claude and training resources.
- Microsoft Copilot is used for professional email refinement and possibly reporting support.
- Claude is a desired tool for more visual and stronger reporting workflows.
- Advertising partners already provide platform-specific AI tooling for targeting, inventory, and efficiency work.
### Barriers / Constraints
- Data security was identified as a primary blocker because programmatic advertising can involve sensitive first-party data.
- Walled-garden platform environments were described as limiting flexibility and forcing users into platform-specific partner AI tools.
- Current AI workflows still involve manual movement across tools, including email rewriting and work outside the active thread/context.
- The session does not verify a specific employer, client, platform contract, private dataset, or internal company process beyond Sara's general professional perspective.
### Agentic Workflow Opportunities
- Market research: Sara proposed an agentic workflow that could gather data across platforms to define target audiences, reach, and frequency requirements.
- Campaign QA: Sara described a need for an agentic auditor that flags human error, monitors performance, and suggests optimizations for budget, reach, or testing.
- Unified interface: Sara imagined managing campaigns and reporting from one place, potentially with Slack-style integration, instead of switching between advertising and research platforms.
- Internal workflow builders: Dekan connected this to medium-to-large organizations letting employees act as internal development teams for specific workflows; Sara found that promising.
### Human Judgment / Brand Work
- Sara was optimistic about AI increasing knowledge and capacity to think beyond existing patterns.
- She cautioned against total automation.
- She emphasized that human checks and balances remain important for brand storytelling and preserving institutional value.
### Follow-up Mentioned In Session
- The meeting summary records a next step for Dekan Brown to share synthesized notes and recordings with the guest once finalized.
- The closing discussion mentions exploring manual advertising workflows in a future follow-up session.
## Candidate Angle Seeds From Direct Evidence
These are not final content recommendations; they are source-backed seeds for angle-plan.
1. **Agentic QA for messy campaign work**
- Evidence: Sara named campaign QA, performance monitoring, budget/reach optimization, and human-error checks as desired agentic functions.
- Useful because it is concrete and grounded in a real workflow pain.
2. **AI adoption happens where operators have room to experiment**
- Evidence: Sara described a grassroots adoption model with Claude access/training but no rigid top-down use mandate.
- Useful because it reframes AI adoption around operator creativity, not generic corporate transformation language.
3. **The blocker is not imagination; it is data boundaries**
- Evidence: data security and walled gardens came up as barriers to cross-platform AI workflows.
- Useful because it connects practical AI ambition with deployment constraints.
4. **Unified campaign command center as a field-note pattern**
- Evidence: Sara wants one workflow surface for campaigns, research, reporting, and possibly Slack-style coordination.
- Useful because it maps well to RaidGuild's interest in building custom tools around real operator workflows.
5. **Human brand judgment remains the high-value layer**
- Evidence: Sara emphasized checks, balances, and brand storytelling as things automation should support rather than erase.
- Useful because it avoids simplistic replacement narratives.
## Outside Context
No outside knowledge was used as evidence for claims in this source pack. The only non-session context captured is Portal continuity metadata: the event is related to the Portal thread `How to RaidGuild - Field Experience from the Edge`.
If later steps use RaidGuild campaign framing, the angle-plan should explicitly label that as editorial context, not session evidence.
## Gaps And Uncertainties
- **Name spelling:** Portal title uses `Sara Brown`; transcript text intermittently appears as `Sarah`. Use Portal spelling unless the guest confirms otherwise.
- **Profile linkage:** the Portal event has no linked speaker profile. Do not imply a verified Portal profile exists.
- **Recording:** Portal event has no direct recording URL. Use summary/transcript artifacts unless a recording resource is added later.
- **Transcript privacy:** early setup chatter includes personal contact details and should not be exposed in public outputs.
- **Platform spelling:** verify `DV3` before publishing; it may be a transcript artifact error or shorthand.
- **Employer/client specifics:** do not infer employer, client names, spend levels, contracts, or campaign performance details.
- **Consent boundary:** public use of the guest's professional examples should remain generalized unless there is explicit approval for more specific publication.
## Candidate Follow-up Questions
1. Should the public-facing materials use `Sara Brown` exactly as the Portal event title does?
2. Is the guest comfortable with public discussion of programmatic advertising examples such as DSPs, walled gardens, and first-party data constraints?
3. Should the final source resource include only the cleaned summary, or should the raw transcript remain private/internal?
4. Is there a recording URL or cleaned audio artifact that should be attached before publish-session-artifacts?
5. Was `DV3` transcribed correctly, or should it be corrected before public drafting?
## Recommended Next Step
Proceed to angle-plan using this source pack. Keep direct session claims tied to the summary/transcript artifacts and keep any RaidGuild editorial framing separate from the guest's statements.