1. Gut check on the patterns we're seeing
0:00 – 0:20 · 20 minutes
Notes for me
Gut check on the four themes. If we're wrong about any of them, what's in proposal-direction.md is mis-aimed. Want pushback, not validation. Listen for what they overweight that we underweight.
Four themes have shown up in nearly every conversation. Want gut reactions: are these right, are we overweighting any, are we missing a fifth?
1.Federal RFP pursuit is the firm's heaviest workload — and the least supported by tools.
Proof
~80 hrs per pursued federal RFP. 12–14-hr days during pursuit weeks. Tricia carries client work concurrently. Go/no-go circle is now Jeff + Tricia, down from three. [Meeting 2]
Business case
~80 hrs × $216/hr blended = ~$17K of internal cost per pursuit. At 4–6 RFPs/year that's $69K–$104K/year direct. But the bigger number is the revenue at risk: each federal contract carries $600K–$800K/year in modifications, so a missed renewal from a burnt-out or deprioritized pursuit is a multi-six-figure event.
Ask
Is this really the heaviest, or is something else louder when you're alone with your inbox?
2.Leadership has no live view of firm performance — every read is a manual stitching job.
Proof
Time lives in Teamwork, dollars in QuickBooks, forecasts in Britt's workbook, utilization in Easy Insight. Anyone wanting a current picture pulls from all four by hand. Over-servicing is caught by Britt's weekly check — intuition, not alert.
Business case
Every $10K of un-caught over-service is a direct hit to the 15–20% net margin. Today the only thing standing between current-state and a margin leak is Britt's weekly forecast-vs-actuals check — and Jeff named that explicitly in Meeting 2.
Ask
Do you want this automated, or does the manual stitching itself have value — your way of staying close to the firm?
3.The most repeatable deliverables are still the most manual.
Proof
Federal monthly progress reports — narrative written by hand from Easy Insight exports. Federal speaking PowerPoints — ~30 YTD for one client, all edited manually off a template. Diagnostic & Communication Plans — three-fold manual assembly across multiple contributors, done for all clients. Commercial proposals — templated but copy-paste-heavy. [Sales meeting · Intake · Meeting 1]
Business case
These are the firm's most reliable, recurring outputs. The templates and inputs already exist (time entries, scope docs, past versions) — what's expensive is the execution. Roughly estimated: federal progress reports $65K–$108K/yr, federal PPT volume $35K–$52K/yr for one client alone. Volume confirmation across the federal book is the gating step to dollarize the full opportunity.
Ask
Where's the gap between "the template exists" and "executing the deliverable" widest? Which of these eats the most of your most senior people's time?
4.Critical knowledge lives in people's heads — not in shared systems.
Proof
"Honestly, mostly in people's heads." [intake form] Strategy drift has already "dinged" RedFlash on commercial accounts where original commitments faded after a mid-engagement pivot. Tricia named that 80% of federal revenue rests on two individual contacts.
Business case
An institutional-memory layer is two things at once: revenue protection on commercial accounts where strategy-drift has already cost trust, AND continuity insurance against the two-contact federal concentration — where losing one contact is a multi-six-figure revenue event.
Ask
Central to what's actually slowing you down, or peripheral?
Closing
Are we hearing this right? Missing a fifth? Overweighting one?
2. Where this could go sideways on the federal side
0:20 – 0:40 · 20 minutes
Notes for me
Federal is 80% of revenue. Any build that mishandles federal posture invalidates itself and the relationship. This is where we get the political shape before we recommend — not after. Putting our read on the table; not asking them to architect.
Four specific speed bumps any RedFlash-AI integration runs into. Want to know which are bigger, smaller, or missing.
1.Recording posture and transcript-driven workflows.
Tricia, M1
"Government side, please no recordings typically… AI notetakers we get away with." [~40:36]
Jeff, M1
"Across the company [we] haven't been as consistent." [~41:00]
Business case
The Client Knowledge Agent, meeting-to-task automation, and RFP-draft-inputs builds all use transcripts as inputs; federal posture cuts inputs on 80% of the book, commercial inconsistency cuts the rest.
Verify
Which input pattern do we design against — notetaker-only, structured manual capture, post-call dictation? Push for firm-wide recording consistency in the proposal, or is that change you'd rather lead internally?
2.Tightening federal-timekeeping controls.
Melissa, M2
(Paraphrased per sensitivity flag.) Federal-timekeeping enforcement isn't happening as required — the late-entry blocking that should run, doesn't. She named AI integration as the fix she'd welcome. [~12:29–13:35]
Business case
GSA audits run annual-to-biennial; current enforcement gap is unmitigated audit-risk exposure on contracts carrying $600K–$800K/yr in modifications each.
Verify
Already on your roadmap, or did her ask surface it? Does an AI-driven enforcement narrative sit comfortably internally and with your auditor — or does putting it in writing create a paper-trail problem? Melissa isn't here; we'll close the loop with her after.
3.QuickBooks Desktop, GSA reports, migration appetite.
Melissa, M2
"If we could use the AI to come up with some of these reports financially. I might be able to make the move over to the online platform." [~13:40–14:21]
Business case
Multiple financial-side builds (leadership dashboards, DCAA automation, mid-month over-servicing visibility) all stall until QB-desktop extraction or QB Online migration is resolved.
Verify
Is QB Online on the table in the next 12 months if we solve the report-replication problem? Who owns that decision — Melissa, Jeff, both, someone else?
4.Two-contact federal concentration.
Tricia, M2
"The revenue really is… resting on the shoulders of two client contacts, right?" [~58:36–58:44]
Business case
80% of revenue rests on two individual contacts; framing a knowledge layer as continuity insurance positions the spend as revenue protection, not overhead.
Verify
Lean into continuity as a named value of that build, or keep it as a quiet operational benefit and skip the framing?
Closing
Which of these is bigger than we're treating it, smaller, or missing?
3. Where the work lands: adoption, ownership, and where the gains go
0:40 – 1:00 · 20 minutes
Notes for me
Sections 1 and 2 pressure-test what we heard. This section answers the question that gates how we sequence the proposal: who absorbs each recommendation, who champions it, and where the reclaimed capacity goes. Lean-crew is a stated posture — that's the constraint we have to design into, not around.
Britt is in the room — every Britt-adjacent reference below is anchored in a quote she heard live in Meeting 2.
Business case · section
Every recommendation in the proposal lands on someone's desk on Day 31. Without an internal owner, automation becomes shelfware — the ROI depends on adoption ownership as much as on the build. And whether reclaimed capacity reads as savings, capacity gain, or growth fuel determines what we should sequence first.
Context going in — not re-asking, just setting the floor:
What we already know
Jake's departure already rippled.
Melissa felt the Fathom-backend gap; the go/no-go circle dropped from three reviewers to two (Jeff and Tricia). [Melissa ~07:14, ~34:24; Tricia ~38:55]
Business caseThe team absorbing our recommendations is already thinner than the one that built the current operating spine. We're designing for the firm that exists after Jake, not before.
Jeff named the automation intent on the record.
"That would be cool if there were some automations around that that wouldn't force Britt to be the person bird-dogging all of that, all day long." [Jeff, Meeting 2, ~09:10]
Business caseWork coming off Britt's plate is a desired outcome, not a side-effect we need to justify. The question is where it goes next.
Lean-crew is named as a choice.
Tricia framed 12–14-hr RFP days as a function of running lean — a posture, not a complaint. [~47:22]
Business caseWe're designing for a firm that has chosen to stay lean, not a firm that's hiring. The automation has to do the absorbing, because no new headcount will.
What we want from you
1.Where does the reclaimed capacity go? If we move bird-dogging-and-reconciliation work off Britt's plate, and 80-hr-RFP work off Tricia's plate, where does that time get redeployed? Higher-leverage work, more federal RFP bandwidth, growing the commercial book back up, simply giving people their evenings back? The answer changes which build we sequence first — capacity-for-growth fuel and capacity-as-relief recommend different orderings.
Business caseWhether reclaimed capacity reads as savings, capacity gain, or growth fuel determines what we prioritize building first — and how the proposal gets evaluated internally.
2.Day 31 owners. Each of the four themes in Section 1 lands on someone's desk after we go live. We have guesses; want to confirm — and surface where we're wrong.
- Federal RFP support — Tricia is the obvious operator, but she's already at 80 hrs/pursuit. Does the build run for her, or does someone else operate it on her behalf?
- Leadership financial view — Britt and Melissa together is our read. Who absorbs the new dashboard layer, and who's the routing point for exceptions?
- Repeatable deliverable automation — likely owners differ by deliverable (progress reports Britt-anchored; federal PPTs ambiguous; comm plans Tricia-anchored?). Worth walking through who picks up each.
- Institutional memory layer — this is the theme without an obvious owner today. Who would champion it, or do we need to write that role into the recommendation itself?
Business caseWithout a named internal owner per build, every recommendation has a higher chance of becoming shelfware. Naming owners now lets us write rollout support into the proposal where the seat is empty.
3.AI champion — formally or de facto. Beyond the day-to-day owners above, is anyone at RedFlash on a trajectory to become the in-house AI lead — someone the team would route AI questions through? If not yet, do you want the proposal to write that role into the recommendations rather than assume it exists?
Business caseThe four themes touch every tier of the firm. Without a connective governance layer, individual build adoption can succeed while firm-level AI maturity stalls — which is precisely the ungoverned-AI pattern you named in the intake.
Closing
Three months after the builds go live, what's the failure mode you'd be most worried about? (We'd rather hear it now than discover it then.)