Context Gathering Report · Dave Vaillancourt · Last 90 Days

Remove Dave’s Repeat Drag.

Dave is spending too much executive time reconciling schedules, budgets, invoices, hardware logistics, benefits, and buried Slack decisions. The fastest win is not a new process. It is a quiet back-office agent that watches the systems, fixes the obvious mismatches, and only surfaces exceptions.

118Dave Slack posts in the Apr 2–14 TeamLeads sweep
40+Dave Slack messages summarized Apr 20–May 4
268visible channels enumerated by ForgeBot
62channels readable with current bot membership
Access reality: ForgeBot’s current bot token cannot use Slack workspace search and hit not_allowed_token_type. It also is not a member of 200 visible channels, so this report combines the live Slack scan I could verify today with existing TeamLeads Slack sweeps and Dave status reports already in the repo. Sensitive executive-only contexts are not quoted.

What Dave’s Posts Say

“This seems to be the biggest pinch point in the sales to production pipeline… If that’s not happening we [are] running projects at high risk.”

“I’m so ridiculously behind with JD RWC that I’m trying to just trust my team leads and I’ll enter in the hours into the spreadsheet.”

“Can you send me the spreadsheet version so I’m less likely to make mistakes when transferring the data to the budget spreadsheet?”

Operating Read

Dave does not need ceremony. He needs fewer loose ends. The useful bot posture is: watch everything, reconcile numbers, draft the fix, update internal trackers when confidence is high, and make the exception obvious. Avoid “AI strategy.” Deliver completed back-office work.

Top 20 Pain Points ForgeBot Can Remove or Partially Remove

1

Sales-to-production hour handoff risk

Signal: Dave explicitly called this the “biggest pinch point” and said projects are high risk if sales numbers are not based on River/Rachel production calculations.

No-HITL path: Closed-loop intake bot compares proposed milestone schedules, production-hour budgets, and named production owners before a proposal leaves draft; it flags mismatches and posts a ready-to-approve correction packet.

2

Budget spreadsheet transfer mistakes

Signal: Dave asked for spreadsheet-form source data because he is less likely to make mistakes moving schedule/hours into the budget spreadsheet.

No-HITL path: Closed-loop spreadsheet normalizer ingests Slack/Excel/Google Sheets, validates totals, maps people/month/project codes, and updates the budget staging sheet with a diff and audit log.

3

Invoice schedule math mismatches

Signal: MRIGlobal invoice schedule totaled $244,570 while the contract total was $244,755; Dave manually found and patched the $185 delta.

No-HITL path: Closed-loop contract/invoice reconciler checks totals, dates, milestone labels, and payment math across SOWs, contracts, invoices, and budget rows.

4

Contract start/end-date ambiguity

Signal: Dave repeatedly asked whether 6/1, 7/1, holiday buffers, start dates, and end dates were acceptable.

No-HITL path: Closed-loop milestone calendar builder proposes schedule options from constraints, holiday buffers, staffing load, and cash-flow needs.

5

Contract paperwork status chasing

Signal: Dave checked Halliburton paperwork, MRIGlobal contract mods, JD support terms, and AES invoices across Slack.

No-HITL path: Closed-loop contract-status tracker watches Slack/email/SharePoint/Salesforce and maintains one live “what is waiting on whom” board.

6

Hardware logistics for John Deere RWC

Signal: Dave coordinated joystick/pedal shipments, serial numbers, 12V power supplies, PCB swaps, enclosures, labels, and receiving confirmations.

No-HITL path: Closed-loop hardware kanban watches Slack + shipment/tracking data, maintains serial inventory, tells each owner exactly what to ship/keep/photograph, and escalates exceptions.

7

Manual procurement verification

Signal: Dave asked whether a $15,000 Brain Exchange invoice was true and expected before paying it.

No-HITL path: Closed-loop invoice guard matches vendor invoices to budgets, events, POs, approvals, and prior Slack context; it returns pay/hold rationale without exposing bank credentials.

8

Benefits and 401(k) admin

Signal: Reports show Kaiser double-coverage resolution, ADP 401(k) setup, incorrect ADP match configuration, and enrollment follow-up.

No-HITL path: Closed-loop benefits admin watcher reads provider/email portals where allowed, drafts exact correction requests, tracks deadlines, and summarizes remaining human-only actions.

9

Contractor invoice completeness

Signal: Reports cite March/April contractor hours, no-invoice cases, late invoice checks, and per-person hour breakdowns.

No-HITL path: Closed-loop invoice/hours reconciler compares reported Slack hours, submitted invoices, project budgets, and missing invoices; sends internal reminders only when authorized.

10

Commission-policy ambiguity

Signal: Dave confirmed Mary’s commission on MRIGlobal/JD support and modeled compensation implications.

No-HITL path: Closed-loop commission calculator applies current policy to opportunities and flags exceptions before leadership debates them in Slack.

11

Trade-show and travel budget tracking

Signal: Dave corrected ConExpo/AES budget numbers and reframed combined trade-show/travel remaining spend.

No-HITL path: Closed-loop event-budget tracker reconciles invoices, travel, booth, shipping, cash-flow year, and remaining allocation.

12

Production authority clarity

Signal: Dave published the River/Brandon/PM/contributor authority chain and role updates.

No-HITL path: Closed-loop org-rules bot detects conflicting assignment language and reminds the channel who owns the decision path.

13

Resource-capacity uncertainty

Signal: Dave asked whether Halliburton start dates were doable with the current team and how work should overlap with ARA.

No-HITL path: Closed-loop capacity forecaster reads budgets, Jira load, staffing, and proposals to score “safe / tight / unsafe” start-date options.

14

Unity license evidence hunting

Signal: Dave said he did not have easy access to contract wording and asked others to hunt license terms.

No-HITL path: Closed-loop license vault search indexes SharePoint/email/Slack for agreements, terms, renewal dates, and client distribution constraints.

15

Payment collection drag

Signal: Dave celebrated JD finally paying a change order and said chasing payment took more hours than the task itself.

No-HITL path: Closed-loop receivables tracker follows invoices, aging, client replies, and change-order status; it prepares escalation notes before cash pain shows up.

16

SOW ambiguity

Signal: Dave found an MRIGlobal contract confusing and said “the SOW isn’t really a SOW,” asking for dates and assigned team.

No-HITL path: Closed-loop SOW sanity checker extracts deliverables, dates, assumptions, project owners, and missing terms before kickoff.

17

Decision history buried in Slack threads

Signal: Dave’s evidence spans long threads, replies, DMs, proposal channels, and production channels.

No-HITL path: Closed-loop decision ledger turns accepted Slack decisions into searchable records with owner, date, source link, and downstream systems touched.

18

Finance category cleanup

Signal: Reports show Dave overhauled Quicken categories for salary, distributions, retirement, and taxes.

No-HITL path: Closed-loop bookkeeping categorizer proposes category mappings from historical rules and flags new/ambiguous transactions for later review.

19

People-compensation timing

Signal: Dave weighed Kris raise timing against invoice processing and contractor negotiation behavior.

No-HITL path: Closed-loop compensation memo builder collects current rates, cash-flow runway, retention risk, and precedent into a concise decision packet.

20

AI opportunity trapped in ad hoc experiments

Signal: Dave tested image generation for technical PCB drawings and tied AI to reducing development hours.

No-HITL path: Closed-loop AI-opportunity scout watches project pain, estimates automation return, and turns “this should be automated” into scoped internal tasks.

Credentials / Access Needed From Adam

1. Slack workspace search/read

Best: a Slack user token or approved app scopes that allow search.messages, private-channel reads where invited, thread reads, and permalink generation. Alternative: add ForgeBot to the specific channels Dave cares about.

2. Microsoft 365 / SharePoint / Outlook

Read access to business email, contracts, SharePoint folders, and calendars. This unlocks contract-status tracking, invoice matching, vendor follow-up prep, and source-document retrieval.

3. Jira + project budget sources

Jira API access plus the budget workbook / ForgeBooks data source. This unlocks capacity forecasts, hour variance checks, milestone validation, and project-risk surfacing.

4. Salesforce / proposal source

Read access to opportunities, proposals, SOW drafts, contacts, contract stages, and expected values. This lets ForgeBot catch proposal-to-production mismatches before they become Dave cleanup.

5. Accounting and invoice sources

FreshBooks/ForgeBooks/Quicken-export access where available, plus vendor invoice folders. Bank movement stays Adam-gated; the bot can still reconcile, flag, and prepare pay/hold packets.

6. Benefits / HR portals where safe

ADP/Kaiser read access or export access is enough for tracking enrollment, match settings, deadlines, and correction status. Any personal or external submission should stay explicitly permissioned.

Five Alignment Questions For Dave

1. What response style will Dave actually use?

2. Where should ForgeBot do the work first?

3. What is Dave’s preferred interruption threshold?

4. How much autonomy should ForgeBot have?

5. What evidence does Dave trust most?