Applied AI Labs for Nonprofits: the rolling agenda

All decisions from this session

• No AI notetaker on WF-hosted meetings. WF security policy; fine when RW Institute hosts.
• Program structure confirmed: 2 cohorts up to 25, seven weekly 90-minute sessions each. Workable facilitation size while reaching more people.
• Curriculum philosophy: strategic capability over tool-specific tactics. Tool tactics go stale within months; judgment about data, model choice, and consequences endures.
• Every participant gets a paid Claude account, with other tools introduced situationally. Free tiers make most of the curriculum impossible.
• WF participants on personal devices. WF systems are air-gapped; personal devices let staff use the full toolset.
• Apple Silicon Macs recommended; hardware question added to intake. The newest tooling performs best there.
• Recruit two operational doers per organization. Pairs double the odds that change sticks back home.
• RW Institute drafts the invite and a readiness self-assessment quiz. Recruitment needs qualifying language and a self-selection tool.
• Session summaries within 48 hours. Standard delivery model.

Left open that day

• Concurrent versus staggered cohorts, final launch date, selection criteria, and whether WF grant materials become shared curriculum content.

All decisions from this session

• Application simplified to a direct apply-now path. Avoids the multi-step back-and-forth. Later replaced by nomination.
• One unified application per pair, not two forms. The organization self-identifies both people at once.
• Two participants per organization required for this run. Breakout homework runs in pairs; two voices are more compelling back home.
• Cohort of about 10 organizations plus 5 WF staff seats, overflow to a waitlist. The waitlist builds the Phase 2 case. Staff seats later moved out of the cohorts.
• No real WF data; practice data or participants' own non-WF data only. A clean, hard line, stated plainly in the room.
• Codex kept out of core curriculum, demo only in Lab 6. Useful context on cost and resilience, secondary to core content.
• Same content twice weekly, Monday and Friday, pairs commit to one track. Mixed days would split the pairs the breakouts depend on.
• Friday session shifted to 10:00 to 11:30 Central. 9:00 Central is very early for Pacific participants.
  Later changed: the confirmed times are Cohort A Mondays 12:00 to 1:30 PM Central and Cohort B Fridays 11:00 AM to 12:30 PM Central.
• Recruitment targeted to the first two weeks of August. Clears WF's late-September internal commitments and gives leads time.
• The entity is called PCI, Philanthropy and Community Impact. Official name, used in all materials since.
• Recruitment comms split into two audiences: nonprofits and internal PCI. Different messages for different populations.

All decisions from this session

• Recruitment split: nonprofit track through the public site, PCI track handled internally. Different processes for different populations; internal buy-in runs through the Foundation's chief administrative officer.
• The role question and all PCI references stripped from the application. Redundant once the tracks separated.
• Cohort sizing confirmed: 10 orgs times 2 people plus 5 PCI staff, 25 per cohort, 50 total. Verified in the room.
• Attendance policy: miss no more than 2 of 7 sessions, stated in the application. Competitive program; a clear bar helps applicants self-select. The room pushed to make it firm.
• Day choice locked per participant. Breakouts need the same recurring pairs.
• The two colleagues in a pair must come from the same organization, and attend the same day. Same tools, same systems, real mutual support.
• Recruitment timing: early August confirmed over late July. Lead time for PCI colleagues, buffer for internal approvals.
• Program start shifts past Labor Day, September 7. The week before is a dead zone; people return with a back-to-business mindset.
• Individual Claude accounts, not a team instance. Team minimums are inefficient at this scale, and participants keep their account and work after the program. Covered within the existing grant budget.
• Briefing package ahead of the Molly Porter meeting. That meeting produced both pivots we absorbed on July 24.

Worth knowing

• The specific calendar dates in the working plan came from the timeline document after this call. In the room, the agreements were early August recruitment and a post-Labor-Day start.

All decisions from this session

• PCI employees out of the funded nonprofit cohorts. The program is funded as a foundation grant, which requires strict separation between nonprofit partners and internal staff.
• Nomination replaces open application. A curated, relationship-led pipeline over mass sign-up. The guidance was to skew restrictive rather than open.
• The no-rejection flow: nominate, score and shortlist, invite to apply, place or hold. Only the nominating relationship manager ever knows the round-one outcome. The nonprofit never sees a rejection.
• Cap of two nominations per PCI member. Around 107 potential nominators; the cap keeps the pipeline strong without flooding it.
• Outreach through individual PCI working-group members, not one broadcast email. More effective and more selective.
• Weighted, scored rubric behind the nomination form. Hard disqualifiers became scored qualifying questions; ranking happens on the RW Institute side.
• Separate PCI-only upskilling track approved: about seven sessions, 90 minutes plus homework, no cost, run outside the grant. Real competency for PCI staff without touching the funded program's scope or budget.
• PCI working-group members may audit recorded cohort sessions. Observation access, cleanly separated from participation.
• Recordings and homework will live in a new participant-facing hub; PCI's SharePoint links out to it. One canonical home, no email distribution.
• Timeline reviewed live: at worst a week of slip, and we can make it up. The launch holds.

What landed across Aug 5 and the Aug 17 leadership review

• Nominations feed a shared, side-by-side review dashboard for the working group. Replaces email submissions; each nomination carries its readiness score and the nominator's rationale.
• Nomination tool tested end to end by PCI, including a live test nomination in the dashboard. Two-button confusion fixed to a single Submit; the review portal address corrected.
• Everything consolidated onto one domain, wf-ai-labs.rw.institute. One host for WF IT to allow, retiring the old vercel and netlify links.
• Timeline confirmed by PCI Leadership: nominations Sep 4, selections and invitations Sep 11, applications Sep 18, cohorts confirmed Sep 24, program starts the week of Sep 28, concludes Nov 16. Leadership had questions, so the initial compressed dates moved out to September.
• PCI staff participation set aside for this round. Leadership chose to emphasize internal AI training; the funded cohorts stay nonprofit-only.
• Scoring rubric extracted to a standalone document for PCI. One qualifier plus five readiness questions, scored 0 to 10, bands internal only.
• Rolling review approved. Nominations can be reviewed and invited as they arrive after Sep 4, rather than waiting for a single gate.
• Storytelling strategy requested by leadership. Sessions are recorded and the work is online, so stories can be mined and shared as the program runs. Approach to be planned with PCI.

All decisions from this session

• The plan was presented to foundation leadership on Aug 17; the nomination call moves the moment the green light lands. The hold point was named plainly so nobody assumed the call was already out.
• Every invited organization gets an off-ramp: take the seat or say it is not the right time, through the same link, on a short response window. A declined seat has to be visible fast so a new invitation can go out.
• The confirmation step collects background on the organization's current tools and AI use. The working group sees the real starting point, and mismatches surface before Lab 1 rather than in it.
• Invitations issued by the nominating PCI member from a shortlist. Relationship-led, matching the nomination model. Revised Sep 2: PCI sends directly with the nominator copied.
• RW Institute sends the calendar series for both cohorts to the PCI leads; how far the invites travel is PCI's call. The dates were held on RW calendars only, and PCI decides nominator visibility.
• Facilitator backup and a contingency behind it for the seven-week run. The dates are rigid, so delivery cannot depend on one person.

All decisions from this session

• The nomination call is out with PCI members, with a reminder to follow. A reminder makes sure every PCI member knows the call is live and open.
• First rolling review on Wednesday, Sep 9. PCI pulls the nominations received from the review dashboard on Tuesday afternoon or first thing Wednesday; its internal Wednesday check-in reviews fit and clears the first invitations. Review, acceptance, and any pass-along take time, and starting before the close spreads that work out.
• Invitations go in reviewed batches, strongest fits first, with 5 to 10 seats held until after Sep 14. Some nominators submit at the last second, and the reserve keeps room for that rush. Fit review weighs relationship history, not the score alone.
• PCI sends the invitation directly to the organization with the nominator copied, naming who put them forward. The fastest path, and it honors the nominating relationship: a nominator who sees the invite go out sees their intention carried through.
• The invitation is an opportunity, not an application: sign up if it fits, decline if not, inside a short window. Deferential language and an easy no keep seats moving to organizations that can use them.
• The Sep 16 biweekly runs the next batch of reviews and invitations. The standing check-in becomes the second review point.
• A communications plan joins the program. No open call, so public promotion starts once the cohort is set and features its members; PCI's communications partners get near-ready material to amplify. Leadership asked for it, and little else like this exists in the ecosystem to point to.
• RW Institute updates the invitation template and the seat-confirmation page to the agreed route and adds a comms section to the hub. The templates in the hub predate this call.

All decisions from this session

• The Lab 1 run-of-show drops the Copilot walkthrough as its opening frame; Copilot and data anonymization are still covered, later in the sequence. Keeps the curriculum's opening on strategic capability, the philosophy set on June 11, rather than one tool's workflow.
• November 16 is the program wrap: Lab 7 has cohorts demo a completed workflow document in breakouts, no homework that day. A separate showcase event for Wells Fargo leadership and the original nominators will be added beyond Lab 7; its length, exact date, and placement around Thanksgiving are still to be worked out. Gives both cohorts a shared ending and gives leadership and nominators a way to see results.
• Nominator communications now include a standard thank-you email to whoever nominated an organization, alongside the invitation to the nominated organization itself. Closes the loop with the people who made the nomination possible, not only the organization invited.
• RW Institute confirmed it sends the cohort calendar series and account invitations to confirmed pairs, consistent with D24; the placeholder invite language is out of date and will be rewritten first.
  Later changed Sep 25: each participant sets up and owns a Claude Pro account, covered by Wells Fargo for six months, so no account invitations go out (D37).

Left open that day

• The partner-organization target per cohort: Wells Fargo to come back with a number.
  Settled Sep 25: 25 nonprofits across the two cohorts (D38).
• The October 12 holiday session: whether Cohort A's Lab 3 moves to Tuesday, October 13, or the Monday cohort is asked directly on their first day. Wells Fargo has not given a final answer.
  Settled Sep 27: the Monday cohort's Lab 3 stays on Monday, October 12, with no poll (D42).
• A date for the storytelling and comms working session.
• The showcase event's length, exact date, and placement around Thanksgiving.