BNEDai · AI Agent Build Sessions Impact Report No. 1 · June 29 to September 30, 2026

AI Agent Build Impact Report

Training women leaders to build their first AI agents in one hour

Prepared by Jacqueline V. Twillie, MBA · bnedai.com

1. What this program is and why it exists

This report covers the first three months of a live AI agent training program built for senior professional women. The curriculum moves participants from passive AI users to managers of a digital workforce: "AI agents are your employees; you are the manager." On Gumloop's no-code platform, participants build, test, and schedule their first working AI agents inside a single structured one-hour workshop, with scoped connectors, human-in-the-loop review, and a schedule set only after live testing. The program closes the AI fluency gap for women while the AI agents go to work on real calendars, inboxes, and searches. It has run without funding or sponsorship to date, taught entirely by one instructor.

This program grows out of a decade teaching women to negotiate and close the gender wage gap. Skills gaps compound the same way pay gaps do: fastest for the people already locked out of the room. This curriculum exists to close the AI skills gap before it hardens into another one.

300+
People taught
Taught across 29 live sessions on Zoom, Google Meet, and in person. Count from the instructor's session records.
95%
Women registrants
Senior managers, PhDs, MBAs, and enterprise decision-makers. From the July registrant roster.
4.95/5
Instruction clarity
Post-session participant surveys on workshop pedagogy.
90.5%
Post-build confidence
Respondents reporting moderate or extreme confidence to build their next AI agent on their own.

2. Who came: the leadership network

Senior professionals from leading global organizations, academic institutions, public agencies, and clinical networks. Participants registered as individuals; no employer sponsored or endorsed the sessions.

Represented organizations
Amazon Web Servicescloud & AI technology
ExxonMobilenergy sector operations
Expedia Groupglobal travel & consumer tech
Moody's Analyticsfinancial risk & analytics
Sony Pictures Entertainmentmedia & distribution
Hallmarkglobal consumer brands
City of Houstonpublic administration & health
NYC Dept. of Educationpublic education
ETSeducational research
Foundation for the NIHmedical philanthropy
Academic and license profile
Ph.D. and Ed.D.research & education leadership
MBAmanagement
MPApublic policy
M.Phil, Oxfordadvising & instruction
M.Engengineering
PMPprogram & operations leadership
DNP, MSN-Ed, RNclinical leadership
M.Acccorporate finance
FINRA Series 7, 63, 24licensed financial operations

3. How word spread: social reach by the numbers

  1. Threads recap, June 30: 15,200 views, 344 replies, 339 likes, 4 reposts. Replies nearly matched likes: people asked to be included rather than scrolling past, and the next two sessions filled.
  2. LinkedIn campaign No. 1, June 29: 1,859 impressions, 1,257 members reached, 19 comments. Viewer profile per LinkedIn analytics: 30% senior, 13% director, 9% C-suite, with founders, CEOs, and professors among the top titles.
  3. LinkedIn campaign No. 2, July 8: 1,396 impressions, 966 members reached, in an audience weighted toward founders, engineers, and senior operators in major tech markets.
  4. Referral traffic: referral links from the June 29 campaign drew 57 visitors; most activation happens live in session, outside link attribution.
What they built Page 2

4. What did participants build?

Working AI agents on their own workflows: meeting prep that reads a real calendar and inbox, a scheduled executive-search scout, and a voice skill that makes drafts sound like their owner.

A working AI agent is the visible outcome. The real return is time: what a manager gets back once the AI agent runs the routine work, and what she does with it instead.

Live sessions teach token discipline, scoped connector access, human-in-the-loop triggers, and how to counter AI sycophancy. Participant builds on record include:

Danyell WellsMulti-business meeting prep · July 20
Runs several companies and chose a separate meeting-prep AI agent for each, every one reading its own calendar and inbox. "They're vastly different. And so I have to keep them separate."
Venicia Figueroa, MPAFirst AI agent plus a master voice skill · June 29
Built her first AI agent in the June 29 session and posted about it the same week. Her tone question became the session's voice lesson: a portable markdown voice file, refreshed monthly, that makes an AI agent's drafts sound like their owner from the first run.
Nicole D.Scheduled search Scout · July 22
A Scout AI agent on Gemini 3.1 Pro targeting consumer-goods executive search firms within a 20-mile Detroit radius, scheduled for Tuesday and Friday at 6am only after individual testing, with a one-week refinement check-in.
A private company, Atlanta metroEnterprise access findings · June 29
Brought a team to build and surfaced the enterprise adoption barrier live: managed-device restrictions blocked the build, and the team left with a WordPress integration workaround plan. Every session now opens with a device check.

5. How long do people stay in a one-hour session?

Across 120 Zoom attendances since mid-July, the median builder stayed 61 minutes, and 85% stayed at least 50 minutes of the hour. Builders have joined from New York, London, Salvador (Brazil), Detroit, Memphis, and Georgia, across 29 live sessions in three months.

The curriculum is built to route around the traps that stall non-technical builders. AI sycophancy, an AI agent confidently confirming work it never did, burns through platform credits and trust in the same afternoon. Naming it early keeps a session on schedule. The guardrails hold across the room: a college-of-nursing professor, a biomedical scientist, attorneys at Warner Bros. and PwC, a Series 7 financial operator, an internal auditor, a voice-over artist, and a salon owner build alongside people from Amazon Web Services, the City of Houston, and Sony Pictures Entertainment, all through the same hour at the same pace.

  1. Retention: across 120 Zoom attendances since mid-July, the median builder stayed 61 minutes, and 85% stayed at least 50 minutes of the hour.
  2. Ratings and repeats: of 20 participant ratings in Luma, 18 are five stars. 39 people have registered for two or more sessions.

Participant feedback and testimonials

"Wow... what a powerful session! Thank you for taking the time [to] teach folks step-by-step on how to build an AI agent." · Kristine Frost, chief human resources officer and board director (5/5, Luma review)

"She made it so easy for me to build my first Agent, which is now working every day to make my life easier." · Alexandria Sarmiento (5/5, Luma review)

"Super great and not as scary as I thought it was going to be." · Susan Quinn, August 10 session

"I walked away with a fully functioning AI agent built for my business." · Berthine Crèvecoeur West, MA, EMBA, CDE®, public LinkedIn post

6. What happens after the hour? The teach-one commitment

100% of surveyed participants made the teach-one pledge: each commits to teach one other person what she learned about building AI agents, a network effect that spreads AI fluency beyond any single room. Follow-through is the measure Report No. 2 will track.

The pipeline already runs deeper than the pledge: Tina Getachew, PMP, came to the June 29 session, joined a Gumloop Learning Cohort, completed the Certified Agent Builder program in July, and shared her certification publicly. Christina Hawkins posted her one-hour build to her network: "If you're AI-curious but not sure where to start, I'd recommend keeping an eye on Jacqueline's posts for future workshops."

Questions this report answers

How many people has BNEDai taught to build AI agents? More than 300 people across 29 live sessions on Zoom, Google Meet, and in person between June 29 and September 30, 2026, counted from the instructor's session records. 95% of registrants on the July roster were women, including senior managers, PhDs, MBAs, and enterprise decision-makers.

What do participants build in one hour? A working AI agent on their own workflow, built on Gumloop's no-code platform: meeting prep that reads a real calendar and inbox, a scheduled executive-search scout, or a voice skill. Each is tested live before any schedule is set, with scoped connectors and human-in-the-loop review.

How confident are participants after the session? 90.5% of surveyed participants reported moderate or extreme confidence to build their next AI agent on their own, and instruction clarity rated 4.95 of 5 in post-session surveys. Of 20 ratings in Luma, 18 are five stars, and 39 people registered for two or more sessions.

Is the program funded or sponsored? It has run without funding or sponsorship to date, taught entirely by one instructor. Partners can support access for underrepresented women leaders, quarterly session series in more cities, and enterprise training modules. Contact Jacqueline V. Twillie to join as an impact partner.

Partnership opportunities

Support expands this program to more cities, quarterly session series, funded access for underrepresented women leaders, and enterprise training modules built around a team's own workflows. On September 21, the program ran a one-hour build in person at the Sistas in Sales Summit in New York, with 32 builders. Next on the calendar: two free build sessions a week through October.

Contact Jacqueline V. Twillie to join as an impact partner.