FutureStack 2026 — Post-Event Report
Generated by: Attendee Experience skill
Event: FutureStack 2026 Dates: October 15-16, 2026 Venue: Austin Convention Center, Austin, TX Final Attendance: 783 (of 800 registered) Theme: AI Infrastructure Report Date: October 24, 2026
Executive Summary
FutureStack 2026 delivered a strong attendee experience across both days, earning a Net Promoter Score of 72 with a 68% survey response rate (532 of 783 attendees). The AI Ambassador concierge system handled 1,247 messages with a sub-20-second average response time and only 23 escalations to human staff. The networking matchmaker generated 89 successful introductions from 156 opt-ins. This report details attendee sentiment, session performance, AI Ambassador analytics, networking outcomes, and recommended changes for FutureStack 2027.
1. Attendee Satisfaction
Net Promoter Score
| Metric | Value |
|---|---|
| NPS | 72 |
| Promoters (9-10) | 78% |
| Passives (7-8) | 16% |
| Detractors (0-6) | 6% |
| Survey responses | 532 of 783 attendees |
| Response rate | 68% |
Overall Satisfaction Ratings
| Question | Avg (out of 5) |
|---|---|
| Overall event experience | 4.6 |
| Session content quality | 4.5 |
| Venue and facilities | 4.3 |
| Food and catering | 4.1 |
| Networking opportunities | 4.4 |
| Value for ticket price | 4.5 |
| Likelihood to attend next year | 4.7 |
2. Top 3 Things Attendees Loved
Based on free-text analysis of 532 survey responses (themes ranked by mention frequency):
1. Keynote Quality and Speaker Depth (mentioned by 74% of respondents)
Dr. Amara Osei's opening keynote ("Building the Foundations That Don't Break") was the single most praised element of the conference. Attendees consistently highlighted the practical depth, real-world failure analysis, and the "no-slides-wasted" density of her presentation. The afternoon panel on build-versus-buy was the second most cited session.
Representative quote: "Dr. Osei's keynote alone was worth the ticket. I've never seen someone explain distributed training fault tolerance with that level of clarity and honesty about what actually goes wrong."
2. Practitioner-First Session Design (mentioned by 61% of respondents)
Attendees appreciated that sessions were designed for practitioners, not executives. The consistent feedback was that speakers shared real architectures, real code, and real failure modes rather than high-level overviews or product pitches. The 45-minute session format with 15-minute buffers was praised for allowing depth without audience fatigue.
Representative quote: "Every session I attended felt like it was written for someone who actually builds this stuff. No fluff, no vendor pitches disguised as talks."
3. AI Ambassador Concierge (mentioned by 48% of respondents)
The SMS/WhatsApp concierge was a standout feature, with attendees calling it "the best conference tool I've ever used." Speed of response, accuracy of answers, and the seamless escalation to human staff when needed were all highlighted. Multiple attendees noted that they did not need to download an app or search for information -- they just texted.
Representative quote: "I texted asking where the gluten-free lunch was and got a perfect answer in 12 seconds. Then I asked about a speaker's background and got that too. I wish every event had this."
3. Top 3 Complaints
1. Afternoon Breakout Room Temperature (mentioned by 34% of respondents)
Rooms 12A and 12B were consistently reported as too cold during afternoon sessions on both days. Multiple attendees mentioned leaving sessions to warm up or bringing jackets specifically for the breakout rooms. The Lovelace Room (Room 16) did not receive the same complaints.
Root cause: ACC HVAC zoning for the 1st floor east wing (Rooms 12A/12B) is on a single thermostat that responds slowly to occupancy changes. Pre-event coordination with ACC facilities was insufficient on this point.
Recommended fix for 2027: Request independent HVAC control for each breakout room. Schedule a walk-through with ACC facilities 48 hours before the event to set and test room temperatures with simulated occupancy load.
2. Lunch Line Congestion on Day 1 (mentioned by 28% of respondents)
The single-entry flow into Hall 2 for lunch created a bottleneck on Day 1, with some attendees reporting 15-20 minute waits. Day 2 was improved after the catering team opened a second entry point, but the damage to Day 1 perception was done.
Root cause: Hall 2 had one primary entrance configured for lunch service. The secondary entrance was blocked by a catering staging table.
Recommended fix for 2027: Pre-configure two entry points from the start. Consider staggered lunch release by session track (5-minute offsets) to distribute the surge.
3. Limited Power Outlets in Networking Lounge (mentioned by 19% of respondents)
Room 10 (Networking Lounge) was popular throughout both days, but attendees reported difficulty finding available power outlets and charging spots. The dedicated charging stations in the foyer and expo hall were sufficient, but attendees wanted to charge while networking rather than at a separate station.
Recommended fix for 2027: Add 2-3 portable power tower units in Room 10. Ensure every table in the networking lounge has a floor-accessible power strip.
4. Session Ratings
Ratings collected via post-session ActionNotes surveys (1-5 scale, sent within 2 hours of each session). Response rates varied by session (42-71%).
Top 5 Sessions
| Rank | Session | Speaker | Track | Avg Rating | Responses |
|---|---|---|---|---|---|
| 1 | Building the Foundations That Don't Break (Keynote) | Dr. Amara Osei, DeepMind | Mainstage | 4.9 | 487 |
| 2 | GPU Cluster Scheduling: Lessons from 10,000 Nodes | James Okonkwo, NVIDIA | Track C | 4.8 | 89 |
| 3 | The Build vs. Buy Decision for AI Infra (Panel) | Chen, Patel, Okonkwo, Sharma + moderator | Mainstage | 4.7 | 341 |
| 4 | Inference at the Edge: Latency Under 10ms | Raj Mehta, Cloudflare | Track A | 4.7 | 82 |
| 5 | Kubernetes for ML Workloads: What Actually Works | Lin Zhang, Anyscale | Track A | 4.6 | 76 |
Bottom 3 Sessions
| Rank | Session | Speaker | Track | Avg Rating | Responses | Issue |
|---|---|---|---|---|---|---|
| 28 | Securing the AI Pipeline: Threat Models for Infra Teams | Alex Durand, CrowdStrike | Track C | 3.4 | 51 | Feedback: too high-level, felt like a product overview rather than a practitioner talk |
| 29 | Cost Optimization for Cloud GPU Fleets | Maria Gonzalez, Doit International | Track B | 3.2 | 44 | Feedback: outdated pricing data, did not account for recent GPU marketplace changes |
| 30 | The True Cost of Training: A Financial Model | Chris Park, Independent | Track C | 3.0 | 38 | Feedback: spreadsheet-heavy with limited actionable takeaways; presenter ran 12 minutes over time |
Session Rating Distribution
| Rating Range | % of Sessions |
|---|---|
| 4.5 - 5.0 | 27% (8 sessions) |
| 4.0 - 4.4 | 47% (14 sessions) |
| 3.5 - 3.9 | 17% (5 sessions) |
| Below 3.5 | 10% (3 sessions) |
5. AI Ambassador Analytics
Platform: AI Ambassador via ai-ambassador.xyz Channels: SMS and WhatsApp Active Period: October 15, 6:30 AM -- October 16, 10:00 PM
Summary Metrics
| Metric | Value |
|---|---|
| Total messages received | 1,247 |
| Unique users (distinct phone numbers) | 412 |
| Messages per unique user (avg) | 3.0 |
| Automated responses (no escalation) | 1,224 (98.2%) |
| Escalations to human staff | 23 (1.8%) |
| Average response time (automated) | 18 seconds |
| Average human response time (escalations) | 3 min 42 sec |
| Longest human response time | 8 min 15 sec |
| Messages outside active hours (not answered) | 14 |
| Opt-outs (STOP) | 2 |
Top 10 Questions by Volume
| Rank | Question Topic | Messages | % of Total |
|---|---|---|---|
| 1 | Wi-Fi password | 189 | 15.2% |
| 2 | Session schedule / "what's happening now" | 176 | 14.1% |
| 3 | Room locations / wayfinding | 143 | 11.5% |
| 4 | Food options / dietary info | 112 | 9.0% |
| 5 | Speaker bios / session details | 98 | 7.9% |
| 6 | Charging station locations | 87 | 7.0% |
| 7 | Parking validation | 72 | 5.8% |
| 8 | Evening reception details | 64 | 5.1% |
| 9 | Restroom locations | 51 | 4.1% |
| 10 | ActionNotes / recording access | 48 | 3.8% |
| Other (long tail) | 207 | 16.6% |
Escalation Breakdown
| Escalation Reason | Count | Avg Resolution Time |
|---|---|---|
| Complex schedule question (multi-session comparison) | 7 | 3 min |
| Code of Conduct concern | 3 | 2 min (immediate staff dispatch) |
| Lost item with specific description | 4 | 5 min |
| Catering allergy emergency (severity confirmation) | 2 | 1 min 30 sec |
| AV/technical issue in breakout room | 3 | 4 min |
| Question outside knowledge base | 4 | 4 min |
Channel Split
| Channel | Messages | % |
|---|---|---|
| SMS | 814 | 65.3% |
| 433 | 34.7% |
Peak Usage
| Time Window | Messages |
|---|---|
| Oct 15, 7:30 - 8:30 AM (arrival/registration) | 178 |
| Oct 15, 12:15 - 1:00 PM (lunch) | 134 |
| Oct 15, 6:00 - 7:00 PM (pre-reception) | 89 |
| Oct 16, 7:30 - 8:30 AM (Day 2 arrival) | 112 |
| Oct 16, 12:15 - 1:00 PM (lunch) | 96 |
6. Networking Matchmaker Results
Platform: AI-powered Networking Matchmaker (integrated with AI Ambassador)
| Metric | Value |
|---|---|
| Opt-ins | 156 (19.9% of attendees) |
| Matches generated | 112 |
| Introductions sent (via SMS/WhatsApp) | 89 |
| Introductions that led to confirmed meetings | 67 (75.3% conversion) |
| Introductions declined or no-response | 22 (24.7%) |
| Average match quality rating (1-5, post-intro survey) | 4.2 |
Opt-in Demographics
| Role Category | % of Opt-ins |
|---|---|
| Senior/Staff Engineer | 34% |
| Engineering Manager / Director | 28% |
| Founder / CTO | 18% |
| Researcher / Academic | 12% |
| Other | 8% |
Top 3 Requested Discussion Topics
- Inference optimization and serving infrastructure (41%)
- Building vs. buying ML platform components (33%)
- Team structure and hiring for AI infrastructure roles (26%)
Attendee Feedback on Matchmaker
"I got matched with an infra lead at a company solving the exact same caching problem I'm stuck on. We spent 45 minutes at the networking lounge and I left with three ideas to try Monday morning."
"The quality of the match was surprisingly good. I asked to talk about GPU scheduling and got paired with someone who literally wrote the scheduler at their company."
7. Recommended Changes for FutureStack 2027
Based on survey data, AI Ambassador analytics, session ratings, on-site observations, and staff debrief (held October 18, 2026):
1. Fix Breakout Room Climate Control
Problem: Rooms 12A and 12B were too cold during afternoon sessions both days, cited by 34% of respondents. Action: Negotiate independent HVAC zoning for each breakout room in the 2027 venue contract. Conduct a climate walk-through with venue facilities 48 hours pre-event with simulated occupancy (space heaters or HVAC adjustments). Include temperature monitoring on the venue ops checklist every 2 hours during the event.
2. Eliminate Lunch Congestion
Problem: Single-entry lunch service caused 15-20 minute waits on Day 1 (28% of respondents). Action: Design Hall 2 lunch service with two entry points from Day 1. Implement staggered lunch release: Track A at 12:15, Track B at 12:20, Track C at 12:25, expo hall and general at 12:30. Communicate staggered timing in the morning announcements and via AI Ambassador.
3. Expand Power Access in the Networking Lounge
Problem: Insufficient power outlets in Room 10 (19% of respondents). Action: Add 3 portable power tower units (8 outlets each) distributed across the lounge. Ensure every networking table has a floor-accessible 4-outlet power strip. Budget estimate: $450 rental.
4. Raise the Bar on Session Vetting
Problem: Bottom 3 sessions averaged 3.2/5.0, with complaints about product pitches and outdated content. Action: Implement a dry-run review process for all accepted sessions. Require speakers to submit final slides 2 weeks before the event. Assign a track chair to review each deck for practitioner-first depth, currency of data, and absence of product pitches. Reject or rework sessions that do not meet the bar. Add a "content freshness" criterion to the CFP rubric.
5. Increase Networking Matchmaker Adoption
Problem: Only 19.9% of attendees opted in, despite high satisfaction among those who did (4.2/5 match quality). Action: Promote matchmaker opt-in more aggressively: include in pre-event emails, mention in opening remarks, add signage in the registration area and networking lounge, and have AI Ambassador proactively suggest opt-in during the first text exchange. Target: 35% opt-in rate for 2027.
Appendix: Survey Methodology
- Survey tool: Post-event email survey (SurveyMonkey), sent October 17, 2026 at 10:00 AM CT
- Reminder: October 19, 2026 at 2:00 PM CT
- Close date: October 22, 2026 at 11:59 PM CT
- Response rate: 532 / 783 = 68%
- NPS calculation: Standard NPS methodology (% Promoters minus % Detractors)
- Session ratings: Collected via ActionNotes post-session micro-surveys (3 questions, delivered within 2 hours of session end)
- AI Ambassador analytics: Extracted from ai-ambassador.xyz admin dashboard, October 23, 2026
- Confidence note: With 532 responses from 783 attendees, results carry a margin of error of +/- 2.1% at 95% confidence
Last updated: October 24, 2026 Attendee Experience skill -- FutureStack 2026