← Home
Partner Basecamp · Cohort 12

San Francisco

May 28-29, 2026
+50
Net Promoter Score
67
Day 1 Responses
80%
Matched Pairs
75%
Day 2 Response Rate
Programme Satisfaction
Strong promoter majority — NPS +50 reflects broad programme endorsement.
+50NPS · n=50
58% Promoters34% Passives8% Detractors
95% CI: +32 → +68  ·  True NPS lies within this range with 95% confidence (n=50 respondents)
Confidence Arc
Strong end-of-Day-1 build confidence — cohort leaves Day 1 ready to apply Claude in client work (mean 4.2/5).
4.2end-of-D1 build / 5
4.2D2 Design / 5
4.1D2 Commercial / 5
4.0D2 Build / 5
Audience
67 participants across 6 organisations — Developer majority with applying in practice the most common AI experience level.
Top Organisations
Deloitte (32) KPMG (28) EPAM (3) Casper Studios (2) Bounteous (1)
Voice of Participant
What participants said — dominant theme from each NPS segment.
Promoters
Participants valued hands-on, real-world exploration of Claude's latest capabilities and ecosystem applications for professional consulting.
Passives
Attendees appreciated foundational content but wanted deeper technical dives, differentiated tracks, or Anthropic engineer insights for advanced learners.
Detractors
Technical setup friction and misaligned audience expectations undermined learning; material felt derivative and better suited for beginners than experienced practitioners.
Facilitator Notes
Room Composition
The KPMG group are hungry for deep technical training on complex agent deployment at scale. They are the first at KPMG so we see that they are wanting to show up well in front of their contingent and are asking highly technical and complicated questions in front of the group (most of the time when we unpack it's a simpler ask). We're leaning in! Planning to give them a role where possible to share their expertise. We'll also share with them the path to CCAF and the roadmap Anthropic shared about additional certifications and resources that will be made available to partners (they weren't aware so that's a good value add). Any tips you have for us on positioning foundational content for a highly advanced audience so they see the value rather than feeling held back by it? I'll also position them as force multipliers for their organizations. Even if they're beyond the foundation level themselves, understanding these concepts helps them better support, coach, and guide the teams that will be building from them. Most importantly, I want to make sure they feel heard and that we're acknowledging the expertise they're bringing into the room. We're doing some stakeholder mgmt with their point person who is lovely, too!
For Next Time
None
KPMG — [{'source': 'Sammy Rogers (SF Basecamp facilitator)'}, "we had a full table of engineers heads-down on other work for most of D1 and all of D2, including the entire 30-minute applied AI Q&A while the rest of the room was fully engaged. I was standing right behind them and they weren't even trying to be subtle.", "One participant at that table asked a question about LangChain, then immediately went back to his LangChain Jupyter buildalong in VS Code lol. A read could be that the table had formed the perspective that the content wasn't valuable and opted out entirely. Understandable for engineers who don't need a Jupyter notebook walkthrough, but my concern is they may have skipped the foundational building blocks that underpin the more advanced work.", "Worth monitoring in future cohorts, though their on-site contact said they'd be more intentional about who they select going forward given this was last minute."]
Survey & Data Quality
Survey & Data Quality
Day 1 responses67
Day 2 responses50 (75% of Day 1)
Matched pairs40 (80% of Day 2)
Orgs resolved67 of 67 respondents matched to named org
Day 2-only respondents10 (20.0%) — suppresses NPS by 2 points; no Day 1 data, excluded from persona & confidence analysis
Organisations
All Day 1 respondents · n=67
Deloitte32 (48%)
KPMG28 (42%)
EPAM3 (4%)
Casper Studios2 (3%)
Bounteous1 (1%)
Pearson1 (1%)
Function × Seniority
All Day 1 respondents · n=67
FunctionPractitionerSr PractitionerMgr / Sr MgrDirector+Partner / MDTotal
Engineering16842131
Architecture2262214
Business Leadership212117
Project / Engmt335415
Experience Profile
AI experience level · n=67 · % of row
By Persona
SegmentExploringApplyingDeliveringFrontiern
Developer19%32%26%23%31
Transformation Lead36%27%23%14%22
Architect21%43%14%21%14
By Seniority
SegmentExploringApplyingDeliveringFrontiern
Practitioner (0-4 years in role)22%43%22%13%23
Manager or Senior Manager47%29%12%12%17
Senior practitioner (5–9 years)21%29%21%29%14
Director, Senior Director, or Principal11%22%33%33%9
Partner, Managing Director, or Executive25%50%25%4
How would you describe your current experience with AI tools prior to Basecamp?
Learning and exploring
17 (25%)
Applying in practice
22 (33%)
Delivering independently
15 (22%)
Operating at the frontier
13 (19%)
Prior to today, how much had you worked with Claude or the Anthropic API?
Not at all
4 (6%)
A little
32 (48%)
Regularly
31 (46%)
Did the depth land for this audience?
Technical depth perception · n=67 respondents · % of row
By Persona
SegmentToo basicAbout rightToo advancedn
Developer39%61%31
Transformation Lead9%82%9%22
Architect14%86%14
By AI Experience Level
SegmentToo basicAbout rightToo advancedn
Applying in practice9%91%22
Learning and exploring88%12%17
Delivering independently60%40%15
Operating at the frontier38%62%13
Overall depth distribution
Too basic
16 (24%)
About right
49 (73%)
Too advanced
2 (3%)
Did the pace work across the room?
Session pace perception · n=67 respondents · % of row
By Persona
SegmentToo slowWell pacedToo fastn
Developer13%77%10%31
Transformation Lead5%68%27%22
Architect7%86%7%14
By AI Experience Level
SegmentToo slowWell pacedToo fastn
Applying in practice9%73%18%22
Learning and exploring71%29%17
Delivering independently13%80%7%15
Operating at the frontier15%85%13
Overall pace distribution
Too slow — could have covered more
6 (9%)
Well paced
51 (76%)
Too fast — not enough time to apply
10 (15%)
How confident were participants to build with Claude after Day 1?
End-of-Day-1 confidence · "How confident are you in your ability to build a client solution using Claude?" · n=67 · 1–5
Mean end-of-D1 build confidence by Persona
Developer4.26/5 · n=31
Architect4.21/5 · n=14
Transformation Lead3.95/5 · n=22
Programme mean: 3.9/5 · 15 cohorts
Mean end-of-D1 build confidence by AI Experience Level
Operating at the frontier4.62/5 · n=13
Delivering independently4.40/5 · n=15
Applying in practice4.00/5 · n=22
Learning and exploring3.76/5 · n=17
Overall end-of-D1 build confidence distribution
1
1 (1%)
2
2 (3%)
3
14 (21%)
4
19 (28%)
5
31 (46%)
How relevant was today's content to your current role?
Content relevance rating · n=67 respondents · 1 = not relevant, 5 = highly relevant
Mean relevance by Persona
Architect4.57/5 · n=14
Developer4.19/5 · n=31
Transformation Lead4.05/5 · n=22
Mean relevance by AI Experience Level
Learning and exploring4.41/5 · n=17
Operating at the frontier4.31/5 · n=13
Applying in practice4.27/5 · n=22
Delivering independently3.87/5 · n=15
Overall relevance distribution
1
0 (0%)
2
4 (6%)
3
7 (10%)
4
26 (39%)
5
30 (45%)
How likely are you to recommend attending this programme to a colleague?
n=50 Day 2 respondents · 75% of Day 1
Score distribution · 0 = not at all likely · 10 = extremely likely
Cohort012345678910
Cohort 1211298920
Detractors 0–6 Passives 7–8 Promoters 9–10
NPS Breakdown
58%
34%
8%
NPS +50 (n=50)
Promoters (9–10)Passives (7–8)Detractors (0–6)
Programme mean: +41 · 15 cohorts
Promoters mean: 51%
Passives mean: 39%
Detractors mean: 10%
NPS by Segment
Organisations, personas, and experience levels · ≥4 respondents shown
By Organisation
Promoters (9–10)Passives (7–8)Detractors (0–6)
Deloitte
83%
17%
0%
NPS +83 (n=18)
KPMG
33%
50%
17%
NPS +17 (n=18)
By Persona
Promoters (9–10)Passives (7–8)Detractors (0–6)
Architect
67%
33%
0%
NPS +67 (n=9)
Developer
50%
44%
6%
NPS +44 (n=16)
Transformation Lead
60%
20%
20%
NPS +40 (n=15)
Programme means · 15 cohorts: Architect +50 · Developer +47 · Transformation Lead +36
By AI Experience Level
Promoters (9–10)Passives (7–8)Detractors (0–6)
Learning and exploring
85%
15%
0%
NPS +85 (n=13)
Operating at the frontier
50%
33%
17%
NPS +33 (n=6)
Applying in practice
43%
43%
14%
NPS +29 (n=14)
Delivering independently
43%
43%
14%
NPS +29 (n=7)
Programme means · 15 cohorts: Applying in practice +45 · Delivering independently +43 · Learning and exploring +40 · Operating at the frontier +38
What is the main reason for your score?
n=43 responses · organised by NPS segment
Promoters (score 9–10)· 24 responses
Participants valued hands-on, real-world exploration of Claude's latest capabilities and ecosystem applications for professional consulting.
9“I learned a lot about sophisticated designs of evaluating agents, models, and prompts. It’s been a great experience for also solidifying what I already believed, and supporting it with stronger skills”
Passives (score 7–8)· 15 responses
Attendees appreciated foundational content but wanted deeper technical dives, differentiated tracks, or Anthropic engineer insights for advanced learners.
8“The course is an excellent all-round introduction to essential concepts and a little strategy. I feel I could promptly summarize learning points to different personas, saving them the trip. After that, the only thing that'd give them a reason to attend would be if you offered differentiated tracks.”
Detractors (score 0–6)· 4 responses
Technical setup friction and misaligned audience expectations undermined learning; material felt derivative and better suited for beginners than experienced practitioners.
5“I felt like there was good information in this course, but as someone with some experience in Claude it felt a lot like just copy and pasting code. There were bits and pieces I thought were insightful, but I feel like the program could’ve been compressed. The audience level was also not exactly understood for me. There should have been expectation setting.”
End-of-Programme Confidence
n=50 respondents · 1 = not confident, 5 = very confident
How confident are you in your ability to design an evaluation for an AI solution?
12345Mean
4%20%32%44%4.16
Programme mean: 4.1/5 · 15 cohorts
Leaving today, how comfortable are you advising clients on AI when questions arise?
12345Mean
2%2%20%32%44%4.14
Programme mean: 4.2/5 · 15 cohorts
How confident are you in your ability to lead a conversation about Anthropic and Claude with a client?
12345Mean
2%2%20%42%34%4.04
Programme mean: 4.2/5 · 15 cohorts
Confidence Trajectory by Persona
D1→D2 delta · apply-AI confidence · matched pairs only · n=40
Positive delta = confidence grew · negative = dropped
Architect (n=9)
+0.22
Developer (n=16)
+0.00
Transformation Lead (n=15)
-0.13
What one takeaway will you share with a colleague or client?
n=28 responses
Participants value understanding model routing trade-offs to optimize cost-effectiveness across deployment contexts.
“The routing of models — you don't always need opus, sometimes the cheaper models are actually better”
Consultants seek to advance beyond prompt engineering toward systematic context engineering for building production agents.
“How to advance skills and iterative feedback into not just better prompt engineering, but better context engineering”
Participants recognize Claude's power but identify operational governance gaps requiring attention before enterprise adoption.
“The tool is powerful. There are some operational / governance gaps that Anthropic itself is still working on”
Participants gain vocabulary and confidence to guide engineers toward responsible AI integration at different system maturity levels.
“You've given me the essential vocabulary of this new field, so I'll be speaking more competently with my stakeholders.”
Participants recognize evaluation-driven development from project inception as critical practice for AI systems.
“Evals from the very beginning of the project is the right approach”
What will you build for yourself or apply at work in the next 30 days?
n=31 responses
Participants want to build and control custom agents, particularly by dynamically curating context to guide agents toward better answers.
“We're very interested in dynamic curation of context, better-guiding agents to the right answers. I and a team are prototyping & reviewing strategies in that area.”
Participants seek structured evaluation frameworks they can immediately apply to prototype and test AI systems in their existing client projects.
“Prototypes to solve development and testing work in existing projects”
Participants value hands-on experience with Claude and recognize Anthropic must operationalize enterprise tools to meet professional services demands.
“The takeaway is that Anthropic has significant work to operationalize tools for enterprises.”
Participants need practical frameworks and skill documentation to apply AI best practices consistently across their architecture and delivery work.
“An ai architect skills folder with everything we've learned over the past 2 days to automatically apply best practices.”
Participants value concrete, reusable artifacts from training that embed learned skills into tools they can deploy in real engagements.
“An ai architect skills folder with everything we've learned over the past 2 days to automatically apply best practices.”
NPS Reason
n=43 responses · grouped by NPS segment
Promotersscore 9–10 · 24 responses
1Informative, educational and hands-on10
2Great team and real world discussions10
3Overall materials10
4Very insightful to learn about the latest models and how their capabilities might differ beyond basic pricing and versions10
5Love the tool people and food10
6Incredible introduction to Claude code and the Claude ecosystem that showed me how it’s changing the world.10
7It was worth the effort to learn and grow10
8Very informative and applicable to daily work10
9This was a great blend of fun learning with both hands on and just written instructions10
10Because I actually learned to build something10
11There was a lot of material we can reference and it had a lot of practical examples to apply this to any client work we’re working with.10
12Much more confident in use of Claude code now than before10
13Heavy content and hands on keyboard learning10
14Tons of good stuff10
15Great hands on exercises and use cases10
16Learned things I wasn’t aware about like prompt caching, & context engineering10
17High value education for the time!10
18Good content and presentation9
19Was really informative9
20I learned a lot about sophisticated designs of evaluating agents, models, and prompts. It’s been a great experience for also solidifying what I already believed, and supporting it with stronger skills9
21Learn how to accelerate productivity and code agents9
22Very fun and informative overall!9
23The program met me where I was and elevated me to a new level while doing so for all participants regardless of tech level.9
24challenging and useful scenarios, i’m a bit slow and would like more time but there are only so many hours in the day9
Passivesscore 7–8 · 15 responses
1The course is an excellent all-round introduction to essential concepts and a little strategy. I feel I could promptly summarize learning points to different personas, saving them the trip. After that, the only thing that'd give them a reason to attend would be if you offered differentiated tracks.8
2It was a great intro but I think there is a narrow audience that this was super helpful for.8
3I enjoyed being surrounded by other practitioners with similar experiences8
4Interactive and insightful.8
5Very helpful!8
6I think it's overall very helpful. I think the audience and content is a little too mixed. Would rather have a separate session dedicated to more advanced topics for FDEs.8
7Overall very good program8
8Very well thought out. I think maybe hearing more from Anthropic engineers on how they use Claude would be better7
9I am expecting to be guided to build an agent7
10How I felt ive learned from this7
11The training was good , however I feel we can reduce few. Exercise and go in details for the ones we do7
12I found it useful, not in depth, but a wide breath of knowledge7
13First day was very introductory. Second day opened more deep dives7
14Some of the concepts were covered through the anthropoid learning modules7
15My team is non technical so I would highly recommend all them take the architect program before coming! Overall I appreciate everyone who made this possible. I think more demos would be beneficial to everyone :)7
Detractorsscore 0–6 · 4 responses
1I felt like there was good information in this course, but as someone with some experience in Claude it felt a lot like just copy and pasting code. There were bits and pieces I thought were insightful, but I feel like the program could’ve been compressed. The audience level was also not exactly understood for me. There should have been expectation setting.5
2I would say that the program needs to be tubed for narrower audience.5
3More technical than business oriented. Had a lot of struggles with setting up environment than learning and brainstorming4
4The audience is helpful for folks with little exposure to AI. Most folks I work with have been keeping up with many agentic tools. Might be of more interest in a deep dive conference. The training material and notebooks seemed to be vibe coded from public documentation.2
Most Valuable
What one takeaway will you share with a colleague or client? · n=28 responses
1Get access to Claude at work or home
2You've given me the essential vocabulary of this new field, so I'll be speaking more competently with my stakeholders. I'll also be advising my prototyping engineers to move very quickly with maximal use of AI, provided they learn. For engineers closer to sensitive production-grade systems, I'll encourage good use of AI assistance, but to remain responsible for commits to the core machine, knowing they are accountable for tough L3 support scenarios.
3Ask Claude about Claude
4Agent context
5How to build an agent!
6Np
7This is going to change the world.
8Use plan mode
9The routing of models — you don’t always need opus, sometimes the cheaper models are actually better
10It’s better than you thought
11Share AI learning
12How to advance skills and iterative feedback into not just better prompt engineering, but better context engineering
13The model is not the bottleneck.
14The fact that you couldn’t use The fact that you can tweak props for different
15Metrics configuration
16The tool is powerful. There are some operational / governance gaps that Anthropic itself is still working on
17The concept of agent containerization and G Visor
18Context
19Ai can be safely adopted
20AI is accessible for all
21Context engineering
22Don’t get frustrated, at first things might not make sense or be difficult but you’ll get there.
23Start with small things and using in both work and personal life, and iterate!
24Evals from the very beginning of the project is the right approach
25Using the proper model can increase savings
26Diverse group in the workshop promotes ideas
27rating model use
28Anthropic doesn’t know their own roadmap
30-Day Intentions
What will you build for yourself or apply at work? · n=31 responses
1Explore agent automation and orchestration
2We're very interested in dynamic curation of context, better-guiding agents to the right answers. I and a team are prototyping & reviewing strategies in that area.
3Would like to build up a library of client material to use to build responses for clients
4Build my own custom agent
5I want to build a Project management agent for my client work
6A lot
7Something cyber related!
8skills and agent, automatic workflows
9Many structured evals
10Automate daily workflow
11RFP response evaluator
12Prototypes to solve development and testing work in existing projects
13I will go back through the capstones and generate similar projects but based on personal ideas and guidelines to push the boundaries, create documentation, and develop something I can use
14An ai architect skills folder with everything we’ve learned over the past 2 days to automatically apply best practices.
15Work related solutions that build on top of the platform
16Executive assistant based on calendars, text messages and emails
17-
18Give code some little projects for selling by to clients
19A complex orchestration agent based off a simple one I created recently
20My first alp
21A workout tracker application
22Anthropic certification
23I will definetely try to work more on evaluators, and see what would be the best strategy to implement lower cost evals
24A few SCM MVPs that I can pitch to our sales team!
25think, design and build together with Claude
26The takeaway is that Anthropic has significant work to operationalize tools for enterprises.
27I work on an agentic AI product - will implement new techniques learned here, esp around Claude Code and evals.
28Swarms, and will put more thought into which model to use as opposed to using Opus for everything
29RFP response system. Accelerator for project implementation
30we’ll see
31Education plugin, watch for it on the marketplace