← Home
Partner Basecamp · Cohort 21

San Francisco

July 16–17, 2026
+24
Net Promoter Score
54
Day 1 Responses
94%
Matched Pairs
91%
Day 2 Response Rate
Programme Satisfaction
Mixed signal — NPS +24 with notable detractor presence.
+24NPS · n=49
39% Promoters47% Passives14% Detractors
95% CI: +5 → +44  ·  True NPS lies within this range with 95% confidence (n=49 respondents)
Confidence Arc
Moderate end-of-Day-1 build confidence — cohort is progressing; Day 2 has room to close remaining gaps (mean 4.0/5).
4.0end-of-D1 build / 5
4.2D2 Design / 5
4.1D2 Commercial / 5
4.2D2 Build / 5
Audience
54 participants across 12 organisations — Transformation Lead majority with applying in practice the most common AI experience level.
Top Organisations
Fractional AI / Ode (18) PwC (12) NEC (6) KPMG (4) Ascendion (3)
Voice of Participant
What participants said — dominant theme from each NPS segment.
Promoters
Hands-on, interactive format with Claude tools effectively built participant confidence and practical AI development skills.
Passives
Technical infrastructure issues and misalignment between Jupyter-heavy content and non-technical roles limited experience despite valuable concepts.
Detractors
Training pitched at beginner level failed experienced engineers and AI practitioners who already mastered foundational topics covered.
Survey & Data Quality
Survey & Data Quality
Day 1 responses54
Day 2 responses49 (91% of Day 1)
Matched pairs46 (94% of Day 2)
Orgs resolved54 of 54 respondents matched to named org
Day 2-only respondents3 (6.1%) — suppresses NPS by 2 points; no Day 1 data, excluded from persona & confidence analysis
Organisations
All Day 1 respondents · n=54
Fractional AI / Ode18 (33%)
PwC12 (22%)
NEC6 (11%)
KPMG4 (7%)
Ascendion3 (6%)
Deloitte3 (6%)
Fractal Analytics2 (4%)
Forgd.AI2 (4%)
Lovelytics1 (2%)
Persistent Systems1 (2%)
DXC Technology1 (2%)
Avaloq1 (2%)
Function × Seniority
All Day 1 respondents · n=54
FunctionPractitionerSr PractitionerMgr / Sr MgrDirector+Partner / MDTotal
Engineering51512124
Architecture252312
Business Leadership145
Project / Engmt3333113
Experience Profile
AI experience level · n=54 · % of row
By Persona
SegmentExploringApplyingDeliveringFrontiern
Developer8%42%25%25%24
Transformation Lead44%39%11%6%18
Architect17%17%42%25%12
By Seniority
SegmentExploringApplyingDeliveringFrontiern
Senior practitioner (5–9 years)4%35%35%26%23
Director, Senior Director, or Principal50%8%25%17%12
Practitioner (0-4 years in role)30%50%20%10
Manager or Senior Manager29%57%14%7
Partner, Managing Director, or Executive50%50%2
How would you describe your current experience with AI tools prior to Basecamp?
Learning and exploring
12 (22%)
Applying in practice
19 (35%)
Delivering independently
13 (24%)
Operating at the frontier
10 (19%)
Prior to today, how much had you worked with Claude or the Anthropic API?
Not at all
3 (6%)
A little
18 (33%)
Regularly
33 (61%)
Did the depth land for this audience?
Technical depth perception · n=54 respondents · % of row
By Persona
SegmentToo basicAbout rightToo advancedn
Developer46%42%12%24
Transformation Lead6%83%11%18
Architect25%75%12
By AI Experience Level
SegmentToo basicAbout rightToo advancedn
Applying in practice11%68%21%19
Delivering independently31%69%13
Learning and exploring92%8%12
Operating at the frontier90%10%10
Overall depth distribution
Too basic
15 (28%)
About right
34 (63%)
Too advanced
5 (9%)
Did the pace work across the room?
Session pace perception · n=54 respondents · % of row
By Persona
SegmentToo slowWell pacedToo fastn
Developer38%46%17%24
Transformation Lead61%39%18
Architect8%92%12
By AI Experience Level
SegmentToo slowWell pacedToo fastn
Applying in practice5%68%26%19
Delivering independently23%69%8%13
Learning and exploring8%50%42%12
Operating at the frontier50%50%10
Overall pace distribution
Too slow — could have covered more
10 (19%)
Well paced
33 (61%)
Too fast — not enough time to apply
11 (20%)
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=54 · 1–5
Mean end-of-D1 build confidence by Persona
Architect4.58/5 · n=12
Developer4.04/5 · n=24
Transformation Lead3.44/5 · n=18
Programme mean: 3.9/5 · 15 cohorts
Mean end-of-D1 build confidence by AI Experience Level
Operating at the frontier4.90/5 · n=10
Delivering independently4.46/5 · n=13
Applying in practice3.63/5 · n=19
Learning and exploring3.17/5 · n=12
Overall end-of-D1 build confidence distribution
1
0 (0%)
2
6 (11%)
3
10 (19%)
4
18 (33%)
5
20 (37%)
How relevant was today's content to your current role?
Content relevance rating · n=54 respondents · 1 = not relevant, 5 = highly relevant
Mean relevance by Persona
Architect4.42/5 · n=12
Developer3.92/5 · n=24
Transformation Lead3.89/5 · n=18
Mean relevance by AI Experience Level
Delivering independently4.38/5 · n=13
Applying in practice4.05/5 · n=19
Learning and exploring4.00/5 · n=12
Operating at the frontier3.50/5 · n=10
Overall relevance distribution
1
0 (0%)
2
3 (6%)
3
11 (20%)
4
22 (41%)
5
18 (33%)
How likely are you to recommend attending this programme to a colleague?
n=49 Day 2 respondents · 91% of Day 1
Score distribution · 0 = not at all likely · 10 = extremely likely
Cohort012345678910
Cohort 2111321310811
Detractors 0–6 Passives 7–8 Promoters 9–10
NPS Breakdown
39%
47%
14%
NPS +24 (n=49)
Promoters (9–10)Passives (7–8)Detractors (0–6)
Programme mean: +43 · 15 cohorts
Promoters mean: 52%
Passives mean: 38%
Detractors mean: 9%
NPS by Segment
Organisations, personas, and experience levels · ≥4 respondents shown
By Organisation
Promoters (9–10)Passives (7–8)Detractors (0–6)
NEC
50%
50%
0%
NPS +50 (n=6)
PwC
42%
50%
8%
NPS +33 (n=12)
Fractional AI / Ode
10%
50%
40%
NPS -30 (n=10)
KPMG
0%
100%
0%
NPS 0 (n=4)
By Persona
Promoters (9–10)Passives (7–8)Detractors (0–6)
Architect
58%
33%
8%
NPS +50 (n=12)
Transformation Lead
50%
39%
11%
NPS +39 (n=18)
Developer
12%
69%
19%
NPS -6 (n=16)
Programme means · 15 cohorts: Architect +51 · Developer +50 · Transformation Lead +36
By AI Experience Level
Promoters (9–10)Passives (7–8)Detractors (0–6)
Learning and exploring
64%
36%
0%
NPS +64 (n=11)
Delivering independently
36%
64%
0%
NPS +36 (n=11)
Applying in practice
40%
47%
13%
NPS +27 (n=15)
Operating at the frontier
11%
44%
44%
NPS -33 (n=9)
Programme means · 15 cohorts: Operating at the frontier +46 · Applying in practice +45 · Learning and exploring +42 · Delivering independently +42
What is the main reason for your score?
n=41 responses · organised by NPS segment
Promoters (score 9–10)· 16 responses
Hands-on, interactive format with Claude tools effectively built participant confidence and practical AI development skills.
9“It’s a great opportunity to learn about gen ai and Claude capabilities, with non obvious insights on optimisations.”
Passives (score 7–8)· 19 responses
Technical infrastructure issues and misalignment between Jupyter-heavy content and non-technical roles limited experience despite valuable concepts.
7“It's helpful for people to know the techniques if they are not familiar with Claude but some technical parts are very complex to understand in a short amount of time, in a small screen.”
Detractors (score 0–6)· 6 responses
Training pitched at beginner level failed experienced engineers and AI practitioners who already mastered foundational topics covered.
2“I am an engineer and this was not a session for engineers, more like for business/sales people. We also had a lot of issues with the WiFi (which was throttling us) and Claude outages, so most of the Jupyter notebook code they provided did not successfully run :(”
End-of-Programme Confidence
n=49 respondents · 1 = not confident, 5 = very confident
How confident are you in your ability to design an evaluation for an AI solution?
12345Mean
2%20%35%43%4.18
Programme mean: 4.1/5 · 15 cohorts
Leaving today, how comfortable are you advising clients on AI when questions arise?
12345Mean
2%24%35%39%4.10
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%20%35%43%4.18
Programme mean: 4.2/5 · 15 cohorts
Confidence Trajectory by Persona
D1→D2 delta · apply-AI confidence · matched pairs only · n=46
Positive delta = confidence grew · negative = dropped
Architect (n=12)
-0.08
Developer (n=16)
-0.25
Transformation Lead (n=18)
+0.83
What one takeaway will you share with a colleague or client?
n=29 responses
Participants recognize Claude's practical value for resolving client issues and see fine-tuning architecture around the core model as key to unlocking possibilities.
“Claude is super helpful in resolving client issues and troubleshooting”
Participants identify evaluation frameworks as critical to programme value, linking model selection directly to cost management and testing requirements.
“Models matter for costs and evals needed”
Participants emphasize understanding agent architecture and context engineering as essential foundations for building effective multi-agent systems.
“how agents work and importance of the AI harness”
What will you build for yourself or apply at work in the next 30 days?
n=32 responses
Participants want to apply agent architectures to specific domain problems like ITSM and healthcare, moving beyond theory to implementation.
“Decision Intelligence using AI agents”
Participants plan to build practical tools addressing real workflows, from business operations to personal productivity integrations.
“I'm going to try to build a software tool to help manage a landscaping business”
Participants identified evaluation frameworks and testing harnesses as critical skills they want to practice and apply immediately.
“Lots of eval harness work!”
Participants are translating training into concrete use cases combining APIs, data sources, and AI capabilities for specific business problems.
“There was a demo I wanted to make, an agent that uses satellite imagery to answer questions in about local business from the Google Places API.”
Participants valued structured practice projects that reinforced engineering skills and leading practices for building production-ready solutions.
“A practice project that helps practice the various skills (like evals)”
NPS Reason
n=41 responses · grouped by NPS segment
Promotersscore 9–10 · 16 responses
1Enjoyed the Basecamp and learned a lot!10
2Interactive and collaborative10
3learned a lot to bring back to pwc, and got more comfortable interacting with claude tools & jupyter notebooks10
4Why not10
5Great information and facilitation10
6Awesome Hands on experience best way to learn10
7Increased learning about effective usage of claude10
8The depth and level at which all hands-on sessions were conducted.10
9Great session10
10Learnt some new ways to use stuff that I use everyday.10
11Explanatory and with a good level of hands on9
12Everything was good9
13its so fun9
14Learned a lot of leading practices on how to build an agent9
15It was great and got me comfortable with using Claude code for the first time9
16It’s a great opportunity to learn about gen ai and Claude capabilities, with non obvious insights on optimisations.9
Passivesscore 7–8 · 19 responses
1Amazing exercise and hands-on8
2There are many components that are bleeding edge that Clients are not ready for8
3Learned a lot of useful information8
4The session was so inspiring and I learned a lot from it.8
5There was a lot to learn8
6It’s well paced and is challenging enough for someone technical but, Claude can walk you through each exercise.8
7I learned a lot from it8
8I think this was very very heavy on Jupyter notebooks which is not relevant to my current role.7
9Hackathon at end7
10Good concepts, would be good for some folks to have more details on the underlying concepts and best practices you are seeing in application. Overall it was good7
11I wish it was even more hands on, and that there is a bit more collaboration with others. The api issues also prevented us from completing stuff7
12Good for ppl with pre-existing technical background but non-technical folks looking to get hands-on ai experience like product folks might feel theres a steep learning curve.7
13The Claude system is responding very slow and had to kill and restart multiple times. Pls have some dedicated env scaled down if needed for training if possible7
14There should have more internal knowledge of Claude7
15It was an interesting session7
16It was useful but I think covered too wide an audience. Some stuff felt boring for engineering or hard for me, a PM7
17Very very useful for the mildly technical but very basic for engineers that do this every day.7
18It was informative , interesting and helped learn and share knowledge7
19It's helpful for people to know the techniques if they are not familiar with Claude but some technical parts are very complex to understand in a short amount of time, in a small screen.7
Detractorsscore 0–6 · 6 responses
1It’s a bit too basic for us6
2Depends on the catalogue. There’sa lot of good stuff here, but it’s pretty basic for most of my coworkers6
3Not differentiated by level of technical skill5
4Super dependant on role and their existing experience in the AI dev space5
5It is a useful training for people starting in the AI applied architect role, but I found it too basic if you already have several years of experience. I was already familiar with most of the topics that were presented4
6I am an engineer and this was not a session for engineers, more like for business/sales people. We also had a lot of issues with the WiFi (which was throttling us) and Claude outages, so most of the Jupyter notebook code they provided did not successfully run :(2
Most Valuable
What one takeaway will you share with a colleague or client? · n=29 responses
1That we don’t need MCPs for every little thing
2Good for pm
3Svg building is cool
4Mcp, eval
5eval is so important
6how agents work and importance of the AI harness
7Models matter for costs and evals needed
8Limitless possibilities
9Cost optimization techniques.
10Agent teams
11Claude is super helpful in resolving client issues and troubleshooting
12AI is real and we are past experiments.
13Evals need to start early
14It's fun and messy to build greenfield together with ai
15Context engineering
16Look for opportunities to simplify AND be more effective
17I learned something valuable
18Endless possibilities and the secret of success is the art of fine tuning the architecture around the core model
19Things are moving fast
20Ask Claude on how to use its features first, learn about the ecosystem which will open up a lot of possibilities. The how part comes after the knowing that they exist.
21The models are ready. Failures are design.
22How powerful Claude code and Cowork are .
23Knowledge of claude code
24Look at how capable Haiku is with the right prompting and evals.
25Roadmap and new features
26Don’t use Claude in a Jupyter notebook
27how to create Claude code agent
28Design Eval before building the systems.
29LLM as a judge evaluation concepts.
30-Day Intentions
What will you build for yourself or apply at work? · n=32 responses
1I’m always building
2Lots of eval harness work!
3Wealth dashboarss
4Data Architecture Solution
5Modernization project
6Key learnings, leading practices on how to build robust solutions
7Expanding my Relationship memory bot
8plan to apply the key lessons from training to day to day tasks - especially creation of evals
9Prob some agents for ITSM managed services.
10A practice project that helps practice the various skills (like evals)
11I’m going to try to build a software tool to help manage a landscaping business
12Decision Intelligence using AI agents
13Email clean up tool
14Degital twinn
15The sky's the limit!
16An agent in healthcare focused on eyes.
17Yes, I will work on build out
18Battling agents over chess
19Some tooling for integration with tmux and other tools I use
20Build a couple of quick products for my personal use and tools that can help me at work. This will give me more confidence in speaking and advocating about this tech.
21Eval agent
22Robotic feedback
23Client Solutions built with CCode, Personal productivity tools
24Data requirements for financial services agent
25I will build some personal projects and help clients with enterprise problems
26AI agent for security
27There was a demo I wanted to make, an agent that uses satellite imagery to answer questions in about local business from the Google Places API.
28Agents
29I do this as my day job, so most of this was relevant
30I will share this program.
31AI agents
32I will use gathered insights to improve even further the development of the gen ai based solutions I am working on.