← Home
Partner Basecamp · Cohort 20

San Francisco

July 14–15, 2026
+35
Net Promoter Score
46
Day 1 Responses
88%
Matched Pairs
104%
Day 2 Response Rate
Programme Satisfaction
Positive NPS of +35 — promoters outweigh detractors with moderate passive presence.
+35NPS · n=48
46% Promoters44% Passives10% Detractors
95% CI: +17 → +54  ·  True NPS lies within this range with 95% confidence (n=48 respondents)
Confidence Arc
Moderate end-of-Day-1 build confidence — cohort is progressing; Day 2 has room to close remaining gaps (mean 3.8/5).
3.8end-of-D1 build / 5
4.1D2 Design / 5
4.2D2 Commercial / 5
4.0D2 Build / 5
Audience
46 participants across 12 organisations — Transformation Lead majority with applying in practice the most common AI experience level.
Top Organisations
PwC (10) Deloitte (10) Persistent Systems (6) DXC Technology (4) Quantium (4)
Voice of Participant
What participants said — dominant theme from each NPS segment.
Promoters
Hands-on Claude experience with real-world scenarios and expert instruction delivered at the right pace excited learners.
Passives
Technical depth and developer focus created engagement gaps for business-oriented consultants who needed different content targeting.
Detractors
Compressed timeline and assumed programming knowledge excluded functional consultants from finance and non-technical consulting domains.
Facilitator Notes
Room Composition
DXC and Deloitte crews were a far less technical audience for the most part. DXC is a new crew this cohort.
What Landed & What Didn't
Agent Build Hackathon landed well — the room had fun with it. Evals fell shortest: only ~2 people made it through to the 'LLM as a judge' section, so most didn't reach the deeper material.
Gaps Filled & Course-Corrections
More facilitators on the ground this cohort to make people feel supported (hoping it shows in the results). Had to troubleshoot tech setup issues — Zscaler blocking on DXC corporate PCs. An engaged Applied AI (Anthropic) partner ran in-person Q&A and did a good job getting out and talking to people.
Room Dynamics
Felt good in the room overall. Left side of the room (facing from the front) was far more chatty and engaged than the right side.
For Next Time
Pre-empt corporate-PC setup issues (Zscaler) before Day 1. Consider more scaffolding/pacing on Evals so more of the room reaches LLM-as-a-judge.
DXC — New crew, less technical for the most part; taking notes to send back to their PM at DXC; hit tech setup issues (Zscaler on corporate PCs).
Deloitte — Less technical audience for the most part.
Survey & Data Quality
Survey & Data Quality
Day 1 responses46
Day 2 responses48 (104% of Day 1)
Matched pairs42 (88% of Day 2)
Orgs resolved46 of 46 respondents matched to named org
Day 2-only respondents6 (12.5%) — suppresses NPS by 3 points; no Day 1 data, excluded from persona & confidence analysis
Organisations
All Day 1 respondents · n=46
PwC10 (22%)
Deloitte10 (22%)
Persistent Systems6 (13%)
DXC Technology4 (9%)
Quantium4 (9%)
NEC3 (7%)
Lovelytics2 (4%)
McKinsey2 (4%)
Ascendion2 (4%)
AWS1 (2%)
EPAM1 (2%)
Forgd.AI1 (2%)
Function × Seniority
All Day 1 respondents · n=46
FunctionPractitionerSr PractitionerMgr / Sr MgrDirector+Partner / MDTotal
Engineering2442–12
Architecture––77–14
Business Leadership––1359
Project / Engmt3132211
Experience Profile
AI experience level · n=46 · % of row
By Persona
SegmentExploringApplyingDeliveringFrontiern
Transformation Lead30%55%5%10%20
Architect7%43%36%14%14
Developer33%33%25%8%12
By Seniority
SegmentExploringApplyingDeliveringFrontiern
Manager or Senior Manager20%53%20%7%15
Director, Senior Director, or Principal14%36%36%14%14
Partner, Managing Director, or Executive14%57%—29%7
Senior practitioner (5–9 years)40%60%——5
Practitioner (0-4 years in role)60%20%20%—5
How would you describe your current experience with AI tools prior to Basecamp?
Learning and exploring
11 (24%)
Applying in practice
21 (46%)
Delivering independently
9 (20%)
Operating at the frontier
5 (11%)
Prior to today, how much had you worked with Claude or the Anthropic API?
Not at all
3 (7%)
A little
19 (41%)
Regularly
24 (52%)
Did the depth land for this audience?
Technical depth perception · n=46 respondents · % of row
By Persona
SegmentToo basicAbout rightToo advancedn
Transformation Lead10%45%45%20
Architect7%93%—14
Developer33%58%8%12
By AI Experience Level
SegmentToo basicAbout rightToo advancedn
Applying in practice10%57%33%21
Learning and exploring9%64%27%11
Delivering independently33%67%—9
Operating at the frontier20%80%—5
Overall depth distribution
Too basic
7 (15%)
About right
29 (63%)
Too advanced
10 (22%)
Did the pace work across the room?
Session pace perception · n=46 respondents · % of row
By Persona
SegmentToo slowWell pacedToo fastn
Transformation Lead—95%5%20
Architect—93%7%14
Developer25%58%17%12
By AI Experience Level
SegmentToo slowWell pacedToo fastn
Applying in practice—90%10%21
Learning and exploring9%73%18%11
Delivering independently22%78%—9
Operating at the frontier—100%—5
Overall pace distribution
Too slow — could have covered more
3 (7%)
Well paced
39 (85%)
Too fast — not enough time to apply
4 (9%)
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=46 · 1–5
Mean end-of-D1 build confidence by Persona
Architect4.36/5 · n=14
Developer3.75/5 · n=12
Transformation Lead3.25/5 · n=20
Programme mean: 3.9/5 · 16 cohorts
Mean end-of-D1 build confidence by AI Experience Level
Operating at the frontier4.80/5 · n=5
Delivering independently4.33/5 · n=9
Applying in practice3.38/5 · n=21
Learning and exploring3.36/5 · n=11
Overall end-of-D1 build confidence distribution
1
1 (2%)
2
3 (7%)
3
13 (28%)
4
20 (43%)
5
9 (20%)
How relevant was today's content to your current role?
Content relevance rating · n=46 respondents · 1 = not relevant, 5 = highly relevant
Mean relevance by Persona
Architect4.50/5 · n=14
Developer4.25/5 · n=12
Transformation Lead3.85/5 · n=20
Mean relevance by AI Experience Level
Operating at the frontier4.80/5 · n=5
Delivering independently4.22/5 · n=9
Learning and exploring4.09/5 · n=11
Applying in practice4.00/5 · n=21
Overall relevance distribution
1
0 (0%)
2
1 (2%)
3
12 (26%)
4
12 (26%)
5
21 (46%)
How likely are you to recommend attending this programme to a colleague?
n=48 Day 2 respondents · 104% of Day 1
Score distribution · 0 = not at all likely · 10 = extremely likely
Cohort012345678910
Cohort 2014912715
■ Detractors 0–6■ Passives 7–8■ Promoters 9–10
NPS Breakdown
46%
44%
10%
NPS +35 (n=48)
Promoters (9–10)Passives (7–8)Detractors (0–6)
Programme mean: +42 · 16 cohorts
Promoters mean: 52%
Passives mean: 39%
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)
Persistent Systems
80%
20%
0%
NPS +80 (n=5)
DXC Technology
50%
50%
0%
NPS +50 (n=4)
Deloitte
56%
33%
11%
NPS +44 (n=9)
PwC
22%
44%
33%
NPS -11 (n=9)
By Persona
Promoters (9–10)Passives (7–8)Detractors (0–6)
Developer
70%
30%
0%
NPS +70 (n=10)
Architect
50%
43%
7%
NPS +43 (n=14)
Transformation Lead
33%
50%
17%
NPS +17 (n=18)
Programme means · 16 cohorts: Architect +52 · Developer +45 · Transformation Lead +38
By AI Experience Level
Promoters (9–10)Passives (7–8)Detractors (0–6)
Operating at the frontier
75%
25%
0%
NPS +75 (n=4)
Delivering independently
38%
62%
0%
NPS +38 (n=8)
Learning and exploring
54%
27%
18%
NPS +36 (n=11)
Applying in practice
42%
47%
10%
NPS +32 (n=19)
Programme means · 16 cohorts: Applying in practice +45 · Learning and exploring +45 · Delivering independently +42 · Operating at the frontier +32
What is the main reason for your score?
n=43 responses · organised by NPS segment
Promoters (score 9–10)· 21 responses
Hands-on Claude experience with real-world scenarios and expert instruction delivered at the right pace excited learners.
10“It covers a wide range of Claude capabilities, from fundamentals to advanced use cases. While it may be challenging for beginners, it is very valuable for people who want to apply Claude to real-world work.”
Passives (score 7–8)· 17 responses
Technical depth and developer focus created engagement gaps for business-oriented consultants who needed different content targeting.
8“The day one prework was really good, and had some quality learnings. Day two was hands on, and you had to move at a decent speed - enjoyed all of the exercises, saw things in practice, were able to discuss things like evals, and the hackathon was good. Preferred day two, score was dragged down by day one being a bit on the simpler side - although this comes from our group who is more hands on than most in the broader cohort with us over these two days”
Detractors (score 0–6)· 5 responses
Compressed timeline and assumed programming knowledge excluded functional consultants from finance and non-technical consulting domains.
5“For a functional consultant who is working with clients on how to deploy agents within their finance organizations, I felt that the training may have been more fit for technical practitioners with extensive programming experience. There was a lot of foundational knowledge we were assumed to have known.”
End-of-Programme Confidence
n=48 respondents · 1 = not confident, 5 = very confident
How confident are you in your ability to design an evaluation for an AI solution?
12345Mean
——23%48%29%4.06
Programme mean: 4.1/5 · 16 cohorts
Leaving today, how comfortable are you advising clients on AI when questions arise?
12345Mean
——21%44%35%4.15
Programme mean: 4.2/5 · 16 cohorts
How confident are you in your ability to lead a conversation about Anthropic and Claude with a client?
12345Mean
—6%15%50%29%4.02
Programme mean: 4.2/5 · 16 cohorts
Confidence Trajectory by Persona
D1→D2 delta · apply-AI confidence · matched pairs only · n=42
Positive delta = confidence grew · negative = dropped
Architect (n=14)
+0.14
Developer (n=10)
+0.20
Transformation Lead (n=18)
+0.22
What one takeaway will you share with a colleague or client?
n=30 responses
Participants valued understanding Claude's distinctive capabilities and how it approaches problems differently from other models.
“Claude does thing differently with skills and agents”
Participants recognized that defining evaluation frameworks upfront transforms AI from experimental to reliably powerful in production.
“Defining the problem, eval and data is first course of action, not just throw ai at everything”
Participants identified prompt and context engineering as critical disciplines for building effective AI agent systems.
“Importance of prompt and context engineering”
Participants learned that technology readiness matters less than selecting the right use case and owning the implementation layer.
“Technology is not a blocker, selecting right use case, own application layer and the way we implement it.”
Participants gained concrete, applied knowledge of connecting development tools and building working use cases within two days.
“My takeaway is how much you can learn in two days about the tool! I am coming away with knowledge on how to connect GitHub to vs code, to Claude code, etc and to build applicable use cases”
What will you build for yourself or apply at work in the next 30 days?
n=33 responses
Participants want to build and deploy agents across personal productivity and managed enterprise contexts.
“Managed agents”
Participants are adopting Claude's specific tools and capabilities to ship production applications with confidence.
“A production grade share trading platform with Claude”
Participants are immediately translating model selection and context management knowledge into client deliverables.
“I will apply my knowledge of switching between Claude models and giving Claude the right context”
Participants are building evaluation frameworks tailored to their specific client domains and workflows.
“Building Evals for client specific use case in my work area”
Participants recognize Claude tools and MCP as essential infrastructure for building production applications.
“Use Claude tools, mcp, and plugins to build apps”
Partner Basecamp Survey Report · Cohort 20 · Generated by analyze.py v2.5
NPS Reason
n=43 responses · grouped by NPS segment
Promotersscore 9–10 · 21 responses
1Very valid real life scenarios10
2Good course to get hands on time with Claude Code and understand how Claude works under the hood10
3Right pace, great content, amazing trainers.10
4Loved the class. Learned a ton.10
5This event was a lot of fun and I see the usefulness of Claude10
6Doing thing different with efficiency10
7Well designed sessions and assignments10
8use cases provided, sharing experiences, ability to ask questions, materials10
9Great exposure10
10Learned a lot of new things especially fundamentals and how to apply ai beyond my dev work in Claude code10
11Knowing the world of possibilities with Claude10
12great hands on exercises across interesting topics at a depth that was at time challenging, which I liked10
13Great interactive sessions and minimal slides. Perfect balance!10
14Great delivery of the helpful attitude of the trainers.10
15It covers a wide range of Claude capabilities, from fundamentals to advanced use cases. While it may be challenging for beginners, it is very valuable for people who want to apply Claude to real-world work.10
16Lesson on Evals9
17It’s a great training event. Not applicable to engine in the org though.9
18Lots of fun and great learning9
19Breadth of various practical concepts covered9
20It was nice to learn usage and how we explain to our customers. This is a nice event to get started with Claude code.9
21Great learning experience9
Passivesscore 7–8 · 17 responses
1Very insightful content8
2Well organized8
3Good coverage8
4Overall balanced8
5Very extensive but very detailed. More focused imo for technical crowd, builders versus business integrators8
6Great overall experience really showed the power of Claude in action with real world examples. Might be too deep for those in business development and client management.8
7Very informative with complex exercises help people understand that non coders can build now8
8It was fun and educational but missing a couple key things - like top Anthropic techies for q&a.8
9The day one prework was really good, and had some quality learnings. Day two was hands on, and you had to move at a decent speed - enjoyed all of the exercises, saw things in practice, were able to discuss things like evals, and the hackathon was good. Preferred day two, score was dragged down by day one being a bit on the simpler side - although this comes from our group who is more hands on than most in the broader cohort with us over these two days8
10The second day was on the money in terms of balancing relevant technical concepts with strategic concepts.8
11Hands on and engaging.7
12Depends ON User level. It's a great program for a coder / software developer level audience. Not for business or strategic level user based on assignment or connections it allows us to make.7
13It was a little too developer focused I felt often times confused and didn’t get to ask too much questions. People walked around but by then I already had forgotten my question. Wish there was a question bank or more dummies down explanations of stuff.7
14Basecamp is tailored for developers. Would be good to have separate session for transformation leads7
15Well it was a lot of great information, nicely put together into 4 main parts across 2 days. I felt it was a bit too aggressive from time perspective, it otherwise great experience.7
16Skills you learn is basically the same as silljar’s, and Japa is little too far away.7
17I felt some of the material was a bit too technical for me to follow, but I overall learned and absorbed the technical vocabulary I need to lead client conversations that I previously didn’t have7
Detractorsscore 0–6 · 5 responses
1Not as directly relevant for my work in finance consulting6
2Very basic .. wish it could help in ontology and use of ontology with Claude to create agents6
3Very technical but a great way to learn what back end developers do and why. Then also brought a great group of people together to network and learn from eachother6
4Lots of useful content, particularly the code bases. But each exercise ran pretty short which meant that there wasn’t a lot of time to engage with each session.6
5For a functional consultant who is working with clients on how to deploy agents within their finance organizations, I felt that the training may have been more fit for technical practitioners with extensive programming experience. There was a lot of foundational knowledge we were assumed to have known.5
Most Valuable
What one takeaway will you share with a colleague or client? · n=30 responses
1Context engineering and Evals
2The 3 pillars
3Various different aspects of claude
4Jump in, jump all the way in. And do it now.
5Invest in AI
6Building with AI is feasible and with the right evals it can be an incredibly powerful tool
7Claude does thing differently with skills and agents
8Model is not the problem, prompt is…
9Importance of prompt and context engineering
10investing in harness and preparation is the key use case needs to be defined
11Carving out learning time
12Technology is not a blocker, selecting right use case, own application layer and the way we implement it.
13The more you use claude. The better at everything you will be
14Defining the problem, eval and data is first course of action, not just throw ai at everything
15Eval and agent building steps in details
16Model consumption tricks that I will study
17Not to judge an AI solution just by the m LLM it uses and how long it extended its thinking is.
18“less” powerful models can outperform “more” powerful models when correctly implemented
19Bring an API key
20Stay open and don’t give up when you think you are not following it
21The memory sharing idea
22Take 10 hours to play with and learn claude and it will change the way you work and communicate about AI
23Fun
24Prompt caching and batching
25Inspiration
26Discuss evals upfront when scoping projects
27My takeaway is how much you can learn in two days about the tool! I am coming away with knowledge on how to connect GitHub to vs code, to Claude code, etc and to build applicable use cases
28The frameworks for classifying AI use cases and approaches.
29Eval is important. Not vibe but score.
30There’s a lot to learn!
30-Day Intentions
What will you build for yourself or apply at work? · n=33 responses
1Agents for personal use
2Already applying
3Managed agents
4POV
5I’ll start playing with a stock picker
6Something to monitor sales channels
7Use Claude tools, mcp, and plugins to build apps
8Spec driven development
9Agentic SDLC
10Personal knowledge wiki
11support agent
12Get more hands on build some agent
13Building Evals for client specific use case in my work area
14Powerpoint presentation reviewer
15Speak to clients about the capabilities and art of possible.
16Optimizing inference is a key new topic as well as memory for what I'll be building for my next agentic client build
17A production grade share trading platform with Claude
18I will use evals in my everyday prompts and also be more cautious of token consumption and what model I am using. I will continue to build dashboards and use Claude code to troubleshoot.
19Agent for transformation
20Start with asking clause to brainstorm with me for the internal hackathon!
21turn my HTML demos into something more real
22Turn the lessons into client demos
23I have a few projects lined up where I will leverage claude
24I will make a workflow that enables me to be confident of Claude’s output
25Use github to share code bases with colleagues. Also built GTM materials to support proposals with the info learned on AI in general and Anthropic specifically
263rd brain
27RAG pipeline, and Full build for a sophisticated conversational agent
28Security solution
29An agentic recommender system for my client
30I will apply my knowledge of switching between Claude models and giving Claude the right context
31An Agent! And applying prompt caching and batching in our workflows.
32Build demo agent by myself.
33Understanding behind the scenes of a lot of the code that was shipped by downloaded the activity files