Partner Basecamp · Cohort 32

Programme Satisfaction

Strong promoter majority — NPS +56 reflects broad programme endorsement.

+56NPS · n=71

61% Promoters35% Passives4% Detractors

95% CI: +43 → +70 · True NPS lies within this range with 95% confidence (n=71 respondents)

Confidence Arc

Strong end-of-Day-1 build confidence — cohort leaves Day 1 ready to apply Claude in client work (mean 4.1/5).

4.1end-of-D1 build / 5

4.1D2 Design / 5

4.2D2 Commercial / 5

4.2D2 Build / 5

Audience

68 participants across 20 organisations — Developer majority with applying in practice the most common AI experience level.

Top Organisations

NEC (19) LTM (8) IBM (5) Provectus (5) PwC (4)

Voice of Participant

What participants said — dominant theme from each NPS segment.

Promoters

Great content and helpful frameworks for client engagements!

Passives

Good mix of lecture, hands-on, and audience variety

Detractors

Poor format. We spent 70% time working on our laptop using Claude prompt. Not a good format for an in person workshop

Function	Practitioner	Sr Practitioner	Mgr / Sr Mgr	Director+	Partner / MD	Total
Engineering	14	5	5	2	–	26
Architecture	1	10	5	5	2	23
Business Leadership	–	–	2	5	1	8
Project / Engmt	2	3	5	1	–	11

Function

Practitioner

Sr Practitioner

Mgr / Sr Mgr

Director+

Partner / MD

Total

Engineering

–

Architecture

Business Leadership

–

Project / Engmt

–

Segment	Exploring	Applying	Delivering	Frontier	n
Developer	12%	77%	8%	4%	26
Architect	13%	35%	30%	22%	23
Transformation Lead	26%	21%	32%	21%	19

Segment

Exploring

Applying

Delivering

Frontier

Developer

12%

77%

Architect

13%

35%

30%

22%

Transformation Lead

26%

21%

32%

21%

Segment	Exploring	Applying	Delivering	Frontier	n
Senior practitioner (5–9 years)	6%	50%	33%	11%	18
Practitioner (0-4 years in role)	18%	71%	6%	6%	17
Manager or Senior Manager	24%	41%	35%	—	17
Director, Senior Director, or Principal	23%	31%	8%	38%	13
Partner, Managing Director, or Executive	—	—	33%	67%	3

Segment

Exploring

Applying

Delivering

Frontier

Senior practitioner (5–9 years)

50%

33%

11%

Practitioner (0-4 years in role)

18%

71%

Manager or Senior Manager

24%

41%

35%

—

Director, Senior Director, or Principal

23%

31%

38%

Partner, Managing Director, or Executive

—

33%

67%

Segment	Too basic	About right	Too advanced	n
Developer	15%	81%	4%	26
Architect	4%	91%	4%	23
Transformation Lead	5%	74%	21%	19

Segment

Too basic

About right

Too advanced

Developer

15%

81%

Architect

91%

Transformation Lead

74%

21%

Segment	Too basic	About right	Too advanced	n
Applying in practice	9%	78%	12%	32
Delivering independently	7%	87%	7%	15
Learning and exploring	9%	91%	—	11
Operating at the frontier	10%	80%	10%	10

Segment

Too basic

About right

Too advanced

Applying in practice

78%

12%

Delivering independently

87%

Learning and exploring

91%

—

Operating at the frontier

10%

80%

10%

Segment	Too slow	Well paced	Too fast	n
Developer	—	88%	12%	26
Architect	9%	78%	13%	23
Transformation Lead	11%	68%	21%	19

Segment

Too slow

Well paced

Too fast

Developer

—

88%

12%

Architect

78%

13%

Transformation Lead

11%

68%

21%

Segment	Too slow	Well paced	Too fast	n
Applying in practice	3%	81%	16%	32
Delivering independently	7%	73%	20%	15
Learning and exploring	9%	82%	9%	11
Operating at the frontier	10%	80%	10%	10

Segment

Too slow

Well paced

Too fast

Applying in practice

81%

16%

Delivering independently

73%

20%

Learning and exploring

82%

Operating at the frontier

10%

80%

10%

Cohort	0	1	2	3	4	5	6	7	8	9	10
Cohort 32				1		1	1	10	15	8	35

Cohort

Cohort 32

1	2	3	4	5	Mean
—	6%	10%	52%	32%	4.11

Mean

—

10%

52%

32%

4.11

1	2	3	4	5	Mean
—	3%	13%	45%	39%	4.21

Mean

—

13%

45%

39%

4.21

1	2	3	4	5	Mean
—	6%	11%	38%	45%	4.23

Mean

—

11%

38%

45%

4.23

NPS Reason

n=54 responses · grouped by NPS segment

Promotersscore 9–10 · 34 responses

1Great content and helpful frameworks for client engagements!10

2I learned great things10

3very informative, liked the activities10

4I liked the interactive/hands on elements10

5Great way to set a strong foundation for claude fundamentals10

6I like the presentation and the content.10

7have same issue10

8Very informative and hands on, great facilitators and location.10

9Course was well structured10

10Insights into the sdk and new ways unlocked to solve problems10

11Very handsome helped understand latest information10

12Great content and energy, pushing the industry forward.10

13The hackathon was a very cool way to apply everything we learned10

14Very enjoyable experience and a lot of learning10

15Depth of evals and inferences , which are the success metrics10

16It was good session and helpful to understand in deep10

17Very structured program to introduce a progressive ramp to advanced concepts. True to its name.10

18Very informative.10

19Relevance and quality of sessions as well as the preparation of team and overall experience.10

20Great hands-on content10

21Very engaging content and delivery.10

22I feel this is a great entry level workshop10

23Because there was communication with other participants.10

24I see many insights about sound architecture and the best way to build AI Solutions with great use cases and examples that can help our customers.10

25Enjoyed the Learning…10

26It provides an end-to-end understanding of the key components needed to build AI agents, and it is also highly practical.10

27Coverage from basics to hackathon10

28The content was insightful. I appreciate the experts sharing experiences and lessons learned.10

29It allows you to reaffirm that AI is a partner in your work10

30learned new things9

31I can learn how Claude works and how I use Claude9

32The main reason was the Amazing instructors.9

33I really like the experience of this basecamp. I learn a lot9

34importance of eval9

Passivesscore 7–8 · 18 responses

1Good mix of lecture, hands-on, and audience variety8

2The content is good to distill the need for different components of the agentic workflow.8

3Interesting applications and hand on exercises.8

4Liked the sessions8

5Very helpful content; would be a bit better if it was more hands on with Claude code vs driven through jupityr notebooks8

6The content is rich and helpful but a bit dense8

7Feel like I learned a lot8

8Content felt a bit introductory. I really enjoyed the hands on exercises but wish there was a bit more debrief. A lot of content was covered so I understand that would be hard to accomplish. Lastly, I’d be interested to go more into architecture and making trade offs of AI solutions.8

9I enjoyed the all program and the enthusiasm is stimulus.8

10Some of the shared code was invaluable as a starting point and reference implementation.8

11I was able to learn systematically learn about Evals and LLM as a judge, and I was able to exchange imformation with people from other company.8

12Liked the engagement of the organizers with the group and the structure of the training.8

13Depends on the type of colleague. Not suited for all7

14If there was more overview of the concepts and stronger instructions on the tasks we did individually I’d be more likely to recommend it. Perhaps a demo of how we work through each task7

15The organization was good, particularly like the exercises. However much of day 1 content was basics in my opinion. I also wished there was more around selling , objection handling and navigating realistic client scenarios for example taking through phase 1 2 37

16fast paced, good program, but need a day to just elongate the same content but provide better heads pace for learning.7

17Great base camp and had some good takeaways and learned some new tools. The only room for improvement would to be a little more hands allowing us to learn from the experts rather than turning us loose for us to dive right in.7

18Great content and practice labs. But I think nobody needs to travel across the globe for this. They can do this from home. Colossal waste of money and company resources.7

Detractorsscore 0–6 · 2 responses

1APIを使った実装は日本のほとんどの人にとって貴重な体験であるため。6

2Poor format. We spent 70% time working on our laptop using Claude prompt. Not a good format for an in person workshop3

Most Valuable

What one takeaway will you share with a colleague or client? · n=44 responses

1The way we can get more out of stronger models while keeping costs down.

2Evals

3model selection really matters, can get around knowledge with tools/prompting

4Eval driven development

5Evals

6The spend of advisors to have long term benefits

7Evals

8Evals.

9Get hands on. Start solving.

10Claude will do many things and applying it wherever the possibility for 95%

11Eval is important and drives quality

12The evalution of Claude is astonishing!!

13Today, evals lead development.

14Claude agent optimization isn’t as simple as just using an older model, it depends on a lot of optimization criteria like cache and triage

15The optimization technique and the leavers

16A lot of the prompt approaches and evals are very interesting and useful for any project

17Eval, prompt rescue and diagnosis is extremely important

18Evals important and how to have that as a best practice when building agent , not after thought

19We should choose right model with evals.

20Evals

21The way we did the hands on. Was great

22agent knowledge

23Implemention depends on multiple factors

24Speculative caching was a new technique that I found to be quite interesting

25how to design eval

26エージェント実装に関する細かい調整。レスポンシビリティとか、コストとか。

27Ask Claude but there are some tips you should know to use it better

28Start using Claude. Limitless possibilities.

29The importance of seemingly trivial details like compaction, model selection, etc.

30How to properly run evals

31Try it and explore and here we are help.

32Increased focus on cost of model performance when designing an agent, and performance/cost analysis

33Compact the context!

34Embrace the possibilities!

35I learn to structure all the best practices and not only based on my experience but in great tools and knowledge.

36Evals are cool. I want to build more.

37eval

38A lot of takeaways from the code examples to share!

39I'll share colleague how to eval

40Practice your craft

41Claude Advancements

42Token Economics

43How to approach problem solving using AI?

44Changes of the times

30-Day Intentions

What will you build for yourself or apply at work? · n=50 responses

1A more robust chief of staff agent for myself!

2Evals

3html pages

4Building an swarming agent

5Probably a lot of internal tooling!

6Continue work on financial chat bot. Will look for opportunities to utilize agents at work as well

7Agentic AI - building agents per the standards explained in the base camp

8Redesign our AI for operations platform

9Parallel agents and better model chennai’s proper evaluation rules

10Trading BOT

11Automated Outlook email action taker

12Eval generation tools

13Negotiations Simulator

14Debt analysis and transform

15Probably an improved eval framework and improvements.

16Multiagent environment

17I apply what I learn about Claude to my or our AI agent to blush up and up the level.

18Healthcare AI solutions

19Agentic solutions

20Se Will Run a poc for an AI briefer using Claude SDK

21agent system

22Evals

23Continue to build and enable clients with their Claude deployments

24an agent that analyze logs and create reports

25プロダクトにAIを組み込みたい。NECではほとんどその実績がないはずなので。

26I am planning to Build a GEO Agent which help to understand AI visibility for your website

27Adopting Claude for full SDLC

28Evals are very important in my daily work

29Use as an agentic coding tool to perform app modernization and migration

30Explore more about multi-agent architectures

31Agent swarm orchestration with 3rd party inference

32Will complete the demo we were building.

33Agentic plugins and eval suites

34Context engineering

35Want to build up my hackathon agent more

36An internal application to demonstrate capabilities to senior leadership

37Evals!!

38I will better myself in using Claude code to develop and better my day to day tasks

39The next step is to build a complex architecture to leverage Claude platform across all the organisation and the our customers.

40Looking to build a personal knowledgebase and software factory.

41I will make project map in which I can collaborate.

42A few more sophisticated eval suites for existing systems.

43incorporate eval

44Apply some of the knowledge on actual client projects.

45Multi Agents, Security, Evals, and more

46I’d like to review and improve the agent evaluation pipeline.

47Lots of stuff

48I will create evals for my agentic AI application.

49More than one thing, it changed on how I approach and use AI will be better

50In system development, ensure that the correction of issues, horizontal deployment, and prevention of recurrence are considered.

Day 1 responses	68
Day 2 responses	71 (104% of Day 1)
Matched pairs	59 (83% of Day 2)
Orgs resolved	68 of 68 respondents matched to named org
Day 2-only respondents	12 (16.9%) — no NPS data; no Day 1 data, excluded from persona & confidence analysis

San Francisco