AI Task Force · Samba ATF · Q2 2026
Lever 04 · Accountability

The adoption leaderboard.

How we measure: efficiency formula (smart usage), not raw volume. Someone spending $40/month using Projects + Skills + Connectors scores higher than someone spending $100/month chatting. Leaderboards rank the smart usage, not the loud usage.

Assigned Seats
336
company-wide
Active Users
284
used Claude in window
Adoption Rate
84.5%
of assigned seats
Samba Employees
332
total headcount
Never Used
62
our gap to close
Data fresh through 2026-05-01 Spend data Jan-Mar only
Cross-Surface MVPs

Users in the top-10 of three or more boards.

True power users — people who use Claude across multiple surfaces (chat, projects, code, connectors, skills). The gold-standard signal for "AI as operating system."

Top 15 cross-surface users

Counts the number of leaderboards each user appears in (top-10).
1Alvir Navin
Executive · COO
6 boards
2Jay Fowdar
Product Delivery
5 boards
3Stephen Park
Programmatic Consulting
4 boards
4Enerico Miensky
Technical Services
4 boards
5Jason Mussman
Commercial Partnerships
4 boards
6Ankitha Ananda
MSCI Strategic Initiatives
4 boards
7Sid Dani
AI Task Force
4 boards
8Sam Lenihan-Franks
Measurement Science
4 boards
9Brian Morel
Marketing
3 boards
10Nithin Srivatsa
Measurement Science
3 boards
11Alyson Sprague
Measurement Science
3 boards
12Szymon Górka
Data Science
3 boards
13Krzysztof Sankiewicz
PM · Data Licensing
3 boards
14Lech Twaróg
Identity & Audience
3 boards
15Ashwin Navin
Executive · CEO
3 boards
Usage by surface

Top 10 in each leaderboard.

Each board ranks usage of a single Claude surface or capability. Eight boards total — the per-surface signal that feeds the cross-surface ranking above.

Messages

Chat messages sent · top 10
1Anusha Mourshed
MSCI
3,715
2Jay Fowdar
Product Delivery
2,750
3Brian Morel
Marketing
2,086
4Tomasz Pilawa
Architecture & Ops Excellence
2,055
5Stephen Park
Programmatic Consulting
2,050
6Evgeny Popov
Commercial Partnerships
2,002
7Enerico Miensky
Technical Services
1,697
8Nithin Srivatsa
MSCI
1,552
9Jagdip Sekhon
PM · Privacy
1,430
10Jason Mussman
Commercial Partnerships
1,413

Conversations

Distinct conversations started · top 10
1Yasmin Sanders
Commercial Partnerships, AU
568
2Alyson Sprague
MSCI
411
3Christopher Villani
Product Delivery
410
4Stephen Park
Programmatic Consulting
354
5Jay Fowdar
Product Delivery
303
6Rose Barry
Agency Solutions, UK
288
7Jason Mussman
Commercial Partnerships
282
8Szymon Górka
Data Science
278
9Ariana Otero
MSCI
256
10Alvir Navin
Executive
253

Code Sessions

Claude Code CLI sessions · top 10
1Ankitha Ananda
MSCI
4,797
2Sid Dani
AI Task Force
2,802
3Jack Tsai
1,867
4Krzysztof Sankiewicz
PM · Data Licensing
824
5Grzegorz Parka
Data Technology
657
6Krzysztof Baranski
Content ID
560
7Alvir Navin
Executive
519
8Michael Zennadi
AI Task Force
510
9Abdulmalak Albeik
Identity & Audience
489
10Lech Twaróg
Identity & Audience
485

Lines of Code

Code lines added + removed · top 10
1Sid Dani
AI Task Force
601,195
2Lech Twaróg
Identity & Audience
138,595
3Krzysztof Sankiewicz
PM · Data Licensing
120,281
4Ankitha Ananda
MSCI
102,012
5Abdulmalak Albeik
Identity & Audience
98,757
6Walter Chen
Commercial Partnerships
78,157
7Max Shiner
Finance
77,500
8Sam Lenihan-Franks
MSCI
68,835
9Michael Zennadi
AI Task Force
64,602
10Jonathan Becker
Data Technology
56,286

Cowork Sessions

Agent-mode sessions · top 10
1Ankitha Ananda
MSCI
159
2Andrew Beames
Commercial Partnerships, EU
102
3Thomas Beasley
MSCI
101
4Sam Lenihan-Franks
MSCI
97
5Jaya Aswani
R&D Executive
92
6Michela Fitten
Marketing
78
7Sid Dani
AI Task Force
75
8Jed Hale
68
9Ashwin Navin
Executive · CEO
66
10Andy Shiels
Marketing
54

Connector Uses

Connector invocations · top 10
1Landis Current
MSCI
11,286
2Justin Lundvall
Core Products
8,871
3Jay Fowdar
Product Delivery
5,757
4Sam Lenihan-Franks
MSCI
3,712
5Alvir Navin
Executive
2,855
6Michela Fitten
Marketing
2,713
7Amanda Riley
Human Resources
2,366
8Stephen Park
Programmatic Consulting
2,358
9Lulu Tian
Programmatic Consulting
2,029
10Anthony Livreri
Programmatic Consulting
1,678

Skill-Days

Distinct skills used per day, summed · top 10
1Alyson Sprague
MSCI
110
2Ashwin Navin
Executive · CEO
80
3Ariana Otero
MSCI
73
4Lewis Abbey
Product Management
67
5Yasmin Sanders
Commercial Partnerships, AU
67
6Jay Fowdar
Product Delivery
62
7Sam Lenihan-Franks
MSCI
61
8Andy Shiels
Marketing
61
9Brian Morel
Marketing
61
10Jason Mussman
Commercial Partnerships
58

Projects Created

Knowledge bases authored · top 10
1Matthew Fenzl
Workplace
16
2Victoria Shrader
Product Delivery
15
3Tomasz Żukowski
Data Technology
14
4Enerico Miensky
Technical Services
14
5Sarah Davil
Account Management
12
6Michael Lazarus
Sales Operations
11
7Alyson Sprague
MSCI
11
8William Murray
Agency Solutions
10
9Alvir Navin
Executive
9
10Krzysztof Sankiewicz
PM · Data Licensing
9
Where the gaps are · among active users

The bigger gap is depth, not breadth.

Most active users have only tried Chat. The under-used surfaces are exactly the ones that drive the most leverage — Cowork for agentic workflows, Claude Code for engineers, and connectors that reach into real systems.

Surface Active users who haven't tried it % Why it matters
Cowork (agentic workflow) 157 of 284 55.3% The biggest leverage gap — over half of active users haven't tried agentic mode
Claude Code 115 of 284 40.5% Engineers who haven't moved from chat to CLI workflow
Connectors / MCP ~36 of 284 ~12.7% People who haven't plugged Claude into Gmail / Drive / Slack / etc.
Chat (Messages) 9 of 284 3.2% Almost every active user is at least chatting — the floor is solid

What this means: Adoption breadth (people using Claude at all) is strong at 84.5%. The real Q2 push is depth — getting the 284 active users to try Cowork and Connectors, not just keep chatting. That's where the 5x leverage lives.

Never-used · the 62

Employees who don't appear in the activity table at all.

62 names from the Samba employee directory don't appear in any Claude activity record. Listed below by department. Internal use only — please don't screenshot or share externally.

⚠ Data integrity caveat

35 of the 62 entries below have no department — meaning the directory match couldn't resolve. These may be: former employees still in the directory snapshot, people with mismatched email formats, or seats that were never actually assigned. The next pipeline pass will reconcile against the live Anthropic seat roster to disambiguate.

Practically: the real never-used number is likely closer to 27 (the 16 named departments below) than to 62. Treat the 62 as an upper bound, not a ground truth.

Unknown department — 35 names · directory ghosts (likely stale)

Alicja Hebel · Andrea Serbanescu · Anna Gramtchev · Artur Konopka · Bruce Quinn · Camilo Gisone · Chrystopher George · Conner McEuen · Craig Sennabaum · Desirè D'Alba · Elisa Sferruzza · Grace Miller · Irene Caroti · Jakub Szeląg · Justyna Tuczkowska · Kimberly Swift · Kirsteen Travers · Lara Fracasso · Leul Gebreselassie · Marianna Coppini · Mary Richardson · Maulik Shah · Michal Gutowski · Mirra Prudowska · Niccolo Presenti · Nitesh Narayan · Oleksandr Verkhovets · Orelle McNamee · Pascal Halim · Redic Thomas · Sitra Mohammed · Stefano Spinosi · Taylor Moore · Wayne Tsai · Wojciech Tarczyński

Data Technology (Semasio · Porto, Portugal) — 6

Ana Peres (Associate Technical Writer) · Hugo Silva (Senior Director, Engineering) · Joao Cardoso (Associate QA Engineer) · Joao Ramos Matos (Senior Software Engineer) · José Vilaça Da Silva (Lead Software Engineer) · Maria Baia (Associate QA Automation Engineer)

Agency Solutions UK (London) — 3

Dan Conneally (VP, UK Agency Solutions) · Madeleine Donnelly (Group Director) · Samantha Mullins (Principal Client Partner)

Agency Solutions (US — LA, Chicago) — 3

John Hoskins (Principal Client Partner, LA) · Mani Davari (Principal Client Partner, LA) · Stephanie Loesch (Senior Account Executive, Chicago)

Finance — 3

Conner Roos (VP, Finance, Remote US) · Davina Unger (Accountant, Semasio, Hamburg) · Jaden Lucero (Revenue Coordinator, SF)

Content ID — 2

Oleksandr Skalozubov (Software Engineer L5, SF) · Sylvia Li (Software Engineer L3, SF)

Executive · HR · Account Mgmt · Core Products · Data Science · TV Platform · Legal · Programmatic Consulting · Commercial Partnerships EU · Product Management — 1 each (10 total)

Executive: Amy Breaux (Executive Assistant, Remote US)
HR: Rafaela Carvalho (People Generalist, Semasio, Porto)
Account Mgmt: Daniela Cinque (Senior Account Manager, London)
Core Products: Irina Croce (Metadata Assistant)
Data Science: Weronika Gąsiorowska (Data Scientist L2, Remote Poland)
TV Platform: Bai Ruei Huang (Software Engineer L4, Taipei)
Legal: Timothy Li (Deputy General Counsel, NYC)
Programmatic Consulting: Aleksandra Moutik (Senior PAM, Hamburg)
Commercial Partnerships EU: Zac Pinkham (SVP, International, London)
Product Management: Alexander Quinn (Consultant)

Closing the gap

What we're actually doing about this

1. Reconcile the data first. 35 of 62 are unresolved directory matches. Pipeline fix is to ingest the live Anthropic seat roster and dedupe against the directory before relying on this number for any decision.

2. Regional coverage. Most named never-used are in EU (Porto, London, Hamburg, Poland) and APAC (Taipei). Champion-led regional office hours (Yasmin Sanders for APAC, Wave 2 EU recruit pending) is the targeted intervention. See Resources → Office Hours.

3. Semasio onboarding. The 6 Data Technology never-used are all at Semasio (acquired company, Porto). That's an integration question, not a champion question — coordinated separately with Bob van Toorn and Hugo Silva.

Methodology

How the efficiency formula works.

Each Claude action is weighted by the quality of usage, not raw volume. The hierarchy runs from basic chat (lowest weight) up through Projects, Connectors, Skills, and the agentic products — Cowork (for knowledge workers) and Claude Code (for engineers). A user spending $40/month using Cowork + Skills + Connectors scores higher than someone spending $100/month only chatting.

ActionTime-saved creditWhy
Conversation session (chat)10 minBasic chat usage — the floor
Web search5 minResearch acceleration on a single query
File upload (10+ pages)15 minDocument analysis
Use a Connector / MCP15 minTool integration — Claude reaches into a real system
Use a Skill20 minAutomated workflow — repeatable structured task
Use a Project30 minKnowledge base usage — Claude has context
Create a Project1 hourPersonal KB invested — long-term leverage
Cowork session30 minAgentic workflow for knowledge workers — Claude does multi-step work end-to-end
Claude Code session45 minFull engineering workflow — Claude builds, refactors, ships

A note on the agentic products: Cowork and Claude Code are Anthropic's agentic surfaces — same engine, different audience. Cowork is the knowledge-worker GUI for multi-step automated work; Claude Code is the engineer's CLI for full development workflows. Both score higher than chat because they reflect a different relationship with AI — delegating real work, not asking questions.

The score formula itself is intentionally not published — Madeline's call. Publishing it would create gaming incentives, where people optimize for the score rather than for smart usage. The leaderboards above show the output of the formula across individual surfaces.