Claude Signals
What leaders and teams rolling out Claude report, with every finding tied to its evidence.
Read the findingsWhat the evidence says
Practitioner sentiment, leaders and teams
+44 Positive from 28 sources that take a side
How this figure is madePositive overall signal, net 44 points
From 27 authors in 24 discussions. Small sample. Run a pilot on your team's recurring tasks to measure how often answers are used without rework.
- Analyzed
- 101 total sources
- Top praise
- Hand-off Tasks (7 of 7 sources) See praise evidence
- Top concern
- Missing Features (6 of 6 sources) See concern evidence
Jan 2025 to Oct 2026.
83 sources from practitioners, by month and sentiment.
The 30 sources that speak for a team or company lean positive. All 7 that discuss hand-off tasks praise them, connectors draw praise in 5 of 8 accounts, and 6 support enabling staff before a rollout. Data and admin controls come up in seven accounts, and 2 raise them as a problem, including a company of about 110 users whose accounts Anthropic suspended at once without warning.
Positive, mixed and negative sources
The score counts them as 14 positive, 11 mixed and 2 negative votes, one per author. See How the numbers work.
See the evidenceBelow: 2 praise themes, 3 concerns and 3 points of agreement.
Findings
Counted for Leaders and teams rolling it out, from 30 sources. Change group
What practitioners praise
Every capability practitioners describe positively, ranked by how many sources say so. Open one to read who said it.
Hand-off TasksAll 7 sources on it praise itHow well Claude completes multi-step work handed to it in Cowork, scheduled tasks, or Slack, and returns real files.
Counted for Leaders and teams rolling it out.
Positive
hungryhobbit, Hacker News commenter
Comment on: Dots: Always-on agentsHacker News comment
The commenter's team uses Claude Tag in Slack to rotate code review assignments and to remind reviewers who have not started within 24 hours.
Positive
Daniel Blum, product manager at Melio
How I AI: How this PM uses Claude to handle 70% to 80% of his workdayPodcast episode, How I AI (Lenny's Podcast Network)
A Melio product manager says Claude and Cowork now handle 70% to 80% of his workday after months of building context files. He finishes a week's work in one focused day, and he turned the setup into a 15-minute onboarding for other Melio employees.
Positive
michelb, Hacker News commenter at a learning and development company
Comment on: Manus will return to operating as an independent companyHacker News comment
Most of the 15-person company uses Cowork or ChatGPT Work almost full time for research and first drafts, sends output to the CMS over MCP, and edits everything before it reaches customers.
Positive
Charles Cook, VP Operations, PostHog
Making Claude Cowork actually usefulCompany blog post, PostHog
PostHog's VP of Operations gets most of his value from scheduled Cowork tasks, such as a weekly check of about fifty team plans for contradictions, which replace about 90 minutes of context-loading. Writing and Google Drive work still fall short.
Mixed
Claire Vo, founder of ChatPRD
What Claude Design is actually good for (and why Figma isn't dead yet)Podcast episode, How I AI (Lenny's Podcast Network)
The ChatPRD founder names marketing landing pages and slide decks as Claude Design's best uses and says Figma still wins in places. She hit the Claude Design limit mid-build and paid $200 to keep going.
Mixed
Staff writer at Every
Vibe Check: Claude Cowork Is Claude Code for the Rest of UsCompany blog post, Every
Every's testers find Cowork's asynchronous work promising but rough, with a learning curve for nontechnical users and missing web or mobile access.
Positive
Scott Monday, operator of Trinity Builder Solutions and kitchen & bath CRATE
Cowork Mondays: The Bet I'm Making in 2026Newsletter, Substack
The operator moved most office staff at two construction companies onto Cowork. After about three weeks of setup, an estimator's two-hour daily bid intake now arrives as an overnight one-page summary.
Connectors & Integrations5 of 8 sources on it praise itHow well Claude reaches the author's tools: connectors, Slack and Claude Tag, Claude in Chrome, and the office add-ins.
Counted for Leaders and teams rolling it out.
Positive, 5
Positive
primitivesuave, Hacker News commenter
Comment on: Manus will return to operating as an independent companyHacker News comment
Nontechnical teammates use Cowork extensively through GUI-configured connectors backed by audited internal MCP access.
Mixed
nsiemsen, Hacker News commenter working in finance
Comment on: ChatGPT for ExcelHacker News comment
Claude for Excel was a main reason their finance company bought Enterprise licenses; it does junior-level spreadsheet work, but heavy users cost about US$100 a day.
Positive
hasyimibhar, Hacker News commenter
Comment on: I still prefer MCP over skillsHacker News comment
A six-person team uses Claude Desktop with Project instructions and audited MCP access for operations and analytics.
Positive
Reid Robinson, Lead Product Manager at Zapier
How this PM uses MCPs to automate his meeting prep, CRM updates, and customer feedback synthesisPodcast episode, How I AI (Lenny's Podcast Network)
Zapier's principal AI product strategist connects Claude to Zapier's MCP server to automate CRM updates and meeting prep, and keeps tool instructions in a Claude Project to make tool use more reliable. He hands file-based data to Gemini.
Positive
michelb, Hacker News commenter at a learning and development company
Comment on: Manus will return to operating as an independent companyHacker News comment
Most of the 15-person company uses Cowork or ChatGPT Work almost full time for research and first drafts, sends output to the CMS over MCP, and edits everything before it reaches customers.
Mixed or negative, 3
No opinion stated
OvercastJeff, GitHub user
Account-scoped auth failure: Gmail and Calendar return "The caller does not have permission" since April 20GitHub issue, claude-ai-mcp
For one account, every Gmail and Google Calendar connector call failed with a permission error for 7.5 weeks from April 20, 2026. Colleagues on the same Workspace tenant were unaffected, and reopen requests went unanswered.
No opinion stated
bbgit123, GitHub user
Salesforce Hosted MCP: "Authorization with the MCP server failed" on production Enterprise orgGitHub issue, claude-ai-mcp
Connecting Claude's custom connector to a production Salesforce org completed OAuth, then failed with an authorization error that Anthropic support acknowledged as an Anthropic-side issue.
Positive
Charles Cook, VP Operations, PostHog
Making Claude Cowork actually usefulCompany blog post, PostHog
PostHog's VP of Operations gets most of his value from scheduled Cowork tasks, such as a weekly check of about fifty team plans for contradictions, which replace about 90 minutes of context-loading. Writing and Google Drive work still fall short.
What concerns them
Every concern practitioners raise, most raised first.
Missing FeaturesAll 6 sources on it raise it as a concernCapabilities the author needed and did not find, or found only on another plan or product.
Counted for Leaders and teams rolling it out.
Mixed
anoop_kumar, Hacker News commenter and startup founder
Comment on: Ask HN: Sub 5 person team - Claude team plan?Hacker News comment
After Anthropic cut the Team minimum to two seats, the founder notes that spend limits, role-based access, and audit logs still sit only in Enterprise.
Mixed
lalithaar, Hacker News commenter
Comment on: Ask HN: Sub 5 person team - Claude team plan?Hacker News comment
Claude Tag comes only with the Team plan, so a team under the five-seat minimum considered sharing a joint account to get it.
Mixed
Staff writer at Every
Vibe Check: Claude Cowork Is Claude Code for the Rest of UsCompany blog post, Every
Every's testers find Cowork's asynchronous work promising but rough, with a learning curve for nontechnical users and missing web or mobile access.
Positive
Katie Parrott, staff writer at Every
Vibe Check: Claude Skills Need a 'Share' ButtonNewsletter, Every
An Every staff writer built six editorial Skills within hours, including one that enforces Every's style guide, and uses them every day. She found that sharing Skills across a team needs features Skills lack.
Mixed
eeemmmooo, Hacker News commenter leading an AI rollout
Ask HN: AI rollout for 100 employees, mostly non-technical - what worked?Hacker News comment
Planning a rollout to 100 mostly non-technical staff, the author finds Claude Enterprise alone lacks the guardrails and shared workspaces needed to see and reuse what people build.
Mixed
Claire Vo, founder of ChatPRD
What Claude Design is actually good for (and why Figma isn't dead yet)Podcast episode, How I AI (Lenny's Podcast Network)
The ChatPRD founder names marketing landing pages and slide decks as Claude Design's best uses and says Figma still wins in places. She hit the Claude Design limit mid-build and paid $200 to keep going.
GILD's fact checks, which don't count as sources
Checked Oct 12, 2026.Claude does not generate photos or illustrations; it builds diagrams, charts, and interactive visuals in HTML and SVG. Can Claude produce images?
Open them in the recordUsage Limits & Cost5 of 7 sources on it raise it as a concernSession and weekly limits, seat prices, usage billing, and the cost of choosing a plan.
Counted for Leaders and teams rolling it out.
Raised as a concern, 5
Mixed
Claire Vo, founder of ChatPRD
What Claude Design is actually good for (and why Figma isn't dead yet)Podcast episode, How I AI (Lenny's Podcast Network)
The ChatPRD founder names marketing landing pages and slide decks as Claude Design's best uses and says Figma still wins in places. She hit the Claude Design limit mid-build and paid $200 to keep going.
Mixed
nsiemsen, Hacker News commenter working in finance
Comment on: ChatGPT for ExcelHacker News comment
Claude for Excel was a main reason their finance company bought Enterprise licenses; it does junior-level spreadsheet work, but heavy users cost about US$100 a day.
Mixed
xlos21, Hacker News commenter
Comment on: Show HN: Claude Tuner - Monitor your Claude usage and find the right planHacker News comment
A developer whose company uses Claude heavily built a usage tracker because limits were opaque, and the company skips buying Team and assigns individual Max plans by measured usage.
Positive
AI at Yale, Yale University
Now Available: Claude Enterprise Licenses at YaleUniversity IT announcement, ai.yale.edu
Yale offers Claude Enterprise at US$20 or US$50 a month with Chat, Code and Cowork, plus a HIPAA environment for ePHI, and notes that Anthropic publishes no usage limits for the licences.
"*Anthropic does not publicly disclose usage limits for licenses. Limits are enforced via a rolling 5-hour window along with a weekly usage cap that resets at a fixed time specific to your account."
Excerpt, Yale University, AI at Yale Positive
Craig Iskowitz, Founder and CEO, Ezra Group
Why We Switched to Claude from ChatGPTCompany blog post, WealthTech Today (Ezra Group)
Ezra Group found most staff working in Claude while its company-wide ChatGPT license sat idle. After polling the team, it made Claude its primary platform for better-formatted documents and fewer reprompts, and kept ChatGPT for formal writing and for lighter use where plan cost matters.
Discussed without concern, 2
Positive
leemoore, Hacker News commenter
Comment on: Gemini 3 Flash: Frontier intelligence built for speedHacker News comment
At the commenter's Fortune 100 financial company, Claude beat OpenAI in an enterprise-wide evaluation, and each engineer gets US$1,000 a month in Claude Enterprise and Claude Code tokens.
Mixed
lalithaar, Hacker News commenter
Comment on: Ask HN: Sub 5 person team - Claude team plan?Hacker News comment
Claude Tag comes only with the Team plan, so a team under the five-seat minimum considered sharing a joint account to get it.
GILD's fact checks, which don't count as sources
Checked Oct 12, 2026.Paid seats have a five-hour session limit and weekly limits shared across chat, Cowork, and Claude Code. Usage credits continue past them at API rates, and Enterprise seats have no per-seat limits because all usage is billed at API rates. How do usage and length limits work?
Open them in the recordData & Admin Controls2 of 7 sources on it raise it as a concernTraining on chats, retention, access to company data, and the admin controls a rollout needs.
Counted for Leaders and teams rolling it out.
Raised as a concern, 2
Mixed
eeemmmooo, Hacker News commenter leading an AI rollout
Ask HN: AI rollout for 100 employees, mostly non-technical - what worked?Hacker News comment
Planning a rollout to 100 mostly non-technical staff, the author finds Claude Enterprise alone lacks the guardrails and shared workspaces needed to see and reuse what people build.
Negative
alpinisme, Hacker News commenter at an agricultural technology company
Anthropic bans orgs without warningHacker News post
Every Claude account in the company was suspended without warning, admins included; the appeal form went unanswered for days while the API account kept billing.
Discussed without concern, 5
Positive
primitivesuave, Hacker News commenter
Comment on: Manus will return to operating as an independent companyHacker News comment
Nontechnical teammates use Cowork extensively through GUI-configured connectors backed by audited internal MCP access.
Mixed
mentorcliq, Hacker News commenter administering an AI rollout
Comment on: Ask HN: AI rollout for 100 employees, mostly non-technical - what worked?Hacker News comment
After moving from ChatGPT toward Claude, the admin keeps staff on the Team account for log access, but notes people use it sparingly and keep personal accounts, so usage needs incentives.
Positive
hasyimibhar, Hacker News commenter
Comment on: I still prefer MCP over skillsHacker News comment
A six-person team uses Claude Desktop with Project instructions and audited MCP access for operations and analytics.
Positive
AI at Yale, Yale University
Now Available: Claude Enterprise Licenses at YaleUniversity IT announcement, ai.yale.edu
Yale offers Claude Enterprise at US$20 or US$50 a month with Chat, Code and Cowork, plus a HIPAA environment for ePHI, and notes that Anthropic publishes no usage limits for the licences.
"*Anthropic does not publicly disclose usage limits for licenses. Limits are enforced via a rolling 5-hour window along with a weekly usage cap that resets at a fixed time specific to your account."
Excerpt, Yale University, AI at Yale Mixed
anoop_kumar, Hacker News commenter and startup founder
Comment on: Ask HN: Sub 5 person team - Claude team plan?Hacker News comment
After Anthropic cut the Team minimum to two seats, the founder notes that spend limits, role-based access, and audit logs still sit only in Enterprise.
GILD's fact checks, which don't count as sources
Checked Oct 12, 2026.Content from commercial products, including Team and Enterprise, is not used for training by default. Free, Pro, and Max chats are used when the user allows it. Is my data used for model training? (commercial products)
Open them in the record
Where the evidence agrees
Statements this group's sources agree on, then the ones close to agreement.
- Recurring point
Instructions, Projects, Skills, and reference files written once make results more consistent than prompting from scratch each time.
6 supporting sources - Recurring point
Many users keep ChatGPT, Gemini, or Copilot alongside Claude, switching for limits, features, or fit to the task.
6 supporting sources - Recurring point
Admins planning rollouts want shared prompts, usage rules, and reasons for staff to move off personal accounts onto the company one.
6 supporting sources
Close to agreement: each needs 6 supporting sources
- Below threshold
Users report large time savings on specific recurring tasks such as decks, reports, briefings, and document reviews.
5 supporting sources
Where it disagrees
Questions the sources split on, with the count on each side.
For leaders and teams rolling it out, no question reaches 5 sources from 3 authors, with 2 on each side.
The evidence
Whose experience this is
The three groups, then every practitioner by the context they work in, and how emphasis changes with scale.
Three groups drawn from the same sources. Choose a group to select whose experience counts. Within that group, each author counts once in the practitioner sentiment, with their latest view. Signals open on Leaders and teams rolling it out.
Leaders and teams rolling it out
30 sources. These accounts focus on seat cost, data and admin controls, connectors to company tools, and how teams were trained, from the accounts of teams, departments, and companies rolling Claude out.
Individual users
53 sources. These accounts focus on hand-off tasks, connectors, Projects and memory, and day-to-day reliability for people using Claude for their own work.
All users
83 sources. This gives the broadest read, but combines individual use with team and company rollouts that can reach different conclusions.
83 sources from practitioners, by the context they work in. Filled rows count for this group.
- Individual userOne person using it for their own work53 sources
Emphasizes Hand-off tasks, connectors, and reliability
- Small team2 to 10 people sharing it6 sources, counted here
Emphasizes Usage limits, plan choice, and shared context
- DepartmentA department, or a named company's team12 sources, counted here
Emphasizes Governance, shared Projects and Skills, and seat cost
- IT & leadershipIT and business leaders deciding the rollout5 sources, counted here
Emphasizes Scheduled hand-off tasks, memory and usage limits, and admin access to logs
- Organization-wideCompany-wide rollout7 sources, counted here
Emphasizes Standardizing on Claude, account suspensions, and Enterprise usage cost
What changes with scale. Individual users focus on tasks and connected tools. Team and company accounts focus on admin controls, plan minimums, and account suspensions. Company-wide accounts report self-measured gains, from a CBTS pilot to an estimator whose daily bid intake now arrives as an overnight Cowork summary.
Where the sources come from
Where this group's sources were published.
Counted for Leaders and teams rolling it out, from 30 sources. Change group
48 sources for this group, 18 of them documentation
- Product Documentation18, 38% of 48 sources
- Online Communities16, 33% of 48 sources
- Company Blogs & Reviews8, 17% of 48 sources
- Podcasts, Talks & Videos6, 13% of 48 sources
Older evidence is retained when it still applies. Sources that depend on a feature that has changed since, or on a plan rule that no longer holds, stay in the history, marked Needs a recheck.
See sources by originThe record
Who each source is, how it knows, whether it still applies and what it counts toward.
Newest first, on its own page. Every count above opens its sources there.
Read all 101 sourcesAbout this read
What changed, and how we count
What changed
- Nov 1, 2026: since Oct 30 the page prints one vote per author, and a figure given on the weighted reading is the second view. Leaders and teams read +44, individual users +30 and all users +35.
- Uses public sources only. Every source links to the page it comes from and names its author and publication date.
- Shows a finding only when at least 5 sources from 3 different authors discuss it, and a disagreement only with at least 2 sources on each side.
- Merges usage limits and plan cost into one Usage & Cost concern, since users raise them together. Drops the question of which assistant is primary, which has a single source on one side.
- Checks every documentation page against Anthropic's current plan, Help Center, and Privacy Center pages.
- Keeps only English-language sources dated January 1, 2025 or later, and adds Hacker News accounts of team rollouts found on October 12, 2026.
- Oct 25, 2026: added Platform Gaps from what existing accounts state: 5 raise devices or tools Claude does not reach, and 1 describes a missing connector without concern.
- Oct 25, 2026: added Value for Money, drawn from the same accounts: 2 say the Pro plan is worth it and 2 say Team seats or Enterprise usage cost too much.
- Oct 25, 2026: adds 6 company blog, podcast, and GitHub sources, including a company-wide Claude Enterprise rollout with self-reported pilot results.
- Oct 26, 2026: adds 8 company blog, newsletter, and podcast accounts from named leaders and teams at Ezra Group, PostHog, Every, Zapier, Tenex, ChatPRD, and two construction companies. They are archived the same way, and 7 name a role stated on a first-party page.
- Oct 28, 2026: adds 8 sources from GILD's research ledger, each with the passage it quotes: Yale's Claude Enterprise launch beside Microsoft 365 Copilot, podcast accounts from Brex, Tiny, Every and a Meta product manager, Till Freitag's switch from ChatGPT, a user-research consultant's 100 repeated analyses and Ethan Mollick's test of Cowork against ChatGPT Work. On the weighted reading, leaders move from +54 to +51, individual users from +30 to +29 and all practitioners from +41 to +39.
- Oct 30, 2026: a second-model check removed Till Freitag's switch post, which markets the consultancy's AI services, and the Ezra Group, Till Freitag and CBTS insight cards for the same reason. It corrected a Meta product manager's account to say only that he opened a Claude project when Meta approached him about a role, and Yale's card and comparison account to what its launch page says. On the weighted reading, leaders and teams move from 31 to 30 sources and from +51 to +50; all practitioners from 84 to 83 and +40 to +39.
Method
A running review of what people using Claude at work report, from individual users to company-wide rollouts. Each finding links to its sources, and each conflict shows both sides.
Counts cover only the sources GILD reviewed.
Who counts
- Leaders and teams rolling it out
- Small team, Department, IT & leadership, Organization-wide
- Individual users
- Individual user
- All users
- Everyone in the leaders and teams and individual users groups
Choose a group to select whose experience counts. Within that group, each author counts once in the practitioner sentiment, with their latest view. GILD files each source in a group from how the account describes its author. A role or employer marked self-described is what the source states about themselves. Documentation checks facts for every group and never counts toward sentiment.
How the numbers work
- Overall signal. Positive votes minus negative ones, per 100 votes from the sources that take a side. Each author counts once, with their latest view, whoever they are and wherever they posted. The older weighted reading, where accounts GILD matched to a named person count three times when GILD lists the author among IT and business leaders and twice otherwise, is shown beside it as a second view. Where an author wrote more than one source, the line beside the split gives the votes, one per author, so the score follows from it. When one author's sources from the same day disagree, that author counts as mixed, and a half point rounds away from zero. Sources with no opinion are listed and not counted. Strong positive is +50 or higher, positive +20 to +49, mixed −19 to +19, negative −20 to −49, strong negative −50 or lower.
- Weighted reading, second view. Counted the older way, where an account GILD matched to a named person counts two or three times, leaders and teams read +50 strong positive (method sentiment/S1). GILD printed that number as "GILD score" until October 29, 2026; most of those matches rest on a name or a self-described role.
- Method change, Oct 30, 2026. GILD changed how practitioner sentiment counts, to one vote per author. Weighted, it read +54 Strong positive; with one vote per author it reads +46 Positive. No evidence changed.
- Unscored. A group gets a score only with 20 sources from 15 authors. Below that, one source moves the score about 5 points, so the read says "Unscored until 20 sources from 15 authors" and shows its counts.
- Independence. A source whose author works for the vendor, for one of its partners or programs, or for a competitor, and a page the vendor published or ran, is listed with that tie and counts toward nothing. One author counts on at most two records per product. The line under the score says how many authors and how many separate discussions the counted sources come from: comments in one thread are one discussion.
- Praise and concerns. How many of the sources that discuss a theme praise it or raise it.
- Consensus. A statement joins when 20% of the group's sources support it and fewer contradict it. Contradicting sources are counted separately. Below that line a statement shows as building. At 40% or more it reads Strong agreement; below, Recurring point.
- Disagreements. The count on each side. GILD names no winner and gives no share of all users.
- Minimum evidence. A finding needs 5 sources from 3 authors; a disagreement also needs 2 on each side. Smaller counts stay in the table below, marked under the minimum.
Labels on a source
A source's date and whether it still applies are separate: a recent post can describe an old version, and an old one can still hold.
- Documented
- Stated in official documentation.
- Reported
- A practitioner's account of their own use.
- Measured
- Backed by the source's own measurements, such as a comparative test or tracked team metrics.
- GILD analysis
- GILD's check of a claim against official documentation.
- Current
- Describes the product as it is now.
- Still applicable
- Older, and the behavior or lesson still holds.
- Needs a recheck
- May depend on a plan, version or setting that changed since it was written. It keeps counting until GILD rechecks it, since the rest of the account usually holds, and its card says so.
- Out of date
- No longer accurate. Kept in the record; counts toward nothing current.
Flagged for a recheck, of the sources that take a side: leaders and teams 4 of 28, individual users 3 of 50, all users 7 of 78.
Limits
- These figures describe reviewed public sources; they do not represent all Claude customers. 51 of the 83 sources are Hacker News comments and posts found through its public search API. The other 32 are 19 English blogs, newsletters, and reviews, a university IT announcement, 4 GitHub issues, and 8 podcast episodes with transcripts or show notes. 30 sources speak for a team or company. Reddit was not reachable for this review.
- Sources name their plan, model, and feature inconsistently; each source shows what it states. Sources that depend on a feature that has since changed, such as Cowork before it merged into one Claude or the Team plan's former five-seat minimum, are marked Needs a recheck.
Every count, and the sources behind it
20 findings for 3 groups
Each row opens its sources for the group you are reading.
| Finding | Leaders and teams rolling it out | Individual users | All users | Sources |
|---|---|---|---|---|
| Current readCurrent sentiment | 15 positive, 11 mixed, 2 negative | 25 positive, 15 mixed, 10 negative | 40 positive, 26 mixed, 12 negative | Sources |
| PraiseWriting Quality | 2 of 3 positive (under the minimum) | 2 of 4 positive (under the minimum) | 4 of 7 positive | Sources |
| PraiseAnalysis & Research | 4 of 4 positive (under the minimum) | 7 of 10 positive | 11 of 14 positive | Sources |
| PraiseProjects & Memory | 2 of 4 positive (under the minimum) | 6 of 10 positive | 8 of 14 positive | Sources |
| PraiseConnectors & Integrations | 5 of 8 positive | 10 of 15 positive | 15 of 23 positive | Sources |
| PraiseHand-off Tasks | 7 of 7 positive | 15 of 16 positive | 22 of 23 positive | Sources |
| PraiseValue for Money | 1 of 3 positive (under the minimum) | 2 of 2 positive (under the minimum) | 3 of 5 positive | Sources |
| ConcernUsage Limits & Cost | 5 of 7 raise it | 4 of 7 raise it | 9 of 14 raise it | Sources |
| ConcernAccuracy | 1 of 1 raise it (under the minimum) | 3 of 4 raise it (under the minimum) | 4 of 5 raise it | Sources |
| ConcernData & Admin Controls | 2 of 7 raise it | 3 of 4 raise it (under the minimum) | 5 of 11 raise it | Sources |
| ConcernReliability | 4 of 4 raise it (under the minimum) | 9 of 9 raise it | 13 of 13 raise it | Sources |
| ConcernMissing Features | 6 of 6 raise it | 7 of 7 raise it | 13 of 13 raise it | Sources |
| ConcernPlatform Gaps | 1 of 1 raise it (under the minimum) | 5 of 6 raise it | 6 of 7 raise it | Sources |
| ConsensusContext set up front pays off | 6 supporting | 2 supporting (under the minimum) | 8 supporting (below threshold) | Sources |
| ConsensusPeople check the output | 2 supporting (under the minimum) | 4 supporting (under the minimum) | 6 supporting (below threshold) | Sources |
| ConsensusAnother assistant stays in reach | 6 supporting | 12 supporting | 18 supporting | Sources |
| ConsensusRollouts need enablement | 6 supporting | 0 supporting (under the minimum) | 6 supporting (below threshold) | Sources |
| ConsensusTime saved on recurring tasks | 5 supporting (below threshold) | 6 supporting (below threshold) | 11 supporting (below threshold) | Sources |
| DisagreementShould people hand Claude whole tasks, or keep it to drafting and questions? | Hand off whole tasks 3, Keep it to drafting and Q&A 0 (under the minimum) | Hand off whole tasks 10, Keep it to drafting and Q&A 2 | Hand off whole tasks 13, Keep it to drafting and Q&A 2 | Sources |
| Current readEvidence mix | 48 sources | 71 sources | 101 sources | Sources |
Fact checks
GILD's checks of claims against official documentation, kept apart from the source counts.
- Usage Limits & Cost. Paid seats have a five-hour session limit and weekly limits shared across chat, Cowork, and Claude Code. Usage credits continue past them at API rates, and Enterprise seats have no per-seat limits because all usage is billed at API rates. How do usage and length limits work?, checked Oct 12, 2026
- Accuracy. Anthropic's Cowork safety guide recommends human oversight in high-stakes tasks because Claude can make mistakes, and treats tools that change things differently from tools that only read. Use Claude Cowork safely, checked Oct 12, 2026
- Data & Admin Controls. Content from commercial products, including Team and Enterprise, is not used for training by default. Free, Pro, and Max chats are used when the user allows it. Is my data used for model training? (commercial products), checked Oct 12, 2026
- Reliability. Cowork sessions in the cloud (beta) let scheduled tasks run with no device online. They are on by default on Team and off by default on Enterprise, where an owner grants them by role. Use Claude Cowork on Team and Enterprise plans, checked Oct 12, 2026
- Missing Features. Claude does not generate photos or illustrations; it builds diagrams, charts, and interactive visuals in HTML and SVG. Can Claude produce images?, checked Oct 12, 2026