Kinetry Privacy Policy
Effective: July 20, 2026. Last updated: October 31, 2026 (Section 16: one connection can read several of a person's workspaces, ticked when connecting, one workspace at a time, and every answer names the workspace it came from). Earlier updates: October 30, 2026 (Section 16: Claude can read each person's job title and who they report to once the workspace confirms it; a refresh token presented again within 30 seconds is treated as a retry; the Connected assistants page shows why a connection ended); October 28, 2026 (Section 16: on the Team plan and higher, Claude can read a pair reading of how two people work together, and the team relationship map); October 26, 2026 (Section 16: in advisor mode, the advisor can save ratings they confirm in Claude, if the workspace turns it on; Section 6 updated to match); October 22, 2026 (Section 16: Claude can read earlier cycles the plan includes, and group reports; Sections 2 and 3 updated to match); October 22, 2026 (Section 16, connecting your workspace to an AI assistant, added, and Sections 1, 2, 3, 6, 7, 8 and 11 updated to match); October 22, 2026 (Section 8: what can remain after an erasure); October 21, 2026 (Section 15, public tools for AI assistants, added).
This policy describes how Kalgren Consulting LLC, doing business as Kinetry ("Kinetry," "Company," "we," or "us"), collects, uses, discloses, retains, handles and protects personal information in connection with the Kinetry service. It is a companion to our Terms of Service. This Privacy Policy applies to visitors to the website, customers who use this Service, employees, contractors or other individuals who voluntarily participate in employer or work-related workplace development coaching type programs, anyone who uses Kinetry's public tools through an AI assistant (see Section 15), or any other visitor to this Site. This Privacy Policy works in conjunction with any master service agreement between Company and any employer or other customers.
The Services provide employers or other entities with behavior-based coaching, workplace culture assessments, workplace surveys, professional development insights, and related analytics. Through the Services, employees (or other authorized participants) may voluntarily provide information regarding workplace behaviors, preferences, experiences, skills, goals, and professional development opportunities. We use technology, including artificial intelligence ("AI") tools, to analyze submitted information and generate coaching insights, skill-gap analyses, and recommendations designed to support workplace development.
1. The two roles we play
- For Customer workspace data (rosters, assessment responses, scores, notes): the Customer organization decides whether to offer this Service, which workers may participate, what surveys, assessments, or coaching activities are offered and what related data is collected and why; Kinetry processes it on the Customer's behalf to provide the Service. Individuals should direct questions about their organization's use of Kinetry — including access or deletion requests about assessment data — to their employer or entity with whom they work. The Customer acts as the business, controller or other similar decision-making entity under applicable privacy laws with regard to participant provided information and what to do with such information. Kinetry is the service provider, and processes such information according to the Customer's instructions and applicable agreements. If the Customer turns on AI assistant access (Section 16), the Customer also instructs Kinetry to let each connected person's AI assistant read what that person's role already shows them. The assistant's provider receives that data under its own terms with that person or the Customer, and is not Kinetry's service provider for it. The exception is when Kinetry's own staff connect as a workspace's advisor, which Section 16 describes. Kinetry treats its records of those connections and requests as Customer workspace data.
- For account, site and tool-usage data (sign-in email, acceptance records, support messages, service logs, and the tool usage records described in Sections 15 and 16): Kinetry determines the purposes and is responsible directly; no Customer is involved when you use the public tools.
2. What we collect
Customers may provide to Company information about participants such as: your organization (name, work email, role, manager); business contact information; department or team information; job related duties or other job related information; participation status; administrative information necessary to operate the Services; any assessment responses and notes submitted by raters; computed scores and AI-generated coaching content; sign-in and usage records (timestamps, IP address, device/browser type); Terms-acceptance records (who accepted which version, when); if the Customer turns on AI assistant access, records of assistant connections and of the requests made through them (Section 16); and support communications.
Further, employee or participant participation is at the direction of the Customer and is voluntary unless otherwise directed by the Customer. Before participating, the Customer should provide to participants:
- The purpose of the survey or assessment;
- The type of information sought;
- How the responses may be analyzed;
- Whether individual, identifiable information may be available within the Customer organization and to whom;
- How AI technology is to be or could be used, including, if the Customer turns on AI assistant access, that people in the roles it allows can have an AI assistant read individual results and group summaries they can see, for the latest cycle or earlier cycles the Customer's plan includes (Section 16);
- Any privacy rights available to participants.
Before participating, participants must acknowledge and agree that:
- their responses will be processed through the Services;
- AI-based analysis is an integral component of the Services;
- submitted information may be analyzed to generate coaching insights and recommendations;
- the Customer may have access to individual-level results depending on the Customer configuration and applicable agreements.
Participants who do not wish to have their information processed through the Services must decline participation or discontinue participation as permitted by the Customer's program configuration. Because AI analysis is an essential feature of the Services, participants who submit responses cannot separately opt out of AI processing while continuing to use the Services.
Participants may be asked to voluntarily provide information about their work and skill sets to allow Services to be completed. Such information may include:
- survey responses;
- workplace experiences;
- behavioral preferences;
- communication styles;
- professional goals;
- perceived strengths;
- development opportunities;
- skills assessments;
- coaching inputs;
- feedback responses.
Participants should avoid submitting information that is unrelated to the purpose of the Services, including sensitive personal information unless specifically requested and authorized by the employer or entity with whom they work. At no time should participants provide any financial account information, religious beliefs, political opinions, union membership or affiliations, or health related or health identifying information including disability information, genetic information, biometric information, or information concerning sexual orientation or sexual activity.
Kinetry, through its systems, may automatically collect information when individuals access or use the Services including:
- IP address;
- browser type;
- device information;
- operating system;
- usage activity;
- log information;
- authentication information;
- cookies and similar technologies.
Note: Kinetry does not collect payment card numbers (payments are processed by our payment provider) and does not intentionally collect data about minors.
3. Customer Access/Use of Information
What a Customer may access depends on its assessment mode, its configuration, and its agreements with Company.
- In multi-rater (360) mode: Customers may access individual participant results (scores and participant-level coaching insights), aggregated organizational reports, benchmarking information, skill-gap analyses, and coaching recommendations. Customers do not have access to the individual assessment responses another rater submits: responses are confidential to the rater, results are computed from combined ratings, and statistical suppression hides results computed from fewer than the minimum number of raters, as described in our Terms of Service.
- In advisor (single-assessor) mode: the Customer's designated advisor authors the assessments, so the advisor and the Customer's administrators may access individual assessment entries and observation notes together with the computed results, participant-level coaching insights, aggregated organizational reports, benchmarking information, skill-gap analyses, and coaching recommendations.
- AI assistant access records. If the Customer turns on AI assistant access (Section 16), its administrators can see the record of requests made through connected assistants: who asked, which tool, when, which cycle, and, for a request about one person, who it was about. The full record is included in the workspace export.
Customers are responsible for determining appropriate access controls, providing appropriate notices and consents to participants, establishing lawful purposes for processing and using participant information, determining who in their organization may appropriately access participant information, and creating and maintaining appropriate security safeguards and confidentiality to ensure that Customer use of participant information complies with applicable laws.
4. How Kinetry uses information
We use information collected to operate the Services (collect assessments, compute scores, generate coaching content, deliver reports); to secure it (authentication, abuse prevention, logging); to support you; to send service emails (sign-in links, cycle notifications, results — not marketing without consent); and to create aggregated, de-identified statistics that cannot reasonably identify any person or organization, which we use for cross-customer benchmarks, service improvement, and research, as permitted by our Terms of Service. We do not use identifiable participant information for any purpose other than to provide Services, nor do we use individual survey responses or other confidential information to train large language models or general-purpose AI models. We may use aggregated or appropriately de-identified information to improve the Services, develop analytic capabilities, improve system performance and enhance user experiences.
Should Company use third-party service providers to assist with operating, maintaining, securing or improving the Services, information may be disclosed to vendors who provide services such as:
- cloud hosting;
- database services;
- cybersecurity services;
- authentication services;
- analytics services;
- customer support services;
- AI infrastructure services;
- communications services.
Company requires vendors to agree to:
- process information only for authorized purposes;
- maintain appropriate confidentiality obligations;
- implement reasonable security measures;
- comply with applicable privacy requirements.
Company will not authorize service providers to sell personal information or use personal information for independent advertising purposes.
We may also be legally required to disclose personal information processed as a part of the Services to comply with applicable laws, in response to lawful governmental requests, to investigate fraud or other security incidents, or to protect individuals or the public from a reasonable threat. We may also use such information as a part of a corporate transaction, and any successor organization will be required to also maintain appropriate protections for such personal information.
5. What we don't do
We do not sell personal information. We do not use Customer workspace data to advertise to anyone. We do not make automated employment decisions — the Services are a development tool, and our Terms prohibit Customers from using it as the sole basis for employment decisions.
6. Who can see assessment data
Visibility follows the Customer's configuration and roles as they determine best (for example, managers see their team's results; executives see organization-level results). Your organization's administrators control these roles and org structure, may grant specific people broader visibility, and may temporarily view the service as one of their own Authorized Users (read-only) to provide support and verify access, and each "view as" action is recorded. If your organization's administrators turn on AI assistant access, people in the roles they choose can also have an AI assistant they connect read what their own role shows them, and those requests are recorded (Section 16). In advisor mode, where Kinetry offers it, they can also let the advisor save ratings from that assistant, and those saves are recorded too (Section 16). Statistical suppression hides scores computed from fewer than the minimum number of raters. Authorized Kinetry personnel, including Kinetry's principal consultant, can access workspace data to operate, support, and improve the service and to deliver contracted advisory services, under confidentiality obligations. Service providers (hosting, email delivery, payments) process data for us under contract. We disclose data if required by law, with notice to the Customer where lawful and practicable.
7. Security
Except as otherwise provided in this Policy, Company agrees to: encrypt data in transit; use single-use, expiring sign-in links (no passwords stored); protect AI assistant connections with one-hour signed access tokens and with refresh tokens that Kinetry stores as one-way hashes (Section 16); enforce role- and tenant-scoped access on every data path; protect scoring internals server-side; and limit access to personnel who need it. No system is perfectly secure; we will notify affected Customers of a breach as required by law.
8. Retention and deletion
Workspace data is retained while the Customer's account is active and for the export window after termination described in the Terms, then deleted from production systems within a commercially reasonable period. Aggregated, de-identified data may be retained thereafter. Acceptance and billing records are kept as long as required for legal purposes.
How this works in practice:
- Export. Workspace administrators can download the workspace's complete data and results at any time (Admin → Company settings → Export workspace data), and a per-person export for any individual (Admin → People → Export data). Exports respect the rater-confidentiality protections described in the Terms — in multi-rater mode they never include assessment responses authored by other raters, and a person's own raw self-assessment entries are included only when that person downloads the export themselves.
- Person deletion. Workspace administrators can permanently erase a removed person and everything recorded about them (Admin → People → Erase permanently). Erasure is immediate in production systems, apart from the residual copies described below, and cannot be undone. Because re-scoring a past cycle from the remaining responses could expose individual assessors, past cycles in which the erased person assessed others are locked and their already-published results stand as-is; other results may later be recalculated without the erased person's data.
- Workspace deletion. Customers can request deletion of their entire workspace by contacting privacy@kinetry.ai; we perform it after any active subscription is cancelled.
- Residual copies. Erased data may persist briefly in encrypted database backups (which age out within about 30 days) and in our error-tracking service (up to 90 days). Our internal operations log, which records automated job runs and email-delivery problems, can name an erased workspace or person in an entry's text; those entries are deleted automatically after about 180 days. Content sent to our AI infrastructure provider (Anthropic, in the United States) to generate coaching output, limited to participant names, computed scores and trends, and advisor and rater notes, is retained by that provider for up to 30 days for trust-and-safety purposes and then deleted; that provider does not use it to train its models. Kinetry keeps no copy of what it sends; the coaching output that comes back is stored with the workspace's results and is deleted with the person or workspace it is about. Text that other people wrote, such as their own notes or messages, can still mention an erased person by name. Records that the workspace or another party still relies on, such as consent and audit records, referral rewards, or a share link made for a client, are kept with the erased person's name, email and account id removed. A minimal record that an erasure was performed (who was erased, who carried it out, when, and the terms-acceptance evidence noted above) is retained for legal purposes.
- Copies outside Kinetry. Deletion and erasure can't reach copies people have already taken out of Kinetry, such as a downloaded export or what a connected AI assistant has already read into a conversation (Section 16).
9. Your Choices; Do Not Sell or Share; How to Opt Out
We do not sell or share your personal information. Kinetry does not sell personal information for money or other valuable consideration, and does not share personal information for cross-context behavioral advertising (sometimes called "targeted advertising"), as those terms are used under the California Consumer Privacy Act (CCPA/CPRA) and similar state laws. We set no advertising or marketing trackers and use no third-party advertising cookies or ad pixels, so there is no sale or sharing of your personal information to opt out of. Because we do not sell or share, we do not display a "Do Not Sell or Share My Personal Information" opt-out link. If that ever changes, we will provide such a link and will honor opt-out preference signals — including the Global Privacy Control (GPC) browser signal — as required by applicable law.
Cookies and analytics. The Service uses only strictly necessary and diagnostic technologies, plus a single first-party analytics cookie that records how a visitor first reached our site, and first-party, cookieless page analytics (page views and aggregate visitor counts, with no cookies and no cross-site tracking). These are described in our Cookie Policy, and you can view, block, or delete cookies through your browser settings at any time.
Service and marketing email. We send transactional and service email (sign-in links, cycle and results notifications, billing receipts). We do not send marketing email without consent; any marketing email we do send includes an unsubscribe control you can use at any time.
Exercising your rights. Individuals: for questions or requests about assessment data held in your organization's workspace, contact your organization's administrator, because Kinetry processes that data on the Customer's behalf; for questions or requests about account or site data, contact us at privacy@kinetry.ai. Applicable state-law rights — such as access, correction, deletion, and, where applicable, opt-out of sale or sharing — will be honored as required by law, routed through the Customer where Kinetry acts on the Customer's behalf. We will not discriminate against you for exercising your privacy rights.
10. Children
The Services are not directed to children under the age of 16. Company does not knowingly collect, nor does it authorize the collection of, information of minors, and will work to delete any such information if it becomes aware of the same.
11. Cookies and Similar Technologies
Company uses cookies and similar technologies to operate the Services and to understand how visitors first find us:
- a session cookie that keeps you signed in (authentication);
- a cookie recording acceptance of the Terms;
- other strictly necessary first-party cookies, each listed in our Cookie Policy, that hold a step or a choice in place: that you completed the participant screen, an earlier cycle you chose to view, an administrator's read-only "view as" session, the email address you confirmed during signup, an AI assistant's connection request to return to after you create an account (Section 16), and that you dismissed the cookie notice;
- browser storage on your own device that keeps an unsent assessment draft until you submit it;
- an error-tracking service that receives technical error reports from the browser so failures can be found and fixed;
- a bot-protection challenge on the signup page;
- first-party, cookieless page analytics (page views and aggregate visitor counts; no cookies, no cross-site tracking);
- a first-party attribution cookie (bpi_src) that records the campaign labels or referring website that first brought you to our site, kept for 30 days and saved with your workspace if you create an account (see our Cookie Policy).
Company does not use advertising or marketing trackers, tracking pixels, or web beacons. Users may control cookies through browser settings or available preference tools; because sign-in relies on a session cookie, disabling cookies will affect core functionality.
12. California Privacy Rights
This section applies to California residents where applicable under the California Consumer Privacy Act, as amended by the California Privacy Rights Act ("CCPA/CPRA").
Categories of Personal Information. Depending on how individuals interact with the Services, Company may collect categories including:
- identifiers;
- professional or employment-related information;
- internet or network activity;
- characteristics or preferences voluntarily provided through surveys;
- professional development information;
- inferences generated through analysis;
- audio, electronic, or similar information where applicable.
Purposes of Collection. Company collects and uses personal information for:
- providing Services;
- processing Customer-sponsored assessments;
- generating coaching insights;
- maintaining security;
- improving Services;
- complying with legal obligations.
California residents may have the right to:
- request access to personal information;
- request correction;
- request deletion;
- request information about collection and disclosure;
- opt out of sale or sharing of personal information;
- limit certain uses of sensitive personal information.
Company does not sell personal information and does not share personal information for cross-context behavioral advertising.
13. Colorado Privacy Rights and Artificial Intelligence Notice
Company recognizes Colorado's heightened requirements concerning privacy, profiling, and artificial intelligence. The Services use AI technologies to analyze participant-provided information and generate coaching-related recommendations. Company does not use AI to make decisions producing legal or similarly significant effects concerning individuals.
AI-generated outputs:
- are informational;
- are subject to human judgment;
- are not employment decisions;
- are not intended to replace human decision-making.
Customers remain responsible for evaluating and appropriately using any insights generated through the Services.
14. Changes to This Privacy Policy
Company may update this Privacy Policy periodically. Changes may be made to reflect:
- changes in Services;
- legal requirements;
- technology changes;
- business practices.
Material changes will be communicated in-product, by email, or through other appropriate channels.
15. Public tools for AI assistants
Kinetry offers a small set of public tools that AI assistants, such as Claude or ChatGPT, can use on your behalf without a Kinetry account. When you use them, your assistant sends Kinetry the tool's inputs: a problem category chosen from a fixed list; an optional short description of a workplace behavior (at most 280 characters); a role label chosen from a fixed list (such as "direct report"); names of behaviors from Kinetry's published list of 24; ratings chosen from a fixed scale; and, for follow-up plans, who makes the commitment and how often you meet, each chosen from a fixed list. The assistant platform may also send standard technical details that Kinetry does not ask for, such as the app's name and version, your language setting, a coarse location, and anonymized user, conversation or organization identifiers. We use those details only to tell which assistant platform a request came from, and we do not store them.
The tools never ask for anyone's name, and they tell the assistant to leave names and other identifying details out of the optional description. If one is included anyway, it is handled like every other input: used only to produce the response and not kept.
- We process the inputs only to produce the tool's response. Apart from the problem category recorded below, we do not store them, use them to train AI models, or link them to any Kinetry account. Error reports from these tools never include their inputs.
- For each use we keep a minimal usage record: the time, the tool name, the assistant platform it came from, how far the tool got, and the problem category, if any. This record contains no description text, ratings, role labels, location, user or conversation identifiers, or IP address.
- Our hosting provider keeps standard request records, such as IP address and time of request, for a short period for security and reliability.
- Your conversation with the assistant is governed by that assistant's own terms and privacy policy. Kinetry does not receive the conversation itself, only the tool inputs and technical details described above. The tool's response goes back into that conversation and is handled under the assistant's terms.
- Links in a tool's response may carry labels that identify the assistant and the tool (for example, utm_source=claude&utm_medium=mcp&utm_campaign=sketch_team_member_baseline). If you follow one to our site, the visit is handled as described in Section 9 and our Cookie Policy. Our cookieless page analytics counts it. Our first-party attribution cookie (bpi_src, kept 30 days) records those labels, or for a link without labels the website you came from, unless an earlier visit already set it. If you then create a Kinetry account or open a demo, that record is saved with it so we can tell which tools lead to signups. It never contains anything you entered into a tool.
16. Connecting your workspace to an AI assistant
If your organization turns it on, people in the roles it chooses can connect Claude, an AI assistant made by Anthropic, to their own Kinetry account. Claude can then answer their questions about the people they can already see in Kinetry and, in advisor mode, if your organization turns that on too, help the advisor record their ratings. This section starts with what Kinetry can't control, then covers who can connect, what Claude can read, saving ratings, what Kinetry records, and how a connection ends. It is separate from the public tools in Section 15, which need no account.
- What Kinetry can't control. What Claude reads goes into the connected person's conversation in their own Claude account, and so does what the person tells Claude there, such as what an advisor describes when saving ratings. That conversation is handled under the terms between that person, or your organization, and Anthropic, not under Kinetry's terms. Depending on that account's settings, it may be kept in chat history, remembered across conversations, or used to improve Anthropic's models. Kinetry's commitments about its own AI, such as not training on your data and deletion within about 30 days, do not cover it. Deleting or erasing someone in Kinetry does not remove what Claude has already read or been told. Kinetry recommends that organizations ask people to use Claude accounts the organization manages. The one exception, when Kinetry's own staff connect as a workspace's advisor, is described two bullets below.
- Anthropic's role. Apart from the staff case in the next bullet, Anthropic is not Kinetry's service provider or subprocessor for these connections: Kinetry has not engaged Anthropic to process this data for Kinetry and does not pay Anthropic for these connections. Kinetry accepts Claude and no other assistant for these connections. Your organization decides, through its setting, whether people in the roles it chooses may connect Claude, and each person decides whether to connect their own Claude account. This is separate from Kinetry's own use of Anthropic to write coaching summaries (Section 8).
- When Kinetry's own staff connect. In advisor mode, a member of Kinetry's staff sometimes serves as a workspace's advisor. If that workspace turns on assistant access, that advisor can connect their own Claude account, as any advisor can. When Kinetry staff connect, Kinetry uses that Claude account in providing the Service to that Customer, so Anthropic handles what Claude reads, and what the advisor tells Claude, such as what they describe when saving ratings, for Kinetry, as Kinetry's subprocessor, rather than under the person's own terms. Kinetry keeps the Claude accounts its staff connect set so that Anthropic does not use those conversations to train its models. Those conversations stay in that Claude account's history until Kinetry deletes them. Erasing someone in Kinetry doesn't remove them automatically, so Kinetry deletes them itself when the person they're about is erased or when Kinetry stops serving that workspace.
- Off until your organization turns it on. AI assistant access is off in every workspace by default. An executive or HR administrator of the workspace, or the workspace's advisor in advisor mode, can turn it on. To do so they must confirm that the people assessed in the workspace have been told their results may be read this way, and that any required notice or consent has been obtained. Kinetry records who confirmed it and when. When Kinetry adds a kind of data Claude can read, it asks for this confirmation again with the updated wording, and Claude reads that kind of data only once it is given; everything confirmed before keeps working. In multi-rater (360) workspaces, rated participants who sign in to Kinetry, other than administrators (who see the setting on the Admin page), also see a one-time notice after it is turned on, and the participant screen mentions it. It is available on every Kinetry plan, including the free plan, and each plan's limits still apply to what Claude can read, such as how far back it goes: the last 3 scored cycles on the free plan, the last 6 on the Team plan, and every scored cycle on the Business plan and higher.
- Who can connect. The administrator chooses which groups can connect: executives, HR administrators and managers, or, in advisor mode, the advisor. In advisor mode the advisor is the one person who signs in, so the advisor is the one person who can connect. Employees can't connect. Neither can consultants who work across client workspaces, external raters, or people who are rated but don't sign in. Demo workspaces can't be connected, and the person connecting must have a confirmed email address. Kinetry accepts connections from Claude through claude.ai and from no other assistant or client, including Claude Code.
- Several workspaces in one connection. Someone who belongs to more than one workspace where they can connect can tick, when connecting, each of those workspaces the connection may read; the workspace they are in starts ticked. Each workspace is handled on its own: everything in this section applies to each workspace separately, under that workspace's own setting, plan and access record. Claude reads one workspace at a time, only what the person's role in that workspace shows them, and every answer names the workspace it came from. A workspace's administrators see the connection and the requests made in their workspace, not which other workspaces it reads.
- What Claude can read. Claude reads through seven read-only tools, and, for an advisor who can save ratings, the two rating tools described below. It sees nothing that the connected person's role doesn't already show them. The tools can read the latest scored cycle or any earlier scored cycle the workspace's plan includes. For every cycle, Kinetry uses the connected person's access as it is today: Claude can read earlier cycles for the people that person can see now, and nothing about people they could see before but can't now, apart from the way a whole-workspace report counts people who have since left, described in the next bullet. For the people that person can see, it can read their names, job titles and teams, who each reports to (naming that manager only when the person asking can see them or is them), how each relates to the person asking, and whether each was scored in the latest cycle. It can also read the names and dates of the scored cycles the plan includes, and how many older scored cycles the plan leaves out. For one person, in a chosen cycle, it can read the overall score and its change since that person's previous scored cycle within the plan, the three strongest and three lowest behaviors with their scores, the trend over the plan's cycles up to the chosen one, whether that cycle's scores came from one advisor or from several raters, which behaviors had too few ratings to show (without a number), how many open flags the person has today, and which behaviors moved since the cycle before, and by how much. It can read commitments someone took on, made in the chosen cycle, or made in an earlier cycle the plan includes and still open today: their wording, status and due date, whether they are overdue, the reason chosen when one was dropped, and how many were kept. Suggested actions nobody took on are only counted. On the Business plan and higher, it can also read a short coaching-loop summary: how many commitments were kept since the previous cycle, one behavior where a commitment was missed, and up to two behaviors that moved.
- Group reports. Claude can also read a report on a group for a chosen cycle. A manager can get one for the people they can see, or for those of them in a team, department or division they name. An executive, HR administrator or advisor can get one for any team, department or division (including the units under it), for the people who report to them (for an advisor, the people they rate), or for the whole workspace. A report compares the chosen cycle with the scored cycle before it within the plan, or with an earlier cycle the plan includes that the person asking names. It can include how many people were scored; the average overall score, and its change for the people scored in both cycles; how many people rose, held steady and fell, how many had no earlier score to compare, and up to five who rose and five who fell the most; up to three strongest and three lowest behaviors on average, up to three behaviors that rose and three that fell the most, and how many behaviors had too few ratings to average; the average in each cycle the plan includes, up to the chosen one; how many commitments someone took on were kept, and how many suggested actions nobody took on; and how many open flags there are today, how many people have them, and, for the latest cycle, how many of the group's current members were scored. A report on anything other than the whole workspace also lists each scored person's overall score and change. A whole-workspace report also gives the average by department (or by team, where there are no departments), and on the Business plan and higher it can show how behaviors with kept commitments moved compared with behaviors where commitments weren't kept. Averages are worked out only from the scores of people the connected person can see, plus, in a whole-workspace report, people who have since left, who are counted but not named. Every average needs at least 2 people behind it. Outside the whole workspace, who is in a group is as of today, for every cycle. When Claude reads an earlier cycle, open flags and commitment statuses are as they stand today, and the reply says so.
- Pair readings. On the Team plan and higher, Claude can also read a pair reading for two people the connected person can see, or for that person and one other, for a chosen cycle in which both were scored. It names the behaviors both are strong in, the behaviors both fall short on, and the behaviors where one is clearly stronger than the other, with who that is: up to three of each, each with a fixed line of guidance, and how many there are of each and in all. It names behaviors only and never includes either person's scores or category names. There is no reading for two people someone in the workspace has marked as not working together, for anyone who is no longer active, or for two people with fewer than 8 behaviors measured for both. In advisor mode, each person's levels are averaged over the last three scored cycles the plan includes, up to the chosen one.
- The team relationship map. On the Team plan and higher, Claude can also read the headline of the relationship map Kinetry shows on the Manager page, for a chosen cycle: for the connected person's own team (in advisor mode, everyone the advisor rates), or, for an executive or HR administrator, the team of a manager they can see. It names the three pairs that compound most and the three riskiest, each with how many behaviors both people are strong in or both fall short on and the top one; says how many pairs improved or worsened since the scored cycle before, naming the pair with the biggest shift; lists shared gaps several pairs have in common, by behavior and number of pairs; and counts the people and pairs on the map, and the pairs someone has marked as not working together. It never includes anyone's scores or every pair on the map, and it leaves out anyone the connected person can't see. In advisor mode, each person's levels are averaged the same way as for pair readings.
- What Claude never gets. Who rated whom; other raters' individual ratings or notes; anyone's private notes about themselves; email addresses or Kinetry's internal ids; how scores are weighted, category scores, or how many people rated someone; any average over fewer than 2 people, or over anyone the connected person can't see other than people who have since left, whom only a whole-workspace report counts; scores, commitments or reports from cycles older than the plan includes, or those cycles' names; risk levels, retention-cost estimates or other dollar figures; and any text Kinetry's AI wrote, such as coaching summaries. Commitments that began as suggested actions are fixed text from Kinetry's own library of actions, not written by AI, even though Kinetry lists them under AI coaching recommendations. Kinetry calls no AI to answer these requests, and nothing in Kinetry can be changed through a connection, apart from the ratings an advisor saves as the next bullet describes.
- Saving ratings (advisor mode only). In advisor mode, where Kinetry offers it, whoever can turn on assistant access can also let the advisor save ratings from Claude. It is off until they turn it on, and it works only while assistant access is on. To turn it on they must confirm that the advisor rates the people in the workspace and that ratings saved this way are the advisor's own judgments, and Kinetry records who confirmed it and when. A connection can save ratings only if the screen the advisor saw when choosing Allow said so. A connection made before saving was turned on stays read-only until the advisor connects again. Saving ratings isn't available in multi-rater (360) workspaces or to anyone but the advisor. The advisor tells Claude what they've seen, and Claude suggests a level for the behaviors it speaks to. Kinetry's tools ask Claude to save only the levels the advisor confirms, each with a short note in their own words. Kinetry requires a note with every saved level, but it can't check that the advisor confirmed a save or wrote its note, so each save is listed for the advisor as described below, and the monthly form shows every saved rating. Kinetry saves up to 6 at a time, for the people the advisor rates, and only to the month that is open now. Choosing Not observed removes a level saved earlier that month. A form the advisor has already submitted can't be changed from Claude. To help, Claude can read the people the advisor rates that month, with their teams, how many behaviors each has rated and open, when the last rating was saved, and whether their form is submitted; each behavior's published definition and first observable sign; and which behaviors were rated that month, when, and whether in Claude, in the weekly focus or on the form, but not at what level or with what note. After a save, Claude gets back the levels and notes just saved, the level and note each of those behaviors had earlier that month, with when and where it was set, and the level the advisor gave it the month before. Saved ratings don't change anyone's scores until the month closes or the advisor submits that person's form, which shows them with their notes to confirm or revise. From then on Kinetry treats them like the advisor's other ratings and notes. Kinetry keeps the saved levels and notes, not the conversation, and runs no AI of its own to suggest, check or save them. An administrator can turn saving ratings off at any time, which stops saves right away and leaves reading as it was.
- What Kinetry records. For each request, apart from ones turned away for coming too quickly, Kinetry records who asked, which tool, when, which cycle, how the request turned out, and, for a request about one person, who it was about. Requests that cover several people, such as the list of people, everyone's commitment tallies, a group report, a pair reading, or the relationship map, record no single person, and Kinetry doesn't record which group a report was about, which two people a pair reading was about, or whose team a map was for. Kinetry does not record the wording of the question. Each save of ratings is recorded the same way, including whether it went through, but the record never holds the levels or notes, which are kept with that month's ratings. The advisor can see their own recent saves from the last 30 days, up to the latest 200, on the Connected assistants page. While assistant access is on, the workspace's administrators can see recent requests from the last 30 days on the Admin page. The full record is in the workspace export, and Kinetry deletes it after 12 months. The export of a person's data lists their own connections and the date, tool, cycle and outcome of each request that was about them alone, without who asked. A group report can name the people who rose or fell the most or list each member's score, a pair reading names two people, and the relationship map names the pairs at its top; none of these is listed in their export, because it wasn't about them alone. Kinetry also records each connection: who made it, which assistant, when it was made and last used, and, if it was ended before it ran out, when and why. Administrators can see each live connection, who made it, and when it was last used, and the workspace export includes the connections and the workspace's assistant-access settings.
- Usage line. Each recorded request also writes the same nameless usage line described in Section 15: the time, the tool name, the assistant platform, and how the request turned out. It contains no workspace, person, or account identifier. When Claude sends a connection's token with any call, including a call to one of the public tools in Section 15, Kinetry checks the token and may update the connection's last-used date. So after a connection ends, Kinetry turns away every call that still carries its token, including calls to the public tools, and asks Claude to connect again, until the person connects again or removes the connection in Claude. The public tools' inputs are still handled as Section 15 describes.
- Connection email. When someone chooses Allow, Kinetry sends an email to that person's Kinetry address naming the assistant and each workspace it can read, and saying so when the connection can also save ratings, with a link to the page where they can disconnect it. If the email can't be sent, the connection still goes ahead.
- Tokens. Connecting uses OAuth 2.1 with PKCE. Claude receives a signed access token that lasts one hour and a refresh token that is replaced each time it is used. Kinetry does not store access tokens, and it stores refresh tokens and one-time codes as one-way hashes, not in readable form. A refresh token that goes unused for 30 days stops working. If a used refresh token is presented again more than 30 seconds after it was first used, Kinetry ends the connection in every workspace it reads; within those 30 seconds it is treated as a retry, such as when Claude didn't receive the reply, and answered with a new pair of tokens.
- How a connection ends. A person can disconnect their own assistant at any time on the Connected assistants page in Kinetry (kinetry.ai/settings/assistants), which the connection email also links to; that ends it in every workspace it reads. A workspace administrator can disconnect any one person on the Admin page, take a group out of the setting, which ends that group's connections, or turn assistant access off, which ends every connection in the workspace. A connection also ends when the person signs out of Kinetry, is deactivated, moves to a role that isn't allowed to connect, changes email address, or is erased, when the workspace is erased, and 90 days after it was made. For a connection that reads several workspaces, an administrator's action, deactivation, a change of role or email, or erasure in one workspace ends it in that workspace only, and it carries on in the others, except that erasing the person or the workspace where the connection was made ends it everywhere. Signing out and the 90-day limit end it everywhere. The Connected assistants page also lists the person's connections that ended in the last 30 days, and why. To keep using Claude this way after that, the person connects again from Claude. Kinetry can also turn this feature off for everyone, and then no connection works.
- Erasure. Erasing a person deletes their connections. The access record keeps each request they made or that was about them, with their name, email and account id removed, as Section 8 describes for other audit records. Erasing a workspace deletes all of its connections and access records. Neither reaches what Claude has already read (Section 8).
- Questions and requests. Kinetry keeps the connection and access records for your organization (Section 1), so questions or requests about them go to your organization's administrator. Questions about what an assistant did with what it read go to the person who connected it, to your organization, or to Anthropic.
- Purpose. These tools are for coaching and development. Our Terms don't allow Kinetry's scores and results, including what an assistant reads, to be used as the sole or primary basis for hiring, firing, promotion, demotion, compensation, or any other employment decision (Terms §7).
Contact: privacy@kinetry.ai · Kalgren Consulting LLC (d/b/a Kinetry), Minnesota, USA