Platform

Capabilities

A readiness assessment is a set of questions somebody in your organization writes and puts in front of the people who know the answers. Each question carries its own type, description, options, and branching rules, so what gets asked next can depend on the answer just given. Assessments are authored at one of three levels, executive, department, or individual, and the level decides who is shown the assessment. Answers are recorded per person, a response holding one answer per question, and a question can also collect ideas from the people answering, with votes on those ideas.

Limits

No instrument ships with the product. A newly provisioned account has zero assessments, and the questions are written by your consultant or your own admins. A level controls who is shown an assessment and nothing more, so no assessment is attached to a department and none carries a department field. Completion is calculated when something asks for it rather than stored, and two different formulas are in use, one over the responses that were started and one over an expected population of people times active assessments, so a completion figure means little without the formula that produced it.

Required roles

Creating, activating, and archiving an assessment requires a consultant or a company_admin, both roles you assign inside your own account, and editing one adds a condition the other three do not carry, that the assessment is still a draft. Our staff can do the same through a site_admin account, which sits outside every customer organization and is not a role you can give anyone in yours. Answering an assessment is governed separately, and a contributor is shown individual-level assessments only.

Capabilities

The secure score is one number for the organization, generated over every assessment in the account and stored with the date it carries. It holds two readings side by side, the model's score with its confidence and its written reasoning, and a reviewer's own score with notes, and where both exist the final score is the model's times 0.6 plus the reviewer's times 0.4, rounded, weights that are constants in code rather than a setting; where only one exists, that one stands. Gaps sit on that same record, each carrying a severity and, where one survived validation against a stored taxonomy of 4 functions, 19 categories, and 72 subcategories, a NIST AI RMF subcategory. Marking gaps addressed adds a severity-weighted bonus, critical counting 4 down to low counting 1, capped at 15 points, and records who marked each one and when.

Limits

There is no score per department, per assessment, or per level. The record has no department column, and even the department readiness report renders the one organization-wide figure. The score is also not a comparison against anything outside your account, since there is no industry average, no peer cohort, and no path by which another client's data reaches it. A score does not mean a control was tested: it reads what people answered, how much of the assessment was completed, how many people sit in each department, and which ideas were voted for, and when the model returns something that cannot be parsed it stores no number at all rather than a guess.

Required roles

Generating a score and adjusting one are limited to a consultant or a company_admin, both roles you assign inside your own account, and neither path admits a site_admin, the account our staff hold outside every customer organization. Adjusting is refused as well while a generation is still running. Regenerating an existing score runs through a different check and does admit our staff alongside those two roles. Viewing the score and exporting it as a PDF extend to executives and department heads.

Capabilities

A roadmap is a generated list of items, typically six to twelve, organized by phase and priority and scoped to whichever assessments you point it at. Each item is a title and a written description carrying one of five categories, security, governance, operations, technology, or training, plus a NIST function and subcategory that has to pass validation to survive. Items can be assigned to departments, given an implementation status once the roadmap is published, and carry file attachments, each one holding a malware scan status. A learning path is the same kind of record generated separately, its items organized by subject across generative AI, predictive analytics, data readiness, decision integration, and responsible AI, over four phases.

Limits

A learning path is not assigned by role: the function that creates one takes no role, and no role track or training level exists in the schema. The platform holds no training material and tracks no completion of any, so an item is a title and a description with no course, lesson, or curriculum behind it, and whether a roadmap contains a training item at all is the model's choice at generation time. Nothing chases the work either, since department assignment is per item rather than per person, and an item's implementation status changes when somebody sets it.

Required roles

Generating a roadmap or a learning path requires a consultant or a company_admin, both roles you assign inside your own account, or a site_admin account held by our staff, and it is refused until assessment completion meets the configured threshold. Publishing requires the roadmap's owner or a company_admin, so a consultant who did not create it cannot publish it, and our staff can publish it as well.

Capabilities

Five report templates ship, and each one is a fixed list of sections. The five templates are the executive overview, the board summary, the NIST compliance report, the remediation plan, and the department readiness report. A report is assembled section by section, each section accepted or edited by a person, and it cannot be finalized until every section has been accepted. The gap summary prints the NIST subcategory beside each of the top ten unaddressed gaps, wherever a subcategory passed validation. A report can be reviewed again through one of four lenses, executive, technical, compliance, or plain English, and a final or published report exports as a PDF.

Limits

The department readiness report is thinner than its name suggests, because its department section prints the department's name and member count while the progress and gap sections in it are organization-wide, including the single organization-wide score. A full-report review returns suggestions and changes nothing on its own, so the text moves only when someone accepts an edit. No report compiles on a schedule, and none appears without a person asking for it.

Required roles

Compiling a report requires a consultant or a company_admin, both roles you assign inside your own account, and it does not admit a site_admin, the account our staff hold outside every customer organization. Publishing is limited to the person who owns the report, and our staff can publish one as well. An executive, a department head, or a contributor with view access can export a final or published report as a PDF.

Capabilities

The register holds what your people enter or adopt from the shared catalog: each tool with its vendor, category, cost terms, license count, and a status of active, inactive, or evaluation, and each vendor able to hold contracts with renewal windows, documents with expiry dates, quotes, evaluations, and a risk tier. Cost analysis is computed the moment you ask for it, converting every tool to a monthly figure, which is what makes two differently priced tools comparable: monthly, quarterly, and annual subscriptions, per-user pricing multiplied by the license count, one-time costs spread over twelve months, usage-based pricing, and free tools at zero. Departments can be allocated a percentage of a tool's cost, and where no percentage has been set, that view charges the department the full cost. Redundancy is reported where two or more active tools in the same category share at least one department, with the potential saving shown as their total minus the cheapest of them.

Limits

Nothing discovers a tool, because there is no scan, no OAuth introspection, no network observation, and no spend feed, and the monthly usage figures for active users and sessions are typed in by a person with every field optional. Approval belongs to the shared catalog, where our staff review a submitted vendor or tool before it becomes visible to everyone, so your own register carries no approval state, no approver, and no approval date, and activating or deactivating a tool is a lifecycle switch that records none of the three. Two same-category tools held by different departments produce no redundancy finding. No return on investment is computed anywhere, because cost is never joined to scores, roadmaps, or outcomes, and where the cost analysis suggests looking at whether wider adoption of a tool used by one department would improve its return, it works out no such figure.

Required roles

Entering a tool or a vendor by hand and editing one afterwards requires a consultant or a company_admin, both roles you assign inside your own account, or a site_admin account held by our staff. Recording a month of usage takes the same permission. A vendor you invite gets a login inside your account once the invitation is accepted, and that login can edit the single vendor record its contact email or accepted invitation matches. Viewing tools and the cost analysis extends to executives.

Capabilities

Actions are written to a log that cannot be rewritten, where rows refuse updates and deletes and each one records who caused it and what changed. Score generations and regenerations, a reviewer's adjustment with the score before and after, gaps marked addressed or reopened, and changes to assessments, responses, questions, ideas, reports, roadmaps, files, tools, vendors, users, and departments are all recorded, from 102 explicit call sites in the code. An AI generation is kept as its own record wherever the code writes one, holding the type, the prompt and the assembled system prompt, the model identifier, the raw result it returned, and the prompt, completion, and total token counts. Department standing persists only as leaderboard snapshots, whose adoption score averages assessment completion and roadmap implementation in equal halves, and those are written when somebody opens the leaderboard rather than on a schedule, so a day when nobody looks leaves no snapshot.

Limits

Authentication is not logged at all, with no login, logout, failed login, password reset, or MFA event recorded, so this is not a login audit trail. Logging is explicit rather than automatic, which means the record is what the code writes and no more: the vendor lookup and the two report review paths, one working section by section and one over the whole report, call the model without writing a generation record, and their tokens go uncounted. The stored system prompt is also not byte for byte what went to the provider, because an organization-level safety prefix is added at call time and is not kept. The log itself does not sit in a database of its own either, since it lives in the shared platform database with a tenant identifier on every row, and a read of your account's log filters on that identifier.

Required roles

The account's activity log is readable by a company_admin, the only role inside your own account that can open it, so an executive, a consultant, a department head, or a contributor cannot. Our staff can read it as well, through a site_admin account that sits outside every customer organization.