Product roadmap
What we are building, and when
We are live now, in beta: institutions can sign up and run a real assessment window today. This is the order everything else is released into the running product, derived from a full teardown of the platform this replaces, so the sequence reflects what institutions actually used rather than what was easiest to demo. Every item says what it unlocks, and several say what they deliberately do not include.
What being live means. Institutions sign up, onboard and run their own branded instance from day one. Everything below is the order further capabilities are released into the live product, published so you can see what lands when instead of finding out later. When a date moves we say so in the build log rather than quietly editing it here.
91% 10 of 11 milestones shipped, the rest released into the running product
- Status
- Open now, in beta
- Roadmap updated
- July 27, 2026
Build log
Subscribe by RSS to the build logDated entries for every status change, including the ones where a date moves. We publish slips here rather than quietly editing a window above.
-
shipped Compare two assessment windows, and a signup bug that only browsers could see
Analytics now compares two windows for the same cohort: the change per competency for the students who took both, self and observer, with a CSV for reviewers. Students who have not submitted get one reminder, timed to the window. The roster has the cohort checkboxes its own instructions promised, plus a CSV template. Staff and anyone holding a share link can download a student's report as a PDF, built off the request so a busy afternoon cannot slow the site down. And a fix we owe an apology for: creating an account or signing in on the main site worked on the server and failed in every browser, because a security header we set forbade the redirect to your institution's own address. The account existed, the welcome mail went, and the screen showed the form again. If that happened to you, your account is there and sign-in now lands where it should.
-
progress We ran a security review, and published what it found
A full review of the platform: cross-tenant isolation, the public demo, student account provisioning, SAML and OpenID Connect, assessment link tokens and the AI path. It found one high-severity problem and we fixed it the same day. Our demonstration account's containment was keyed to the institution subdomain, and the demo signs visitors in on our main application host where no institution is bound, so somebody who navigated back to the team-invitation page there could have sent mail from our domain to any address they chose. It is now keyed to the signed-in account instead, and institution pages are not served on the main host at all. The full write-up, including a low-severity finding and one concern we investigated and dismissed with a measurement, is in the trust centre. We publish these because "we review our own security" is worth almost nothing as an assurance and quite a lot as a dated record with findings attached. It is still not an independent test: a review of our own code by the people who wrote it has a blind spot exactly the shape of the reviewer, and that is the part still outstanding.
-
shipped Try it without signing up, and a full walkthrough
There is now a demonstration institution you can open without giving us anything: no signup, no email address, no card. You are signed straight in as the careers director of Riverside State University, a fictional college that has already run a term, so every screen has real work on it rather than a "no data yet" message. Change whatever you like; it rebuilds itself every hour. Alongside it there is a complete walkthrough at /guide, with screenshots taken from that same demo so the pictures match the numbers, and an FAQ at /faq covering cost, FERPA, data deletion, single sign-on and student accounts. The FAQ also lists what we do not have: no SOC 2, no ISO 27001, and no completed penetration test. An analyst who discovers that later stops believing the rest of the page, so it is on the page.
-
shipped Student accounts, and help on every page
Students can now create an account and sign in, with a password or with Google, Microsoft or LinkedIn. This is optional and always will be: the assessment link your institution emails works on its own, with no account and no password. An account simply keeps your assessments and reports in one place. Institutions decide who can have one: an account can only be created for somebody already on their roster, so nobody can sign up their way into an institution they do not attend, and a student who lands on the wrong address is routed to the right institution by their email rather than hitting a dead end. Staff can also add a student by hand now rather than editing a spreadsheet. Alongside it, every page in the product gained a plain-English panel explaining what it is for, what to do, and what usually goes wrong. This is free software with nobody staffing a support desk, so anything you cannot work out on the page is a question that never gets answered. We would rather put the answer next to the thing.
-
change We said the penetration test would happen before launch. It did not.
The trust centre and the HECVAT response both stated that an independent penetration test was scheduled before launch. Signups opened without it. Rather than quietly reword that into the future tense, both pages now say what happened: the test is booked work rather than completed work, and our security posture is self-assessed until the summary letter exists, at which point it will be published here. Retention, deletion, export and security headers shipped in July and are real, but none of that is an independent review and we will not let it stand in for one. If your process needs a completed pen test before you put student data in a system, we do not clear that bar today.
-
shipped Signups are open, and the platform is labelled beta
Any college or university can create an institution now, at app.careerreadinessreport.com/register: your own subdomain, your logo and colors, the NACE instrument live, and no demo call, purchase order or cost. This is eighteen days ahead of the August 15 date we published, because the assessment engine, rater feedback, reporting and analytics all shipped early. That date has since been taken off the site, because a product you can use today should not advertise a future launch; it stays here, in the log, rather than being edited away. Everything now carries a beta label, and that is a statement of fact rather than modesty: the platform is built and tested, but no careers office has run a real term through it yet, so the first institutions to sign up are the first to do that. The label comes off when a cohort has finished a term, not on a date.
-
Public trust center and legal documents
Everything a security review normally requires an NDA and a sales call to obtain, published in full and free.
What this unlocks
- HECVAT 4.1.5 response across all seven domains, including the AI/ML governance domain
- VPAT 2.5 accessibility conformance report against WCAG 2.2 AA, Section 508 and EN 301 549
- Data processing agreement with a FERPA school-official addendum, ready to sign
- Privacy policy, terms, accessibility and security statements
What it does not include
- No SOC 2, no ISO 27001, no completed third-party penetration test, no cyber liability insurance. All stated on the trust center rather than omitted.
Why hereBuilt first because it is the cheapest thing that removes the largest obstacle. An institution can rule us in or out today, without waiting for a single feature.
-
Launch: institutions can sign up
We open. Institutions sign up directly, onboard, and stand up their own branded instance without a call or a procurement cycle.
What this unlocks
- Self-signup for any college or university, free, no demo call and no purchase order
- Your institution set up with your logo, colors and your own subdomain
- The whole cycle ready on day one: a 26-item instrument on the eight NACE competencies, 360 rater feedback, student reports, cohort analytics and accreditation exports, all shipped ahead of schedule
- A place in the first pilot cohort
- Direct input into sequencing while the order is still cheap to change
What it does not include
- Signing up puts your institution in the first cohort using this in production. It has been built and tested but not yet run by a real careers office through a real term, and we would rather say that than discover it with you.
Why hereThe plan was to launch on onboarding while the assessment engine was still being written. It shipped early instead, along with rater feedback, reporting and analytics, so launch now opens a complete cycle rather than a placeholder. The reason for launching on onboarding still holds: the cheapest time to hear that something is in the wrong order is while it is still cheap to change.
-
Institution accounts and your branding
The tenancy layer: your institution, your people, your roles, your logo and colors, on your own subdomain.
What this unlocks
- Self-signup for an institution, with roles and permissions for staff
- Your logo, colors and institution name throughout, with live preview
- Your own subdomain, and a custom domain if you want one
- Tenant isolation enforced in the data layer and covered by automated tests
Why hereShips with launch. Multi-tenancy is not a feature you retrofit, and isolation is enforced so an unscoped query fails rather than quietly returning another institution's records, and that has to be true from the first table.
-
The NACE competency instrument
All eight NACE competencies as a versioned instrument, plus your own custom competencies alongside them.
What this unlocks
- The full eight-competency NACE set, not a subset
- Versioned instruments, so changing wording does not silently invalidate last term's data
- Custom competencies for programs with their own outcomes
Why hereVersioning is here rather than bolted on later because the single most common way longitudinal assessment data becomes unusable is an instrument that changed without anyone recording that it changed.
-
Student assessment
Rosters, cohorts, terms and the student-facing assessment with save and resume.
What this unlocks
- CSV roster import, organized into cohorts and academic terms
- The student assessment, saveable and resumable part way through
- One scoring implementation, with tests, used everywhere a score appears
What it does not include
- No mobile app in v1. The interface is responsive and the API stays clean so an app remains possible.
Why hereOne scoring service is a direct lesson from the teardown of the previous platform, which had fourteen scoring implementations that disagreed with each other.
-
360 employer and supervisor feedback
Students invite the people who actually watched them work, and those people respond without creating an account.
What this unlocks
- Single-use, expiring, cryptographically signed rater links, no password and no signup
- Custom rater fields per institution
- Automated reminders, because response rate is the binding constraint on any 360 program
- Self versus observer gap analysis, which is the part students actually act on
- Anonymity thresholds you control
Why hereRequiring an employer to register is the single largest cause of response-rate collapse, and a pile of half-created external accounts is a security liability in its own right.
-
Student dashboards and reports
What the student sees, and the PDF they keep.
What this unlocks
- Student dashboard with competency visualizations
- A report as real HTML first, with the PDF generated from it
- Queued batch generation for a whole cohort
- Private storage, so a report URL is not a public URL
Why hereHTML first, PDF second. The previous platform generated PDFs that silently fabricated content when batch generation failed, which is worse than producing nothing at all.
-
Cohort analytics and accreditation exports
The reporting an assessment office actually has to produce, and the evidence a reviewer actually asks for.
What this unlocks
- Filter by cohort, major, term and your own custom fields
- Exports built for AACSB, ABET and regional review rather than for a dashboard
- Competency to program-learning-outcome mapping, with partial mappings marked as partial
- Response rates, which reviewers do ask about
- Term-over-term comparison on a stable instrument: two windows side by side, the change per competency for the students who took both (shipped September 2026)
What it does not include
- No platform can produce accreditation evidence on its own. This supplies the data and the mapping. The analysis, the curricular response and the narrative are the program's work.
Why hereThe teardown showed accreditation is what schools were really buying, judged by what they configured rather than what they said. This is the payload, not a reporting afterthought.
-
AI development narratives
A written development narrative generated from a student's competency results, with the governance stated up front.
What this unlocks
- A development note written from the scores, labeled AI-generated wherever it appears
- The student's name, email and ID are never sent to the model. Only scores, the scale and the institution name are
- Regeneration on request by a student or an administrator
- An institution-level off switch, reachable by an administrator without asking us, with the rest of the platform fully functional without it
- Spend caps checked before each request rather than reported after, so an AI feature cannot quietly become the largest line item
- A failure is recorded and shown as a failure. Nothing is ever invented to fill the space
What it does not include
- No model makes or influences any decision about a person. Scoring, aggregation, benchmarking and access control are deterministic code. Our provider agreement prohibits training on submitted content.
Why hereDeliberately late, and built only once the instrument, the scoring and the reporting were finished. An AI feature is the easiest thing to demo and the least important thing to get right first, and putting it before the instrument would have been building the wrong product impressively.
-
Campus single sign-on
SAML 2.0 and OpenID Connect, configured per institution, optional, and settable to required.
What this unlocks
- SAML 2.0 for Shibboleth and ADFS, OpenID Connect for Microsoft Entra ID, Okta and Google Workspace
- Live SP metadata at your own subdomain, so your identity team can configure their side first and nothing needs re-sending when it changes
- Signed assertions required and verified against your certificate, with single-use assertion IDs
- Just-in-time provisioning, with an optional domain allowlist and a default role you choose. Ownership is never granted this way
- Optional enforcement, which closes password sign-in for everyone but the institution owner
- A staged rollout: enforcement does not take effect until the connection works, so turning it on early cannot lock anybody out
What it does not include
- Verified against a live campus identity provider with our first pilot institution, not before. The protocols are implemented and tested here, but we will not claim a real Shibboleth or Entra ID tenant has signed in until one has.
Why hereThe build is short. The coordination is not: each campus identity team is measured in weeks. Building it early means the metadata document exists now, so an identity team can start their side the day an institution signs up rather than waiting on us.
-
Independent security review and audit
Independent penetration test, load testing, and the trust documents re-issued against the running system rather than the planned one.
What this unlocks
- Independent penetration test, with the summary letter published
- Our own security reviews, published with their findings rather than summarised as an assurance (first one July 2026)
- Full audit logging and retention controls (shipped early, July 2026)
- The VPAT re-issued to cover the application, not only the website
- The HECVAT response re-issued against the shipped system
- Peer benchmarking across institutions, with minimum group sizes and an institution-level opt out, once enough institutions have run real terms to build a comparison set from
- The full feature set complete, still free, still self-signup
What it does not include
- This milestone stays open until an independent tester has actually looked. We do run our own reviews and we publish them with their findings, including a high-severity one we found and fixed on 2026-07-28, but a review of our own code by the people who wrote it has a blind spot that is exactly the shape of the reviewer. That is what independence buys and it is the part still outstanding.
Why hereThe current trust documents describe a system being built. They are dated and scoped for exactly that reason, and they get re-issued here against something an auditor can actually poke.
Nothing matches that combination.
This roadmap is a plan, not a contract, and dates move. When one does we change it here rather than quietly leaving the old one up. If something you need is missing or in the wrong order, that is worth telling us while the order is still cheap to change.