Sana Labs alternatives after the Workday acquisition

Workday completed its acquisition of Sana in November 2025, in a deal Workday announced at roughly $1.1bn, and relaunched the product as part of the Workday suite in March 2026. If you chose Sana because it was the best AI-native learning platform on your shortlist, and your HR system is not Workday, your renewal is now a different decision from the one you originally made. The question is no longer whether the product is good. It is whether you want to be a standalone customer of something that is now a module of somebody else's suite.

Two things before anything else. Sana was a good product and still is — built around AI before that was a category, and the people who bought it were not wrong. And an acquisition is not a scandal. Most of what follows applies to any product absorbed into a larger suite, and most of it is about your contract and your requirements rather than about Sana.

What actually happened

Workday completed the acquisition in November 2025 and relaunched Sana as part of the Workday suite in March 2026. Sana's Core plan is listed at about $13 per user per month with a 300-licence minimum — roughly $46,800 a year at that floor, with the enterprise tiers quoted rather than published. Figures checked against G2's Sana Learn pricing page in August 2026.

Everything beyond that is a question for Workday rather than something to infer from a press release. What the standalone roadmap looks like, whether standalone pricing survives the next catalogue revision, how Sana's people data reconciles with Workday's — your account team is where to get those answers. Get them in an email you can forward to your finance director, not on a call.

What changes when a product is absorbed into a suite

Three things move, usually in this order, and none of them appear in a release note.

Roadmap priorities move to the centre of the suite

A standalone product builds whatever its own customers ask for most. A suite module builds whatever makes the suite more coherent. Those are not the same list, and the gap widens.

• Work that only matters inside the suite goes first: shared identity, shared people data, shared reporting, one admin surface.
• Work that only matters outside it — a connector to the HR system you actually run — moves down the queue.
• Overlapping features get reconciled. Where both the product and the suite have a skills model, a reporting layer or a content library, one of them gives way.
• The parts that made the product distinctive are the parts most likely to be re-described, in time, as capabilities of the platform.

Pricing stops being a standalone price

A 300-licence minimum at about $13 per user per month is a coherent standalone pricing decision. It says: we are aimed at large companies, and we would rather not sell to a 60-person training provider. What changes inside a suite is that your line item becomes one row in a much larger negotiation.

• Discounts get tied to what else you buy. Buy nothing else from the suite and you have less to trade.
• Published list pricing tends to go quiet once a product becomes a module — quote-only pricing is easier to bundle.
• The smallest customers of the standalone product are the least interesting customers of the suite.

None of it is certain, and some may never reach you. It is the usual pattern, though, and cheap to plan for.

Integration assumptions invert

Before an acquisition, a standalone product has to work with everything, because its customers run everything. After one, the reference architecture is the parent's. Documentation, sample configurations and the integration roadmap all assume the surrounding systems are the acquirer's. If yours are not, you have done nothing wrong — you are simply no longer the customer the manual is written for.

What to check in your contract this week

Before evaluating a single alternative, spend an hour with the agreement you already signed. Most of the decision turns on dates and clauses you can read today.

1. Term end date and notice window. Thirty, sixty or ninety days before the anniversary. Missing it by a day is how you end up owning a year you did not want.
2. Auto-renewal language. Another full term, or month to month?
3. Assignment and change of control. Most agreements expressly permit transfer to a successor, so the contract simply continues with the new owner. Read yours rather than assuming it either way.
4. Price protection. Any cap on the renewal uplift, and whether it survives an assignment.
5. The service description. Does the contract commit the vendor to a named product, or to "the service" as the vendor defines it from time to time? The second is far broader than it looks.
6. Seat minimums and true-up terms. A minimum you comfortably cleared can become the whole cost if your headcount moves.
7. Data export. What you can take out, in what format, and for how long after termination. Run a test export now, not under time pressure at the end.
8. Sub-processors, support and SLA. Ownership changes bring infrastructure changes: check what notice you get, and which legal entity now owes you the support commitments.

Then put the notice date in a shared calendar with a reminder six weeks ahead, owned by a named person, not a team inbox. A renewal nobody chose is a diary failure, not a product failure.

How to evaluate alternatives without chasing the same feature list

The common mistake here is to write a requirements document that transcribes the incumbent's feature list. It feels rigorous. It guarantees one of two outcomes: you re-buy the incumbent, or you buy its closest imitation, including the parts nobody used.

Start from the work, not the product

Write down what your team actually did last term or last quarter, in the order it consumed time. Not capabilities — tasks. A real list looks like this:

• Built four onboarding modules from a policy document and a slide deck.
• Marked 900 submissions against a rubric, then moderated the borderline ones.
• Issued certificates that an external body later asked to verify.
• Reported completion by department to someone who wanted it by cost centre instead.
• Onboarded a new client company and gave them their own branded space.

That list is short, specific and awkward to fake in a demo. Send it to every vendor before the call — the incumbent included — and ask them to do those five things with your material, not their sample data. Forty minutes of that beats a two-week feature comparison.

Break "the AI was good" into what the AI actually did

This is where evaluations go wrong when you are leaving an AI-native product, because one word is covering four capabilities with four different quality bars:

• Drafting — turning your own documents into modules and assessment items, with a citation back to the source page.
• Tutoring — answering a learner's question mid-lesson without handing over the answer.
• Grading — producing a suggestion against a rubric that an instructor can defend, confirm and release.
• Search — answering a staff question from internal material well enough that people stop asking a colleague instead.

Judge each separately, on your own material. Then ask the three questions that decide whether any of it can go near a live programme: which model provider is used and can you choose it, what happens to the text you send, and whether the output is a draft a person accepts or a write that lands on its own. Unattended writing is a different category of risk.

Ask who the roadmap is written for

One question predicts the next three years better than any feature table: which kind of customer produces most of this vendor's revenue, and are you that customer? Roadmaps follow revenue. A school group buying from a vendor whose money comes from large corporates will spend the year after next explaining why its requests keep sliding.

Where Lurno fits, and where it does not

Full disclosure: we build Lurno. This section is here because "what are the alternatives" is the question in the title, and skipping it would be coy.

Lurno is a standalone learning platform built by Augmental Learning Inc. It is not part of an HR suite, and there is no larger product for it to be folded into. It is priced per organisation rather than per seat — $119 a month on the entry tier with 1,000 members included, and no licence minimum — which suits the organisations a 300-licence floor was designed to filter out: schools, academies, and training providers running a separate branded academy per client company. Each of those is a real nested sub-organisation with its own members, branding and custom domain, kept separate underneath the platform rather than by a filter on top of it. The AI drafts from documents you uploaded, tutors learners and proposes grades, and every AI write is a draft a person accepts — how that works.

The limits, because you will find them anyway. There is no one-click importer from Sana; content moves as its source material, which is slower than a migration tool and tends to produce better courses. Self-serve signup and trials are in development — accounts are created by invitation today, and payments and checkout are designed but not built. SCORM, xAPI and LTI 1.3 are modelled but the runtime is still being built. Partner sign-on (silent SSO) ships today; SAML and OIDC single sign-on, MFA and passkeys are on the roadmap and not available.

And the case for staying, which a vendor is not supposed to make: if Workday is already your HR system of record, learning inside that suite is worth more than any feature comparison anyone can put beside it. Roster changes arrive without an integration project, and your identity, procurement and security reviews are already done. The side-by-side, including where Sana is the better choice, is on our Sana comparison page; our pricing is published rather than quote-only, so you can rule us out without a call.

The short version

1. Read the contract this week: term end, notice window, assignment, price protection, export rights.
2. Put the notice date in a calendar, owned by a person, with six weeks of warning.
3. Ask, in writing, about standalone roadmap and standalone pricing for your next two renewals.
4. Write down five things your team actually did last quarter, and make every vendor — the incumbent included — do them on your material.
5. Split "the AI" into drafting, tutoring, grading and search, and judge each on its own.
6. Ask each vendor which customer type pays most of their bills, and decide whether you are it.
7. Then decide. Staying is a legitimate answer.

An acquisition is a good reason to re-open a decision you made two years ago with less information than you have now. On its own, it is a poor reason to spend a term migrating.