Skills intelligence software is two products under one name
Most skills software asks people to type their skills in, or reads them from HR records. Only 18 of 139 products work skills out on their own, without anyone typing them in. Only two of those sell to smaller companies. Check what a tool actually reads before you trust a skill on someone's profile.
Two questions decide most of this purchase. What does the product actually read, and does a person have to type the skill in by hand? A product's name answers neither question. Two products can both call themselves skills intelligence platforms. One reads the work people produce. The other sends out surveys and waits for the answers.
Most tools read HR records or ask people, not the work
107 of the 139 products say enough in public to sort them. They fall into four groups, or a mix. The group decides a product’s cost, its risk and who it suits. Two groups matter most here. One works skills out. The other asks people.
Reads the work
It reads what people produce at work. Nobody types a skill in.
- What it reads
- tickets · documents · codesupport cases · meetings
Someone types it in
The company builds a skills list. People rate themselves, or their manager rates them. This is what most people mean by a skills matrix.
- What it reads
- self-rating · manager rating
work output
Someone sits a test
A person sits a test. The test shows skill only in what it covers.
- What it reads
- test scores · simulations
anything untested
Sells market data
It sells data about the job market. It shows what employers advertise for and what people put on public profiles. It never looks at your own staff.
- What it reads
- job adverts · public profiles
your staff’s skills
At least half of the 18 products that work skills out alone do not read people's work. What they read instead is CVs, job descriptions, HR records or job-market data.
Nine vendors set out where their skills data comes from.
Where each vendor’s skills data comes from
9 vendors- Beamery
“Skills from Beamery AI are inferred skills based on the primary job title.”
- Oracle Dynamic Skills
It pulls from HR records. Those are profiles, candidate data, requisitions, job descriptions and courses, not the work itself. Oracle said in August 2026 it is building a tool to read work systems directly. Treat that tool as not yet available.
- Degreed
A skill can reach a profile four ways. The employee adds it, or it comes from their CV. The company assigns it from a role. Or Degreed’s AI suggests it.
- Nestor
Someone can confirm a skill three ways. They rate themselves. Their manager rates them. Or a colleague backs them up.
- SAP SuccessFactors
Skills reach a profile from the job role, learning, performance reviews, goals and recruiting. Self and manager ratings get entered by hand too. Suggesting skills from a CV or written achievements needs a separately purchased AI units licence. All of it is an HR record, not the work itself.
- Neobrain
Its AI reads CVs, career paths and LinkedIn profiles. It drafts a first skills list from org charts and job descriptions. Neobrain says setting each person’s level “is fundamentally human”. Its Teams assistant collects feedback from managers and colleagues.
- Workday Skills Cloud
It suggests skills already in Workday. Those are job profiles, job history, projects, courses and public feedback. An employee or admin adds them. It reads only what is in Workday. Data from work tools such as Jira, GitHub or Slack counts only if it is loaded in.
- Cornerstone
Its Skills Graph attaches skills to each job title, using data from more than 250 million profiles. A skill only counts once confirmed. Levels come from self, manager and peer ratings, plus a system estimate.
- Visier
You can bring in a skills list from partners like TechWolf, Fuel50 and Lightcast, or from almost any system. Its own tools also pick skills from job adverts, CVs and job titles.
Beamery's AI suggests skills from a job title alone. What you get is a job-title lookup. That is useful for what a role usually needs. It is poor for judging one person.
Reading work does not give the full picture either. Microsoft calls these skills a partial picture. In its own words, they “aren’t intended to be a comprehensive reflection of a person’s capabilities”. Confirming the results is a separate problem. In TechWolf's own T‑Mobile case study, confirming was left to employees and never made compulsory. Confirmation stands at around 32% across the company. It reaches closer to 70% in back-office functions. The case study says the gap is “concentrated in frontline retail and care”. That project still worked from HR and learning records and job descriptions. The case study calls reading work systems directly the next step. Get the same numbers from every vendor on your shortlist.
Only two of six jobs need an answer about one person
Whether this matters depends on what you want the data for. For two of these six jobs, the wrong kind of product gives a confident, wrong answer. For the other four, you do not need software that works out each person's skills at all.
Move someone into a different role
About the personA job-title lookup tells you what someone in that role would usually know. It does not tell you what this person has actually done. If the two differ, you may only learn it after you have made the move.
Find who can do X, right now
About the personSoftware that works skills out for itself can answer this. So can a self-reported matrix, if somebody keeps it up to date. Both go out of date. A matrix gives no warning on screen when that happens.
Show an auditor a certification is still valid
Needs sign-offSoftware that works skills out for itself will not help you here. An auditor does not want to know what a system concluded about somebody. They want a dated record with a name on it. That is what matrix tools are built to keep. AG5, for example, tracks when each certificate expires and logs every change made to it.
Size a skills gap across the org
About the roleHere you need the size of the gap for each role, not an exact reading of each person. A job-title lookup gives you that. Workday, SAP, Oracle and Cornerstone each sell a skills tool that reports on gaps across a workforce. Cornerstone, for one, charges extra for the add-on that shows gaps across teams and the whole business.
Target learning by role
About the roleThe same suites use their skills data to recommend learning. Some suites sell their learning tool separately. Oracle and SAP both do.
Build job architecture or pay bands
About the roleHere you are describing roles, not people. You do not need to know what each employee can actually do.
Pay for software that works skills out for itself only when you need an answer about a person. Three of those six jobs are about roles, not people. A fourth needs a dated, named record instead of something a system worked out.
Most of the kind that works skills out sells to large companies
Which kind you need is only half the decision. What you can buy depends on how big your company is.
| How skills get in | Enterprise only | Enterprise + mid | Mid‑market | Smaller | Unclear | Total |
|---|---|---|---|---|---|---|
| Software works it out | 8 | 5 | 0 | 2 | 3 | 18 |
| Mixed | 3 | 4 | 1 | 0 | 0 | 8 |
| A test | 7 | 10 | 3 | 0 | 2 | 22 |
| People type them in | 5 | 18 | 15 | 4 | 3 | 45 |
| Market data only | 8 | 4 | 0 | 1 | 1 | 14 |
| Couldn't determine | 17 | 8 | 2 | 1 | 4 | 32 |
In 15 of 21 mid-market products, people type their skills in themselves. Among the mid-market products that publish how they work, none works skills out entirely on its own. Of the 18 that do, eight sell only to enterprises. Five more sell to both enterprises and mid-sized firms. 365Talents, now Docebo's advanced skills product, is built for organisations that typically have 1,000 employees or more. The smallest customer TechWolf names a size for has about 5,000 people. Gloat says it supports organisations of all sizes. The smallest company in its own case studies has 4,200 employees. Eightfold publishes no price and no minimum size. It counts Fortune 500 companies among its customers. It has one US federal contract on record. That is a single-year Air Force contract worth about $330,000.
Two products break that pattern. Talentguide, a Belgian product, has paid plans from 50 seats. Microsoft People Skills works out skills from people's documents, emails, chats and meetings in Microsoft 365. Getting skills worked out this way costs $0 extra, so far, with Microsoft Copilot, its paid AI assistant, or certain Viva licences. Microsoft publishes no minimum number of seats for those. A standard Microsoft 365 licence on its own does not include it.
Get Microsoft’s method for skill levels in writing before you use them. People can use it to find colleagues by the skills they share on their profiles. Microsoft says its performance may be limited for staff who spend little time in Microsoft 365. That includes frontline and field workers. It also says People Skills is not designed for promotion or pay decisions.
The vendors split by what they actually read
There is no single ranking of these vendors. A matrix tool and the automatic kind are not doing the same job.
TechWolf and Microsoft are built to read work
TechWolf publishes no price and no minimum size. Its own blog says it is built for large enterprises. By default, employees confirm their own skills through its assistant in Microsoft Teams or Slack. The results feed into Workday, SAP and Visier. It connects to work tools such as Jira, ServiceNow and Teams. It also connects to HR and learning systems. At T‑Mobile, 78% of the skills it suggested for jobs got approved. That run worked from HR and learning records and job descriptions. Get TechWolf to name the systems it would read at your company. Microsoft People Skills does much the same inside Microsoft 365. Start with these two if you want software that works skills out for itself.
Weighing these two? Read TechWolf vs Microsoft People Skills. The decision comes down to whether those skills must reach your HR system, as TechWolf’s do and Microsoft’s do not.
These platforms differ in what they actually read
Large employers use several of them to move staff internally and to plan job structures. Gloat, for example, builds profiles from HR records, CVs, LinkedIn profiles, and learning and recruiting systems. It also adds optional ratings. Its agents add skills from the work itself, such as projects and code. If you run one of these suites, its skills tool is the sensible place to look first.
A skills matrix suits most mid-market jobs
AG5 suits manufacturing and regulated frontline work. Skills Base gives you a straightforward inventory. MuchSkills builds one from employees rating their own skills. Managers then check the skills that matter most. SkillsTX is built on SFIA, the IT and digital skills framework. It suits IT and digital staff, not the whole workforce. Fuel50 began as software for planning people’s next roles. It now pairs a skills inventory with a talent marketplace. That is a place inside the company where employees can see and apply for other roles or projects. The inventory starts from employees rating their own levels. TalentGuard has employees rate their own skills. A manager must confirm or change every rating. It then measures how ready each person is for the role they want next.
Three vendors test people directly
Workera tests technical subjects such as machine learning, AI, data science, software engineering and data literacy. Its plans start at $40 a user a year. The US Air Force bought 7,000 Workera licences for one year. It paid $910,000, or $130 each. iMocha does both testing and profile-building. It offers over 10,000 tests across technical, functional and leadership skills. It also fills in a person’s profile. That profile pulls from their CV, certificates, courses, project records and performance records. It adds ratings from the person and their manager, plus data from work tools such as Jira and GitHub. Kahuna handles frontline competency. An auditor may ask to see this kind of record.
Where published, prices run from $0 extra to $1.60 a user a day
Almost nobody publishes a price. The few public figures need a second look before you rely on them.
If the skills tools sit inside a suite you use, read your contract before you assume they are included. SAP and Cornerstone both charge for parts of theirs. Either way, budget for the time. TechWolf says it takes three to six months before every job and every employee has a confirmed set of skills. Only two to six weeks of that is technical work. The rest goes on confirming the results and on managing the change.
Five vendors here have shut down, sold or cut staff
Weigh the vendor's position as carefully as the product. The five cases below all date from 2024 on.
Hand-built skills frameworks can stall too. Etsy published an engineering career ladder in October 2019. It accepted two small typo fixes later that month. It has not updated the public version since. It is now archived. More than 20 people made 69 changes to Monzo's public framework. Its main published text last changed in September 2019. At the Financial Times, the engineering progression framework had lived in a spreadsheet. That spreadsheet split into at least six competing versions. Some named roles that no longer existed. A working group of five FT engineers rebuilt it. They released a first trial version roughly fifteen months after starting. The FT still maintains that version seven years on.
Whatever you buy or build, someone has to keep it up to date. At the FT, one of the engineers from that original working group was still updating the framework in January 2026.
Buying gets you a working system faster
The ‘taxonomy’ is the standard list of skills and occupations that people get matched against. It costs nothing. The European Union's ESCO classification publishes 13,939 skills across 28 languages. The US O*NET database covers 1,016 occupations. Both are free to download and free to use, as long as you name the source. O*NET is updated every quarter.
The hard part is working out who has which skill. Independent studies and TechWolf both give numbers here.
In 2024, researchers tested OpenAI’s GPT-3.5 and GPT-4 AI models on six sets of job-advert data. The AI models fell well short of older, purpose-built software on five of the six sets. Part of the problem is the answer key itself. Both research teams blame overlapping skill names, which make the right answers hard to pin down. In published studies, trained reviewers agreed well on which words in a job advert named a skill. Experts label adverts differently when they use a standard skill list instead.
There is also a legal point to weigh. Under the EU AI Act, software that works out skills for employment decisions carries legal duties. Normally the vendor carries them. If you build your own, those duties move to you.
Before you decide to build, compare it against buying. In TechWolf's own T‑Mobile case study, skills were suggested for every employee in under six months.
Here is what building it looks like, agent by agent, with the data each one needs and who decides. The HR Agent Map
Try the skills tool you may already own before you buy
Switch on People Skills and see what it produces before you talk to any vendor. It costs $0 extra, so far, with Copilot or certain Viva licences. It will show you what your Microsoft 365 activity says about your people. Without one of those licences, People Skills will not work out anyone's skills for you.
Check what your contract already covers. Dynamic Skills comes at no additional cost for Oracle Global HR customers, so far. SAP's own add-on for finding skills automatically needs a separately purchased AI units licence. Cornerstone charges extra for its Skills Architect and People Graph add-ons. Workday switches Skills Cloud on automatically under only one of its two contract types. Turn on what your suite includes. Check whether it answers your question before you shop anywhere else. At Cornerstone, a suggested skill does not count until the employee confirms it. Expect to ask people to fill in their own profiles.
For compliance or frontline competency, a skills matrix tool does that job. It is where most mid-market purchases should land. Some matrix tools publish their prices. Skills Base lists paid plans from $2.25 a person a month.
For a large organisation, start with TechWolf or Microsoft People Skills. For a smaller company, two of the 18 products that work out skills on their own sell to smaller companies. They are Microsoft People Skills and Talentguide. Three more do not make clear who they sell to.
Buy nothing extra. A job-title lookup, or the suite in your contract, tells you what a role usually requires.
For 32 of the 139 products we could not tell from public material how skills get in, so we recorded them as unclear rather than guessing.