Textio has made a substantial turn.
The original Textio helped companies improve the language they used to attract candidates. Its central insight was simple and important: words change who responds. Language that seems neutral often is not. Better language widens the pool.
Lavalier, Textio’s new interview intelligence product, moves downstream. It works inside the interview process, where hiring decisions are made and remarkably little usable evidence is produced.
That is the continuity between the old Textio and the new one. Textio began by augmenting writing. Lavalier augments judgment. This is a much bigger ambition.
The missing record of hiring
Companies spend a great deal of time interviewing people. Then they document the results badly. Interviewers neglect scorecards. Hiring managers provide incomplete feedback. Debriefs are shaped by the first confident speaker, who is often also the most senior person in the room. Nine months later, the company may know whether a new employee succeeded.
It usually cannot reconstruct why that person was hired. There is an enormous amount of activity and very little durable information.
Textio first tried to improve the feedback people wrote. The problem was more primitive than the quality of the writing. People were not writing anything. Lavalier starts there.
The product listens to role-intake meetings and interviews. During an intake conversation, it builds the job specification while prompting the recruiter to ask for missing detail. Does the employee own the campaign budget? What does the person actually do each week? How will success be measured?
These are ordinary questions. They are also the questions that mysteriously disappear from many hiring processes.
During an interview, Lavalier offers follow-up questions based on what the candidate says. Afterward, it organizes the evidence around the skills and responsibilities defined for the role. The user can return to the transcript, inspect the supporting conversation and change the assessment.
The human remains responsible for the decision. The machine improves the record on which the decision rests. That distinction matters.
Interview training at the moment it is needed
Companies spend billions teaching managers how to interview. Much of the training evaporates before the next interview begins.
Lavalier puts the guidance inside the conversation. If a candidate gives a vague answer about analyzing campaign results, the system can suggest the next useful question: What did the analysis reveal? What changed because of it? How did you know the campaign worked? This is interview training delivered when the interviewer can use it.
It may also create a way to evaluate interview quality. Organizations could identify inappropriate questions, inconsistent candidate experiences and interviewers who routinely fail to collect useful evidence. Eventually, they could identify their best interviewers and help everyone else improve.
That may be more important than another round of interviewer certification.
The ATS becomes plumbing
Lavalier integrates deeply with the applicant tracking system. Roles, candidates and stages move between the two systems. But Colleen Gallagher’s goal (She’s the new CEO and a long time Textio employee.) is clear: once Lavalier is connected, hiring managers and interviewers should rarely need to enter the ATS. This points toward an uncomfortable future for ATS vendors.
The ATS may become the data layer underneath hiring rather than the place where hiring work happens. That sounds sensible until every part of recruiting acquires its own attractive interface. Then the recruiter has a bucket of applications and the privilege of carrying information between them.
Software architecture has a fondness for turning organizational problems into navigation problems. Some recruiting tasks may disappear into Slack, Teams, Claude or another conversational interface. A sourcing request can often be expressed in a sentence. Interview planning, candidate comparison and hiring analysis require something more structured.
People use reports and spreadsheets for a reason. Sometimes you need to see the whole field at once. Lavalier sits near that boundary. It has conversational capabilities, but its larger value depends on a consistent reporting environment.
The interesting question is no longer whether chat replaces applications. It is which work requires a durable interface and which work can safely dissolve into conversation.
Hiring probably needs both.
The actual job and the official one
The most interesting part of Lavalier may be its skills matrix. Most large companies have several versions of every job. Recruiting has one. Compensation has another. Learning and development has a third. The hiring manager carries a fourth, usually in their head.
Each version is correct for its purpose. The trouble begins when they meet.
The compensation description determines the level and salary. The learning description connects the employee to development programs. The recruiter’s version attracts candidates. The hiring manager’s version describes the work that must get done on Tuesday morning.
These descriptions rarely reconcile cleanly.
HR data is strongest near administration. Payroll knows exactly what someone is paid. The HR system generally gets the employee’s name right. Move closer to the work—what the person actually does, which skills matter and why one employee succeeds—and the data becomes thin remarkably quickly.
Lavalier begins to capture information closer to the work. It records what the hiring manager says is required, what interviewers ask and what candidates demonstrate. That gives the organization a more detailed account than the formal job description usually provides. But local detail is not enough.
A hiring manager may accurately identify the person who fits today’s immediate need and still hire the wrong person for the organization’s longer-term direction. Six-month success is useful evidence. It is not the same thing as proof that the company hired the right people.
The real opportunity is to connect the evidence collected near the work with the larger systems used by compensation, development and workforce planning. The corporate taxonomy supplies the frame.
Lavalier supplies the detail. That combination could make skills data genuinely useful instead of merely decorative.
Better evidence, not mechanical certainty
There is an obvious temptation to turn all of this into a score.
Textio’s original score gave writers a visible target. An interviewer score could do the same. A candidate score would be more dangerous. A system that claims one applicant deserves an 87 while another deserves a 73 creates an appearance of precision that the underlying judgment cannot support.
Lavalier’s current approach is more restrained. It groups candidates by priority and links its observations to transcript evidence. That is a better use of the technology.
Hiring decisions will remain judgment calls. They should. The goal is not to remove judgment. It is to make the evidence visible, improve the questions and leave a record that survives the debrief.
The practical next step for Textio is to keep moving in both directions. Go deeper toward the work by improving the evidence gathered during intake and interviews. Then connect that evidence upward to the formal definitions used across HR.
If Lavalier can do both, it will become more than a pleasant interface sitting on top of an unpleasant ATS. It could become the place where companies finally learn what they mean when they say they are hiring for skills.