What it was like before. The skills assessment already allowed for the generation of an independent report with graph and PDF, but the detection was based solely on the semantic analysis of the conversation, without explicit methodological guidance or anchoring to behavioral indicators.

What changes. Three additions that bring the system closer to the BEI (Behavioral Event Interview) methodologies already used by selection teams:

1. Questions related to competencies. When configuring a competency in the template, you can now define a list of linked questions. These questions serve a dual purpose: during automated assessment, they facilitate the identification of critical incidents in the conversation, since the system knows where to look; and for the consultant, they act as interview manual, indicating what to ask according to the type of profile and the competence that you want to explore.

2. Behaviors associated with levels. Three established methodological practices are formalized and unified —anchoring to levels, weighting of indicators and derailment factors— in a simplified model. Each behavior is associated with a competency level, with a polarity (positive/negative) and a weight (low/medium/high). The assessment engine adds a new phase that detects these behaviors in the transcript, more accurately classifying the candidate's level. The model is optimized for automated assessment by LLMs, but with intuitive semantics for HR consultants who define the competencies.

3. Customization of the skills chart. The radar chart that accompanies the report now allows for two adjustments: change the colors (available colors from the platform palette plus the main brand color) and modify the values of each skill by clicking directly on the levels, in case any need manual correction before exporting or sharing.

Furthermore, it is now possible delete a skills report existing and reprocess the interview to obtain a new one, useful when the initial result was not accurate or the competencies were incorrectly specified.

What it was like before. The opportunity search module was linked to a specific candidate and offered results without advanced filtering.

What changes. Five improvements that expand its scope and usefulness:

Standalone use with documents. It's no longer necessary to start with a candidate. Now you can upload documents directly—resumes, job descriptions, briefings—and search for opportunities based on them. This separates the module from the core interview functionality and transforms it into a standalone prospecting tool.

New filters. Two filters to refine the results: avoid opportunities to large companies (where it would be difficult for a recruitment consultancy to compete commercially) and avoid opportunities that other recruitment agencies They are already working.

Signs of opportunity. Predefined indicators —active offer, new product, recent investment, geographical expansion…— that act as detection criteria to prioritize the most promising opportunities.

New interface. Complete redesign that enables independent use of the module and improves usability: clearer navigation, better structured results, and quick actions per opportunity.

Export to Excel. The search results can be exported to an Excel table to save them, share them with the team, or work on them outside the platform.

What it was like before. Since the general activation of Plan Mode in Q1, the engine generates significantly more accurate reports, but processing was slower than the previous workflow, especially when many meetings and/or competitions coincided.

What changes. Four optimizations that together double the processing capacity:

Duplication of concurrent meetings. The transcription stage goes from 5 to 10 simultaneous meetings and the summary from 3 to 6This reduces bottlenecks during peak demand.

Parallel skills assessment. Until now it was sequential: each competition took between 1 and 4 minutes, and the total was the sum of all of them. Now they are run in parallel, with a maximum limit of ~4 minutes regardless of the number of competitions evaluated.

Parallel sections (configurable). In the plan ProSection parallelization is always activated. In the plan HRIt is optional and is controlled from the meeting creation or reprocessing dialog, since activating it disables the cross-context between sections, which can slightly affect the quality of the writing.

Elimination of the audit stage in Pro. The loyalty audit consumed a significant amount of resources and is only maintained in the plan. HRwhere quality must always be ensured. In ProThe time savings are immediate.

What it was like before. Each page of the app required opening a new browser tab or going back to change the view.

What changes. Now the app includes its own tabs in the top navigation barEach time a new page is opened, it appears as a tab within the app, up to a maximum of 10 open tabs. When that limit is exceeded, the oldest tab is automatically closed.

The result: moving between interviews, templates, metrics, and settings without losing the context of what was open, and without accumulating browser tabs.