Bill Fischer (CTO VONQ.) meldde enkele dagen geleden nog cryptisch:

We just might have some interesting announcements in this space….

Met this space wordt LLM & HR bedoeld. Gisteren was dat moment blijkbaar gekomen, al geeft Fischer aan dat het niet een zelfgekozen moment is:

We were keeping this under our hats but, sadly, the excellent engineering team at LinkedIn has just released a paper on Embedding Based Retrieval [EBR] We at @VONQ believe that #EBR is a game changer for #HRTech.

It’s vector based search and if you remember Soh Cah Toa, it’s the Cah that’s at play.

Zoals gewoonlijk bij Fischer het geval is, kan je heel snel het spoor bijster raken. Laat ik voor mezelf praten, ik raak het spoor meestal snel bijster. Maar tegenwoordig komen LLM’s to the rescue, een vraag aan ChatGPT+ leverde dit op:

Supereng plaatje

En er zaten nog heel wat formules bij maar die heb ik maar achterwege gelaten. Het is duidelijk dat dit volstrekt onduidelijk is, behalve voor de mathematici onder ons. Aangezien ik nog nauwelijks mijn kleinzoon van 8 jaar kan bijhouden bij rekenen is dit far above my paygrade.

Gelukkig hebben we in deze situatie ook nog de marketingfunctie (hoewel ik dat niet zonder hoesten kan uitspreken) die met een meer toegankelijke (en zwaar over the top) uitleg komt van Fischers cryptische posts: een persbericht van 9 oktober met als titel VONQ leads the HRTech revolution with the first to deliver a comprehensive ChatGPT Copilot.

Zitten we er klaar voor? Daar gaan we dan de tekst van het persbericht (tekst in bold is ook van VONQ.:

VONQ, the leading provider of recruitment marketing solutions, is pleased to announce the introduction of HAPI Copilot, its most advanced and intuitive suite of Talent Acquisition (TA) solutions.

This revolutionary product marks a significant milestone as VONQ is the first to introduce a full-featured ChatGPT copilot to the Enterprise Human Resources Technology (HRTech) market. HAPI Copilot transforms how talent acquisition teams work, offering several innovative features to simplify and streamline recruitment.

HAPI Copilot meets the growing demand for Gen AI-powered talent acquisition, in line with the intentions of nearly two-thirds of HR teams in the US and Europe, according to recent Bullhorn AI research (October 2023). This innovative solution offers recruiters the potential for over 40% productivity gains by providing real-time candidate assessment and communication, effectively addressing the issue of slow response times, a common cause of candidate drop-off.

For the successful deployment of Generative AI, distribution is everything. CEO Arno Schaefer said: “Embedding these tools into our partners’ existing HR workflows is at the heart of our strategy. Any point solution will struggle to gain traction with enterprise customers in this environment.”

VONQ HAPI Copilot seamlessly integrates with ATS and HCM platforms through VONQ’s Hiring API (HAPI) and will be available to more than 80% of US/EU talent acquisition teams through VONQ’s 100+ ATS partners.

Bill Fischer, CTO of VONQ, further highlighted the superiority of the transformer-based architecture in HAPI Copilot, stating, “At every step, using the transformer-based architecture as a foundation was better than the current state of the art AI/ML. It wasn’t even close.”

Wat hiervan te maken? Twee dingen. Allereerst ga ik VONQ vragen of ik hier eens mee mag spelen. En ten tweede, ik heb het persbericht aan ChatGPT+ aangeboden met de vraag om daar een evaluatie van te maken. En dit is daarvan het resultaat:

Strengths:

1. First-Mover Advantage: VONQ claims to be the first to introduce a full-featured ChatGPT copilot to the Enterprise Human Resources Technology market, potentially giving them a competitive edge.

2. Integration: The suite aims to seamlessly integrate with existing ATS and HCM platforms, addressing a crucial point for many enterprises that rely on existing systems.

3. Data-Backed Demand: The press release cites Bullhorn AI research to justify the product’s relevance, adding credibility to the offering.

4. Productivity Gains: Promises of over 40% productivity gains are compelling and directly address a key pain point in talent acquisition—slow response times.

5. Technical Superiority: The mention of transformer-based architecture adds a layer of technological sophistication.

Weaknesses:

1. Buzzword-Heavy: The press release leans heavily on buzzwords like “revolutionary,” “Gen AI-powered,” and “transformer-based architecture,” which could be perceived as marketing fluff unless substantiated with empirical evidence.

2. Lack of Specifics: While the release mentions productivity gains and integration with existing systems, it lacks specific details on how these will be achieved.

3. Unverified Claims: Statements like “it wasn’t even close” regarding the superiority of their AI model may raise scepticism unless backed by comparative data or third-party validation.

4. Limited Scope: Although the press release cites research, it only focuses on the U.S. and EU markets. It would be beneficial to understand the product’s applicability in other geographical markets.

5. Assumption of Distribution Ease: CEO Arno Schaefer mentions that “distribution is everything,” but there’s no information on how they plan to overcome potential distribution challenges, especially given that any point solution will struggle to gain traction.

Questions Raised:

1. Validation: How are the productivity gains measured?

2. Compatibility: How seamlessly does the system integrate with various ATS systems, and what’s the transition experience for a company switching to or adding HAPI Copilot?

3. Data Security: Given that AI will be parsing and reasoning, how secure is the candidate data?

4. Cost: What is the cost structure for implementing HAPI Copilot?

Overall, while the press release makes ambitious claims that could revolutionize talent acquisition processes, critical evaluation and third-party validation are essential to assess the true value and applicability of HAPI Copilot.

Slotopmerkingen

Wat leert dit ons? Ten eerste dat ik voortaan mijn reviews van persberichten door ChatGPT laat schrijven. Biasfree en wie is nu de azijnblogger? Ten tweede dat VONQ. de plank misslaat door nogal Amerikaans ronkend op de borst te kloppen zonder met bewijs te komen. En dat is zonde, want voor hetzelfde geld staan we aan de vooravond van een evolutie (inderdaad, geen revolutie) van het recruitment domein (inderdaad, niet HRTech).

Ik ga in ieder geval niet wachten, want er is hopelijk wat onderzoek te doen!