ChatGPT Is Useless for Enterprise Procurement – Here's What Works

01 /The Concept

The market for software products is so saturated that choosing the right tool – for a specific person, a specific workflow, or a specific tech stack – has become genuinely difficult.

- Reviews go stale fast. They're also often written with a promotional agenda, subtly pushing a particular product.

- Google searches and ChatGPT queries pull from the same reviews. Neither can guarantee they've surfaced every product in the market – and there's no way to verify coverage.

- Consultants are expensive and slow. And the quality varies wildly: some are biased, others are behind on recent releases.

Dragonfly has built an AI assistant for software selection – one that knows everything, has no promotional bias, and can compress a selection process to a fifth of the time it takes through traditional methods.

The platform's knowledge base tracks and continuously updates information on 250,000 software products. It charges no fees to vendors for promotion or higher placement in results. Queries are written in plain language – for example, "suggest the best project management platform for a five-person team that integrates with Google Calendar and has a free tier" or "show me a personal data processing platform for financial companies that's fully compliant with EU regulations and can run on AWS."

For immediate compatibility screening, the platform can be pointed at a company's own servers – it will map the existing software stack and filter recommendations accordingly.

Beyond reactive search, the platform can proactively suggest tools the company doesn't yet have but should consider, based on what's already in place – surfacing gaps the team may not know exist.

Selecting software isn't a one-shot query. The assistant supports a full conversation, in the same way ChatGPT does. It can ask clarifying questions, accept additional criteria mid-conversation, compare specific options head to head, and go as deep as the buyer needs.

Dragonfly was founded in December last year. The platform is currently in beta, with a full enterprise version planned for launch by year-end – at which point pricing structure will be announced.

The startup has raised a first round of £2.6 million (approximately $3.5 million).

02 /Why It Matters

The enterprise software buying funnel has changed significantly. In the past, companies largely bought what vendors sold them. Today, they research independently first – reading social media, querying ChatGPT, exploring vendor websites, testing demos.

The result: in 81% of cases, companies compile their software shortlist before having any contact with a sales rep. By the time they reach out, they've already decided who the candidates are and just want help making the final call.

Smart buyers know that social media, Google, and ChatGPT don't give a complete or current view of the options available. Specialized platforms like Dragonfly can fill that gap – functioning as a genuinely comprehensive and up-to-date source.

Dragonfly isn't the only startup to spot this trend.

Stackfix ([related review](/review/pomogi-im-pokupat-na-1-trillion-dollarov-v-god)), another UK startup, raised £2.4 million last December on a similar platform. Its key differentiator: AI-generated recommendations are supplemented by commentary from practitioners who have hands-on experience with the tools being evaluated.

Elion ([related review](/review/novoe-rozhdenie-drugih-marketplejsov)) raised $12.6 million – including $9.3 million in August – on a software selection platform that went deliberately narrow: it focuses exclusively on software for healthcare institutions.

Sagetap ([related review](/review/ne-pomogaj-prodavat-luchshe-pomogi-pokupat)) raised $12.5 million on a platform where buyers can not only research and compare software but also submit questions directly to vendors – while remaining anonymous. That anonymity lets buyers do real due diligence without triggering a sales process before they're ready.

03 /Opportunities

The narrower play is to build a software selection platform using the models and differentiation angles described above. The enterprise software market is enormous, the buying funnel is clearly shifting, and the window to establish a credible position is open.

Worth noting: the term "sales funnel" is starting to lose relevance here. What's actually emerging is a "buying funnel" – companies don't want to be sold to. They want to buy on their own terms.

The broader play is more interesting: build a comprehensive, current, and expertise-backed alternative to ChatGPT for research and selection of anything – not just enterprise software.

The range of possible topics is enormous, from niche hobbies to industrial procurement. Consider audio equipment – headphones, amplifiers, DACs – where hunting through forums for current information is exhausting and ChatGPT regularly hallucinates specs and availability. Any category with that problem structure is a candidate.

What topic would you build this for? And what's actually stopping you – given that the tools and the reference examples already exist?