All Case Studies

CASE STUDY

M&A / ADVISORY

Broker introduces independent technical review — inflated AI claims exposed before heads of terms

Result

AI claims corrected pre-HoT. Deal closed on accurate terms.

Timeline

3 weeks

Sector

Technology Services — GCC

The Context

A boutique M&A advisory firm based in Dubai was managing the sale of a regional technology services business — a 45-person company offering managed IT and, increasingly, AI-assisted workflow automation to mid-market clients in the GCC. The asking price was positioned at a premium, in part because the seller had been marketing the business as having a proprietary AI platform powering its service delivery.

The acquirer — a UAE-based private equity group — had proceeded through initial conversations and was approaching heads of terms. Their investment committee, however, was uncomfortable with the AI narrative: the platform was central to the valuation justification, and no one on either side had independently verified what it actually was or did. The broker, concerned about a late-stage collapse, introduced Polar Frequency to conduct a focused technical assessment before HoT were signed.

The Challenge

AI claims in deal processes are increasingly common and increasingly difficult to verify without technical expertise. Sellers — and sometimes their advisors — use language like "proprietary AI platform," "machine learning pipeline," and "automated intelligence" in ways that buyers cannot readily interrogate. For a broker, the risk is asymmetric: if technical problems surface post-HoT, the retrade damages both sides and costs the advisor the relationship.

The broker needed a clear, commercially framed view of the technology before commitment — not a lengthy technical audit, but a focused, fast assessment that could be shared with both the buyer's investment committee and, selectively, with the seller's team to reset expectations if needed.

Our Approach

We scoped a three-week technical review structured around the specific claims being used to justify the premium valuation:

Phase 1

Claim mapping

We extracted every substantive AI and technology claim from the seller's information memorandum, pitch materials, and the oral representations captured in meeting notes shared by the broker. Each claim was categorised as verifiable, unverifiable without access, or structurally implausible.

Phase 2

Technical access and review

With NDA in place, we conducted two structured sessions with the seller's technical lead, reviewed the platform architecture, source code samples, infrastructure setup, and the data pipelines underpinning the "AI" functionality. We examined three live client deployments and one recently delivered project.

Phase 3

Commercial-language findings report

We produced a findings document structured for a non-technical audience: each claim rated as substantiated, partially substantiated, or not substantiated, with a plain-language explanation of what the technology actually was, what it did, and what the gap between claim and reality meant in commercial terms.

What We Found

Of the seven substantive AI and technology claims made in the information memorandum, two were substantiated in full, three were partially substantiated, and two were not substantiated:

Not Substantiated

"Proprietary AI platform" was a configured third-party SaaS product

The platform referenced throughout the IM was a customised deployment of a commercially available workflow automation tool. No proprietary model had been trained. No custom ML infrastructure existed. The seller's team had built client-specific configurations and integrations, which had real value — but these were not a "proprietary AI platform" in any meaningful sense.

Not Substantiated

"AI-driven insights" in client reporting were rule-based templates

The "AI-generated" monthly reports delivered to clients were produced by a series of conditional logic rules and templated outputs, not a machine learning model. The outputs were consistent and commercially useful — but describing them as AI-driven was misleading, and client contracts made no reference to any AI functionality.

Partially Substantiated

Automation coverage was real but significantly narrower than claimed

The seller claimed automation covered "80% of routine service delivery." In practice, automation handled approximately 35–40% of delivery tasks across three service lines, with significant manual intervention still required in the remaining lines. The automation that did exist was solid and defensible — the headline figure was not.

The Outcome

The broker shared a summary version of the report with both sides before heads of terms were finalised. The seller's principal accepted the findings without dispute — acknowledging that the marketing language had "gotten ahead" of the underlying technology. The deal proceeded, but on corrected terms.

The AI premium was removed from the valuation basis — price adjusted to reflect a well-configured managed services business rather than a proprietary AI platform

The two substantiated automation capabilities were retained as genuine value drivers and documented accurately in the revised IM

Heads of terms were signed within two weeks of the report being shared, with no further technical disputes during formal diligence

The broker preserved relationships with both sides — framing the correction as a mutual benefit rather than an adversarial finding

The buyer's investment committee approved the revised deal without condition, having seen the technical picture before commitment

"Bringing in Polar Frequency before HoT was the right call. We got a clean, readable report that both sides could work with. The deal didn't fall apart — it just got honest. That's a much better outcome than finding the same problems six weeks later."

— M&A advisor, Dubai

Managing a deal where technical claims need independent verification?

We give brokers and advisors a fast, readable technical perspective — before heads of terms, not after.

Talk to Us

NDA available before any materials are shared. Responds within one business day.

We use analytics cookies to understand how visitors use our site. No advertising or tracking cookies are set. Cookie policy.