AI arrives in valuation: sentiment, speed and scepticism

Private markets used to move to a quarterly rhythm; increasingly they do not. As reporting cycles compress and AI-enabled tooling spreads through valuation workflows, the profession is doing something sensible before declaring a direction: measuring its own sentiment. The early readings suggest expectations are running well ahead of readiness.

The end of the quarterly rhythm

The clearest account of the shift comes from Fair Value in the Fast Lane, an IVSC interview with Abhishek Pandey of fintech firm 73 Strings. Private markets, he argues, are no longer quarterly. A poll run during an IVSC/73 Strings webinar found that over half of respondents expect monthly, weekly or on-demand valuations — yet fewer than one in six firms feel equipped to re-run valuations in real time during macro shocks.

That gap between expectation and capability is the story. Demand for faster valuation is arriving before the governance, data pipelines and review processes needed to deliver it credibly — particularly during periods of heightened market volatility, when rapid repricing of assets puts the most strain on process.

Taking the profession’s temperature

Rather than guessing how practitioners feel about AI, the IVSC is tracking it. The Valuation Pulse, launched with 73 Strings in May 2026, is an ongoing sentiment tracker on AI and technology in valuation. It is open to valuers, firms, clients, investors, regulators and academics; responses are anonymised; and the findings feed the IVSC’s AI and Technology Working Group.

Sentiment data will not settle the open questions, but it shows standard-setters where uncertainty and unease are concentrated — which is exactly the information a standards programme needs. It also gives firms a benchmark: knowing how peers, clients and regulators are thinking about AI helps a valuation team judge whether its own caution — or its own enthusiasm — is out of step with the wider market.

Judgement stays with the valuer

On the central question — what happens to professional judgement when models do more of the work — the direction is consistent. Proposed revisions to IVS 105, discussed in the IVSC’s article on private credit valuations, make explicit that judgement stays with the valuer, not the model.

Pandey’s formulation is blunter: “A valuation that cannot be explained or traced back to its assumptions isn’t a valuation — it’s an output.” His broader point, and the IVSC’s, is that AI should augment judgement, not replace it. Speed is welcome; unexplainable speed is not. For valuation teams, that turns the AI question into a documentation question: however a number is produced, the assumptions, data and reasoning behind it must remain visible and reviewable.

Standard-setters moving together

Notably, this is not one body’s position. The IPEV Valuation Guidelines 2025 — published in December 2025 and effective for periods from 1 April 2026 — include a Valuation Standards section acknowledging the relationship with the IVSC, and the IPEV Board records that it concurs with the conclusion of the IVSC Standards Review Board on automated valuation models and AI.

For practitioners, the signal is reassuringly boring: the tools will keep changing, but the expectations — explainability, traceable assumptions, accountable human judgement — are converging across standard-setters. Firms investing in AI capability should invest in exactly those disciplines alongside it.