In AI answers, Shell is a transition and emissions-disclosure name, seldom a service or reliability one

This is a snapshot from 29 September 2026: 250 questions that never named a brand, put to ChatGPT, Claude, Gemini, Perplexity and Grok. Shell was named in 160 of the 1,250 answers. Grok named it most often (49 of 250) and Claude least (21 of 250). ExxonMobil (169) and Chevron (158) sit in nearly the same band, so Shell's overall presence is typical for the group. What differs is where it shows up.

By question intent, Shell's visibility was highest on leadership and sustainability (31% each), followed by employer (23%), investment returns (22%) and innovation (21%). It was near zero on service, plan and billing, reliability and safety, and 0% on clean energy options. These figures cover every answer. The 14 stored answers reviewed here are only a sample.

The emissions-reporting answers gave Shell a place among the more forthcoming majors. Claude's answer said "BP and Shell position themselves as leaders in Scope 3 disclosures," then added that these companies "continue to invest heavily in fossil fuel infrastructure." Perplexity's answer listed Shell among four majors that publicly reported Scope 1, 2 and 3 emissions in a 2022 review. The same answer said none provided adequate data to verify their claims.

On transition spending, ChatGPT's partial answer said Shell's low-carbon platform capital of 20.3 exceeded its buybacks of 13.9. It called Shell the only major where that held, and cautioned that this was not purely renewables. Gemini grouped Shell with ExxonMobil among majors with high investment-grade credit ratings. Perplexity's layoffs answer said Shell has pursued restructuring expected to cut its workforce by about 20%.

Consumer-facing questions were different. In three Grok answers about EV charging networks, Shell Recharge appeared in a list of networks with higher problem rates, at about 48% of sessions in the Consumer Reports 2025 data Grok cited. Gemini's cheapest-gas answer led with warehouse clubs and mentioned Shell only in passing, as a chain with loyalty programs.

PeakMetrics' news index shows 16,227 Shell mentions over the last 30 days, 2,377 of them with Shell in the headline. That is more than BP (17,714 mentions is higher, but BP was named in only 120 answers) and roughly half of Chevron's 30,821.

The five recent headlines supplied are about transactions and projects. They cover a planned expansion of company-owned convenience retail sites through the Tri Star Energy acquisition, the sale of the Rhode Island State Energy Center to Constellation Energy for $715M, and BP's share price amid a Shell asset sale. They also cover a Trinidad natural gas project put on hold and an industrial-scale electric heater deployed at the Rheinland energy and chemicals park.

The press story is about deals, divestments, retail expansion and project decisions. The AI story is about how Shell reports emissions, how its low-carbon spending compares with buybacks, and how its workforce and credit profile compare with other majors. None of the five supplied headlines touches emissions disclosure.

The retail expansion in the headlines has a counterpart in the answers, but not a matching one. The gasoline answer sampled here gave Shell only a passing mention, and the EV charging answers surfaced a problem-rate figure from a third-party survey. On the day of this snapshot, coverage volume and answer content were driven by different subjects.

A comms team would take three points from this snapshot. First, the assistants tie Shell's name to disclosure and transition-spend questions, so the accuracy of those third-party comparisons matters, including the caveats the answers attach, such as "not purely renewables" and "asset sales."

Second, the EV charging figure is a survey number relayed through Grok. It is worth tracing to its source, since the answers cited it without qualification. Third, the retail and charging questions are where Shell was barely present, even as the press headlines point to retail growth.

These are single runs of live answers, and results vary between runs. Repeat the benchmark questions before treating any of this as a pattern.