Skip to content
GLOBALANCE INTELLIGENCEThe economy ahead.
This Week Index 28/100 Sign in
Money & Investments › GI 100 Index
The Great Value ShiftInvestment ThemesGI 100 IndexGI PortfolioStart Future Check
AI FOOTPRINTlimitedASSESSMENT 2026GLOBALANCEINTELLIGENCE
Globalance AI Footprint

Assessment as of November 2026 · next assessment 2027.

Result of the assessment by the Globalance research team under the published methodology. Not a certification or quality label.

How we assess ›

How this rating is builtVideo · 0:52
GI 100 Index · Profile

Comfort Systems USA

USAEnablersAssignment FullData quality Very high

Key messagelimited AI Footprint

Comfort Systems USA builds HVAC, electrical and modular plants, including for data centres, and is piloting a voice assistant using artificial intelligence that helps service technicians with troubleshooting and passes uncertain cases on to experts. For protecting its IT, the company describes oversight by the board and regular reviews, but publishes no rules of its own on the use of artificial intelligence.

Opportunities

  • Safetysolid

    The cyber programme is described in concrete terms in the company's own mandatory publications: board oversight, a CISO reporting directly to management and separate from IT, a Risk Committee and at least annual information of the Audit Committee (S04); audits, tabletop exercises, threat modelling, …

    The cyber programme is described in concrete terms in the company's own mandatory publications: board oversight, a CISO reporting directly to management and separate from IT, a Risk Committee and at least annual information of the Audit Committee (S04); audits, tabletop exercises, threat modelling, vulnerability testing and independent reviews reported to the board, and zero client data losses in 2024 and 2025 (S05, S06); Item 1C is available only in a secondary account (S02, S03). No cyber incidents or regulatory proceedings are documented in the research period (S14, S21). For the voice AI pilot, escalation to humans and reliance on manufacturer manuals are documented by the company itself (S05), while testing and accuracy (60–70%) are available only as a vendor statement (S09); information security training coverage was only 19% in 2025 (S05). The rating “solid” rests essentially on sub-criterion b, is not AI-specific and is flagged for review.

Risks

  • Responsibilitylimited

    General risk governance is documented (board oversight of risk management, a management Risk Committee including the CISO, annual review of the ERM matrix by the Audit Committee, a Compliance Committee, an anonymous helpline; S04, S05, S07), but it contains no AI-specific responsibility, policy or approval.

    General risk governance is documented (board oversight of risk management, a management Risk Committee including the CISO, annual review of the ERM matrix by the Audit Committee, a Compliance Committee, an anonymous helpline; S04, S05, S07), but it contains no AI-specific responsibility, policy or approval. Although the group runs a voice AI pilot and is planning AI call routing (S05), the proxy statement, code of conduct and privacy policy do not mention AI (S04, S07, S08); the 10-K names the misuse of AI tools by employees only as a risk (S01). Indications of data governance for AI come from a job advertisement (S25).

  • Transparencylimited

    In its 2025 sustainability report, Comfort Systems discloses a voice AI pilot including its escalation principle and planned AI call routing (S05), but names neither the vendor, scope of use, accuracy nor limits; the accuracy figure comes only from the vendor (S09).

    In its 2025 sustainability report, Comfort Systems discloses a voice AI pilot including its escalation principle and planned AI call routing (S05), but names neither the vendor, scope of use, accuracy nor limits; the accuracy figure comes only from the vendor (S09). The 10-K mentions AI only as a risk (S01), and the privacy policy contains no statement on automated decisions, AI or voice data (S08). Model and system cards are not applicable as there are no in-house models; the remaining disclosure is limited to general statements.

  • Risk area

    AI-assisted troubleshooting of building services in the field, with possible safety impact

  • Risk area

    Protection of client, plant and personal data across more than 45 operating companies

  • No serious controversy recorded.

What lies behind the assessment

Research of 3 November 2026
41searches
31pages and documents read
8primary documents
0controversies checked
5open points, named openly
The full research: pillars, evidence, controversies and open points
AI Footprint · Enabler · As of 3 November 2026
Comfort Systems USAOverall rating: limited

Comfort Systems USA builds building services and modular plants for data centres and is piloting a voice AI assistant for service technicians. Its cyber programme is described, but no AI policy or AI responsibility is documented.

Responsibility

40 %
limitedsolidstrong

General risk governance is documented, but there is no AI policy or AI responsibility.

  • +Risk Committee with CEO and CISO reports on the risk matrix Source ↗
  • +Anonymous 24-hour helpline with a ban on retaliation Source ↗
  • –No AI policy and no approval procedure for AI found
  • –Proxy, code of conduct and privacy policy do not mention AI Source ↗
  • –Business ethics training coverage only 4% in 2024 Source ↗
All sources for this pillar (8)

Safety

40 %
limitedsolidstrong

The cyber programme is described in detail, but AI testing exists only as vendor claims.

  • +CISO reports directly to management, separate from IT Source ↗
  • +Audits, tabletop exercises, vulnerability tests and independent reviews Source ↗
  • +Voice AI escalates uncertain answers to experts Source ↗
  • –Voice AI accuracy of around 60–70% only as a vendor statement Source ↗
  • –Only 19% of employees trained in information security in 2025 Source ↗
All sources for this pillar (9)

Transparency

20 %
limitedsolidstrong

The AI pilot is disclosed, but not its vendor, scope, accuracy or limits.

  • +Sustainability report describes the AI pilot and planned call routing Source ↗
  • –AI vendor known only from its own marketing Source ↗
  • –Privacy policy says nothing about AI or voice data Source ↗
  • –No rules on labelling AI to callers
All sources for this pillar (5)
What is missing for “strong”

These steps are in the company’s own hands.

1Responsibility
Set an AI policy and responsibility

Publish an AI policy with an approval procedure for purchased AI and name a responsible function in management.

2Safety
Document own testing of the voice AI

Publish its own validation, accuracy measurement and monitoring of AI troubleshooting, including how safety-relevant faulty guidance is handled.

3Transparency
Disclose AI use to customers

Disclose vendors, limits and data use of its AI applications and label the AI call routing to callers.

Context: 101 companies assessed

Distribution of overall ratings, no ranking
strong
none yet
solid
limited
← Comfort Systems USA is one of 66

No company reaches “strong” today. Disclosure on how companies handle AI is still young across the market.

Controversies
0 none found

No deduction for controversies.

Largest risk area
Faulty AI guidance in troubleshooting

The voice AI assistant “Mark” gives technicians troubleshooting help on equipment, while an internal test accuracy of around 60–70% is available only as a vendor statement.

How thoroughly assessed
22 Evidence

8 primary documents read, 41 searches.

Read the full reasoning and all sources
Comfort Systems USA · Full reasoning

Documented cyber oversight, but no rules for the use of AI

Overall rating “limited”. No deduction for controversies. The superscript numbers link directly to the source.

1 · Responsibility

limitedWeight 40%

Risk oversight by the board and management is documented, but it contains no AI-specific responsibility, policy or approval, although the group already uses AI.

What is documented
  • The board oversees risk management; a Risk Committee including the CEO and CISO reports on the risk matrix at least annually1 2
  • The Audit Committee reviews the risk matrix annually; the skills matrix lists technology, cybersecurity and innovation experience1
  • An anonymous helpline, a ban on retaliation and ethics audits at all sites are described2 6
  • The 10-K names the misuse of AI tools by employees as a risk5
The case against
No AI rules or responsibilities
  • No AI policy, risk classification or approval procedure for in-house or purchased AI was found
  • There is no named AI responsibility at board or management level; only an operational innovation unit is visible1 3
  • The code of conduct warns against unapproved cloud services without mentioning AI6
Weak embedding
  • Business ethics training coverage was 4% in 2024; no AI training or AI figures were found8
  • No external audit or certification of AI governance was found
Sources for Responsibility (8)
  1. Comfort Systems USA, Inc. (SEC EDGAR): Definitive Proxy Statement (DEF 14A) 2026 · 18 May 2026 · primary source
  2. Comfort Systems USA, Inc.: Sustainability Report 2025 · 3 November 2026 · primary source
  3. Aquant: Customer Story: Comfort Systems · 3 November 2026 · secondary source
  4. The Service Council: Preparing Your Service Organization for Conversational AI: From Use Case to Real Impact (Webinar) · 3 November 2026 · secondary source
  5. Comfort Systems USA, Inc. (SEC EDGAR): Form 10-K for the fiscal year ended December 31, 2025 · 19 February 2026 · primary source
  6. Comfort Systems USA, Inc.: Code of Conduct (Last Revised 8/5/2026) · 5 August 2026 · secondary source
  7. Teal (jobs portal): AI & Automation Engineer – Comfort Systems USA, Houston, TX · 3 November 2026 · secondary source
  8. Comfort Systems USA, Inc.: Sustainability Report 2024 · 3 November 2026 · primary source

2 · Safety

solidWeight 40%

The rating rests on a generally described cyber programme that is not AI-specific, while testing and accuracy of the voice AI come only from the vendor.

What is documented
  • The CISO reports directly to management and works separately from IT; the Audit Committee is informed at least annually3
  • Audits, tabletop exercises, threat modelling, vulnerability testing and independent reviews reported to the board are described1 4
  • According to the company, there were no client data losses and no privacy complaints in 2024 and 20251
  • The voice AI “Mark” relies on more than 700 manufacturer manuals and hands uncertain cases to Fix Center experts1
  • No cyber incidents and no regulatory actions are recorded in the research period7
The case against
AI testing only from the vendor
  • Stress testing and test accuracy of around 60–70% come only from the vendor Aquant; the company's own test results are missing2
  • The voice AI learns from interactions without any monitoring being described1
Gaps in system protection
  • Item 1C is available only in a secondary account; no framework or certifications are named5 8
  • Only 19% of employees were trained in information security in 2025, after 8% in 20241 4
  • No public channel for reporting vulnerabilities and no rules on personal data in AI tools were found
Sources for Safety (9)
  1. Comfort Systems USA, Inc.: Sustainability Report 2025 · 3 November 2026 · primary source
  2. Aquant (Micaela McPadden): Comfort Systems USA is Redefining Service with Voice AI · 28 October 2025 · secondary source
  3. Comfort Systems USA, Inc. (SEC EDGAR): Definitive Proxy Statement (DEF 14A) 2026 · 18 May 2026 · primary source
  4. Comfort Systems USA, Inc.: Sustainability Report 2024 · 3 November 2026 · primary source
  5. board-cybersecurity.com: COMFORT SYSTEMS USA INC 10-K Item 1C. Cybersecurity – 2026-02-19 · 19 February 2026 · secondary source
  6. Comfort Systems USA, Inc. (SEC EDGAR): Form 10-K for the fiscal year ended December 31, 2025 · 19 February 2026 · primary source
  7. board-cybersecurity.com: Company tracker: Comfort Systems USA · 3 November 2026 · secondary source
  8. Boardroom Alpha: FIX 10-K 2026-02-19 (account, accession 0001104659-26-017530) · 19 February 2026 · secondary source
  9. Comfort Systems USA, Inc.: Comfort Systems USA Announces First Quarter 2019 Conference Call and Webcast; Discloses Ransomware Attack · 18 April 2019 · primary source

3 · Transparency

limitedWeight 20%

Comfort Systems discloses the voice AI pilot and the planned AI call routing but names neither the vendor nor accuracy or limits, and its privacy policy is silent on AI.

What is documented
  • The 2025 sustainability report describes the AI pilot “Mark” and planned voice call routing1
  • The report names more than 700 manufacturer manuals and HVAC fundamentals as the pilot's data basis1
  • The 10-K names AI as a competitive and cyber risk5
The case against
Missing information on AI use
  • The company does not itself name the vendor, scope of use, accuracy or limits of its AI applications1 2
  • Whether technicians' interaction data is used for training by the vendor is not documented
Customers not informed
  • The privacy policy says nothing about automated decisions, AI or voice data4
  • No rules on labelling the planned AI call routing to callers were found
Sources for Transparency (5)
  1. Comfort Systems USA, Inc.: Sustainability Report 2025 · 3 November 2026 · primary source
  2. Aquant (Micaela McPadden): Comfort Systems USA is Redefining Service with Voice AI · 28 October 2025 · secondary source
  3. Aquant: Customer Story: Comfort Systems · 3 November 2026 · secondary source
  4. Comfort Systems USA, Inc.: Privacy Policy (Last Updated and Effective Date: September 18, 2026) · 18 September 2026 · secondary source
  5. Comfort Systems USA, Inc. (SEC EDGAR): Form 10-K for the fiscal year ended December 31, 2025 · 19 February 2026 · primary source

4 · Why the harm rule does not apply

The harm rule does not apply: there is no judgment, authority decision, fine or pattern of harm from AI. No controversy relating to AI, data or cybersecurity was found in the period.

5 · Controversies (0)

No AI controversy found.

6 · What remains open

  • AI-specific responsibility at board or management level
  • Group-wide rules for the use of generative and purchased AI, such as a policy, approvals and approved vendors
  • The company's own information on testing, accuracy and monitoring of the voice AI and the Aquant triage
  • Handling of voice and interaction data and labelling of the planned AI call routing
  • Original text of Items 1C and 3 of the 2025 10-K and the legal information in the 10-Q for the third quarter of 2026

7 · How it was assessed

Cut-off date: 3 November 2026 · 41 searches, 31 fetches.

Primary documents read (8)
  • Form 10-K FY 2025 (cover page, Item 1, Item 1A up to 94,555 characters, four retrievals)
  • Definitive Proxy Statement 2026 (up to and including the CD&A, two retrievals)
  • Sustainability Report 2025 (up to 100,000 of 116,577 characters, two retrievals)
  • Sustainability Report 2024 (up to 100,000 of 104,467 characters)
  • Form 10-Q Q3 2026 (up to Note 3, three retrievals)
  • Form 8-K voting results AGM 2024 and AGM 2025
  • Code of Conduct (as of 2026-08-05), Privacy Policy (as of 2026-09-18), Sustainability website
  • NLRB-Fallseite 28-CA-373300; Justia-Docket 4:2026cv00849; Leagle-Auszug 2024 Ark. App. 491

Research of 3 November 2026 under Globalance AI Footprint rulebook version 0

We assess documented practice in dealing with AI; this is not an investment recommendation. Three pillars, weights 40/40/20, harm rule.

Research of 3 November 2026 under Globalance AI Footprint rulebook version 0. Documented observations from public sources, no investment recommendation.

Important Notice. This material is provided for information purposes only and does not constitute an offer or solicitation to buy or sell any financial instrument, nor does it constitute investment advice or a recommendation. The GI 100 Index is a rules-based strategy index of Globalance Bank. Investments involve risk, including the loss of capital invested. The final product documents (including FinSA) shall prevail in all cases.

Your company? Information on inclusion in the GI 100 Index →

What comes next