A facilities engineer at an operator in KL asks ChatGPT which contractors can do ROV inspection and riser repair on platforms off Bintulu. The answer names three firms. Two are your direct rivals, one is a company you have never seen bid, and yours is not there, even though your crews have worked those fields for years. That is the gap in Malaysia offshore AI shortlists: the names an assistant offers before anyone opens a tender portal.

This article is for marketing and business development leaders at inspection, maintenance and repair (IMR), diving, survey and light construction contractors. It explains why AI tools favour some offshore firms, the prompts buyers type, an audit you can run this week, and the specific fixes that make your capabilities easier to find and quote.

Why rivals win Malaysia offshore AI shortlists

Rivals win because AI assistants shortlist from public text that matches the prompt, and their text names basins, scopes and assets while yours says "integrated offshore solutions". Tender history and client trust do not reach the model unless they are written down somewhere it can read.

Two inputs: training and live retrieval

An assistant answers from what the model learned in training, which is fixed at a point in time, and from live retrieval, where it searches the web and cites what it finds. The engines document the second route. OpenAI says OAI-SearchBot is used to surface websites in ChatGPT's search features, and that sites which opt out of it will not be shown in ChatGPT search answers. Perplexity says PerplexityBot is designed to surface and link websites in its search results.

Google says that to appear as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to show with a snippet, and it lists making important content available in textual form as a basic practice. It also describes "query fan-out", where several related searches run across subtopics. A prompt about "saturation diving Sabah deepwater" can fan out into searches on the scope, the basin and named fields. The firm with a page answering each of those is the one that gets cited.

Why offshore contractors are hard to match

Offshore service firms tend to publish the least. Project details sit under client confidentiality, capability statements are PDFs sent by email, and the website leads with a vessel photo and a slogan. Meanwhile, a rival's project announcement, a conference paper or a trade article states that it completed pipeline inspection in the Malay Basin. That one sentence is a better match than your entire homepage.

The market adds pressure. The Star reported in February 2026 that analysts expect a gradual downtrend in offshore maintenance, construction and modification (MCM) activity over 2026 to 2028, with upstream maintenance, drilling and OSV guidance for 2026 lower. When fewer packages are let, being on the informal shortlist early matters more.

What buyers in Malaysia actually ask AI

Offshore buyers write prompts the way they write emails to colleagues: a scope, a basin or field area, an asset type and sometimes a licence condition. Most are in English; some local contract staff use Bahasa Malaysia.

Early scoping

  • "Which Malaysian contractors do ROV inspection and riser repair offshore Sarawak?"
  • "Saturation diving contractors with experience in Sabah deepwater fields"
  • "Kontraktor IMR berlesen PETRONAS untuk platform di Terengganu" (PETRONAS-licensed IMR contractors for platforms off Terengganu)

Comparing firms

  • "Compare [your firm] and [rival] for pipeline survey and light construction in the Malay Basin"
  • "Which offshore survey companies in Malaysia have their own DP2 vessels?"
  • "Who has done hook-up and commissioning work in Central Luconia?"

Checking a name

  • "Does [your firm] hold PETRONAS licences for subsea inspection work?"
  • "Recent offshore projects by [your firm] in Malaysia"

Only the last two name you directly. The rest are open questions where the assistant decides who belongs on the list, and those are where rivals with basin-specific content take the slots.

How to check where you stand

You can build a baseline in a week with a spreadsheet and two people from business development and operations. Our measurement method uses the same structure with larger prompt panels.

Step 1: Build a basin and scope grid

List your real scopes down one side (ROV inspection, diving, pipeline survey, light construction, MCM, HUC, decommissioning support) and the waters you work in across the top: Peninsular Malaysia, Sarawak, Sabah, and cross-border areas if relevant. Each filled cell gets two or three prompts in the phrasing your clients use.

Step 2: Check crawler access

Open robots.txt on your corporate site and any subsidiary domains and look for rules covering OAI-SearchBot, PerplexityBot and Googlebot. Then ask IT whether a firewall or bot-protection service challenges them. Engineering firms often lock sites down after a security review and block search crawlers along the way.

Step 3: Run each prompt several times

Run every prompt three times in ChatGPT, Gemini and Perplexity, in a clean session. Answers vary between runs and users, so one result proves little. Note whether the assistant searched the web, since that tells you live pages are in play.

Step 4: Log what matters

For each run, record whether you were named, whether the scope and waters described were right, which competitors appeared and which sources were cited. The cited sources column is the most useful: it shows which news sites, papers, directories and rival pages the engines trust for each basin.

Step 5: Test your own pages

For each prompt where you were missing, open the page that should answer it. Could a stranger tell, from visible text alone, that you do that scope in those waters? If the answer lives in a PDF or a sales deck, that cell is your first fix.

How to fix it

Give AI tools a clear, public statement of what you do and where, in the vocabulary buyers use, then make sure third parties say the same. The fixes below suit IMR, diving, survey and light construction firms.

Publish basin and scope pages

Create one HTML page per major scope, and a short section or page per region where you have real experience. Open each with a plain sentence such as "We provide ROV inspection, repair and maintenance for fixed platforms and pipelines offshore Sarawak and Sabah." Then list asset types, water depths you have worked in, typical vessel spreads and the year range of your work. Where client names are confidential, describe the asset class and region instead. Our content work focuses on pages like these.

Use the licensing vocabulary buyers know

PETRONAS explains that a licence allows a company to supply goods or services to both upstream and downstream, while registration covers downstream only, and that SWEC is the list of work categories used across upstream and downstream. Buyers think in those categories. If you hold current licences, state your licence status and the work categories in words on your capability pages, and keep it updated. Do not paste internal codes without a description; a human and an AI tool both need the plain-language scope next to it.

Make fleet and equipment facts quotable

Describe your vessels and equipment at the level you are comfortable sharing: vessel type, DP class, ROV class, diving system type, home port or supply base. A short table works well:

Fact What to state
Scopes IMR, diving, survey, MCM, HUC, decommissioning support
Waters Peninsular Malaysia, Sarawak, Sabah, other regions
Assets Fixed platforms, pipelines, risers, FPSOs, subsea structures
Fleet and kit Vessel type, DP class, ROV class, diving systems
Bases Kemaman, Labuan, Bintulu, Miri, or wherever you operate
Licensing Current licence status and work categories, in words

Fix the entity basics

State your legal name, trading name, parent or joint venture partners and head office the same way everywhere: your site, LinkedIn, Bursa announcements if listed, and industry directories. Google's Organization structured data documentation describes how to mark up a company's name, logo, address and other identifiers. Markup must match visible text. It does not guarantee a mention, but it removes doubt about who you are. Our technical work usually starts here and with crawler checks.

Earn coverage that names basin and scope

Use the cited sources from Step 4 as your target list. In Malaysian offshore work that tends to mean business press reporting contract awards, trade publications, conference proceedings and the LinkedIn articles engineers share. When you win or complete a package, ask whether a short public note is permitted, and make sure it names your firm, the scope and the waters in the first lines. A technical paper by your engineers on a real problem, such as riser inspection in a specific environment, often outlasts a press release as a citation. Our authority work covers this.

Keep Bahasa Malaysia consistent

If you publish a Bahasa Malaysia version for local partners, translate the scope and region terms consistently and keep company and vessel names identical. Inconsistent translations can make one firm look like two.

What to measure

Track mention rate, accuracy and citation share on your basin and scope grid every month, using the same prompts. Read the results as trends, not verdicts.

  • Mention rate: share of prompt runs where you are named, split by scope, region and engine.
  • Accuracy: share of mentions where your scope, waters and licensing are described correctly. Fix errors first; a wrong scope sends the wrong enquiries.
  • Citation share: how often your own pages appear among the cited sources, against rivals, press and directories.
  • Competitor presence: which firms appear in cells where you do not, and which sources back them.

Report the method with the numbers: prompts, runs, engines and dates. Expect movement to be uneven. A new Sarawak page may start being cited in one engine within weeks and not appear in another for months.

Common mistakes

  • Leading with slogans. "Integrated offshore solutions" matches no buyer prompt. Scope plus basin plus asset does.
  • Keeping capability statements in PDFs emailed to buyers instead of published as text.
  • Blocking crawlers by accident through firewall rules set during a security review.
  • Listing licence codes without descriptions, or leaving lapsed licence claims online.
  • Publishing nothing about completed work when a short, client-approved note was possible.
  • Judging visibility from one screenshot instead of a repeatable prompt grid.

How Shubi helps

Shubi is an AI visibility company based in Malaysia. For offshore contractors we build basin and scope prompt grids, check crawler access, trace which sources the engines trust for each cell, and help publish capability pages buyers and AI tools can quote. Our AI Visibility Intelligence service is the usual starting point, and we do not promise rankings or guaranteed mentions. Message Shubi on WhatsApp