The Entity Brief: the one paragraph that decides whether AI recognises your firm

Executive summary

An Entity Brief is a single canonical paragraph, roughly 80 to 120 words, that states who a firm is, who it serves, where, and what makes it different, and is then deployed word for word on every surface an AI engine reads. AI engines do not rank firms; they name the ones they can describe with confidence, and confidence comes from seeing the same facts about the same entity across many independent sources. Most professional firms have five or six slightly different descriptions of themselves scattered across directories, LinkedIn and their own site, which reads to an engine as five weak entities rather than one strong one. The Entity Brief fixes this. It is the highest-leverage, lowest-cost asset in any AI visibility programme, and this post shows exactly how to build and deploy one.

What is an Entity Brief?

It is the firm's official description of itself, written once and never paraphrased.

Not a tagline. Not a mission statement. Not marketing copy. A paragraph of verifiable facts in a fixed order: canonical name, founding year and credentials, client segments, services, location and service area, one factual differentiator, contact details. Once written, it is pasted, unchanged, into every listing, profile, directory, bio and schema block the firm controls.

The concept comes from how search systems have handled entities for a decade, but AI engines have made it decisive. In classic search, inconsistent listings hurt local rankings a little. In generative search, they can remove you from the answer entirely.

An Entity Brief is not copy about the firm. It is the firm's identity record, and AI engines treat it as such.

Why does an AI engine need one paragraph?

Because it has to decide whether you exist before it decides whether to recommend you.

When a prospect asks ChatGPT for "a good CA in Andheri for NRI taxation", the engine does not look up a ranking. It fans the question out into sub-queries, retrieves sources for each one, and then tries to resolve the firms it finds into distinct entities. For every candidate it is asking a quiet question: is the "Sharma Tax Consultants" on this website the same business as the "Sharma & Co CA" on JustDial and the "R. Sharma, Chartered Accountant" on LinkedIn?

If the answer is uncertain, the engine's confidence in that entity drops. Low confidence means the engine does not name you. It names the competitor whose five listings all say exactly the same thing.

The data supports how much this matters.

• LinkedIn is the most-cited domain for professional queries across ChatGPT, AI Overviews, AI Mode, Copilot and Perplexity, appearing in 14.3% of ChatGPT Search responses (SEMrush, 325,000-prompt study, March 2026). Your LinkedIn description is one of the first things an engine reads about you.
• Domains with profiles on review and listing platforms have roughly 3x the chance of being chosen by ChatGPT as a source (SE Ranking, November 2025). Each listing only helps if it corroborates the others.
• SparkToro found less than a 1 in 100 chance that ChatGPT returns the same list of brands twice for the same prompt (January 2026). In a system that unstable, the firms that appear reliably are the ones the engine is most certain about.

There is also a hallucination angle. When an engine has partial or conflicting information about a firm, it fills the gaps. I have watched ChatGPT describe a Mumbai CA firm as offering services it has never offered, at an address it left in 2019. A consistent Entity Brief gives the engine a complete record to work from, which is the best defence against being described wrongly.

What goes into an Entity Brief?

Seven fields, in a fixed order. Every one of them is a fact a prospect might ask an engine about.

The full working document holds a little more than the paragraph itself, because different surfaces have different length limits.

• A 10-word version for social bios and directory headlines
• A 30-word version for listing summaries
• The 100-word canonical paragraph, which is the Entity Brief proper
• A structured services list with one-line descriptions
• Partner names, credentials and short bios
• A change log with version numbers and dates

Everything downstream is derived from the canonical paragraph. Nobody writes a fresh description for a new listing. They pull the version that fits and paste it.

What does a before-and-after look like?

Here is a composite drawn from audits of Mumbai CA firms, with the firm fictionalised.

Before: five surfaces, five descriptions

• Website: "Sharma Tax Consultants: your trusted partner for all financial needs"
• JustDial: "Sharma & Co CA, Andheri East. Tax, audit, accounting, company registration"
• LinkedIn: "R. Sharma, Chartered Accountant with 10+ years experience"
• ICAI directory: "Sharma Rakesh & Associates, Mumbai"
• Google Business Profile: "Sharma CA" at the pre-2019 Vile Parle address

Asked to describe the firm, ChatGPT produced a paragraph that got the name wrong, listed "company registration" as a primary service, and gave the old address.

After: the Entity Brief, deployed identically on 18 surfaces

The paragraph in Fig. 2. Same name, same services, same address, same partner credential, on every listing.

Asked the same question sixty days later, ChatGPT described the firm accurately, named the NRI and founder-company focus, and, for two of the five test queries, included the firm in its recommended list for the first time.

AI engines do not recommend the best firm. They recommend the firm they can describe with the most confidence, and confidence is built from repetition of the same facts.

Where is the Entity Brief deployed?

Everywhere an engine might read about the firm. For a Mumbai professional services firm that is typically 15 to 20 surfaces.

For clinics the professional group changes to Practo and Lybrate. For law practices it changes to Bar Council listings, with the compliance note below.

How do you write an Entity Brief?

Five steps. The first three take an afternoon. The fourth takes two weeks. The fifth never stops.

1. Draft from a partner interview, using verified facts only. Founding year, ICAI or Bar Council or medical registration numbers, actual client segments, actual services, exact address. If a fact cannot be verified against a document, it does not go in. Aim for 80 to 120 words.
2. Test it on the engines before deploying. Paste the draft into ChatGPT and Perplexity with the prompt "Summarise this firm in two sentences." Whatever the engine drops or garbles is what is unclear. Rewrite until the summary comes back accurate every time.
3. Lock it. One source document, a version number, a date, partner sign-off. Derive the 10-word and 30-word versions from the locked paragraph, never the other way round.
4. Deploy to every surface, word for word. Log each surface, URL and deployment date in a tracking sheet. Where a platform imposes a character limit, use the 30-word version rather than trimming the paragraph by hand.
5. Monitor monthly. Ask each engine to describe the firm. Compare the answer to the brief. Drift, hallucinated services and old addresses get corrected at the source and re-deployed. Any real change to services, partners or address restarts the cycle at step one.

What are the most common Entity Brief mistakes?

1. Writing marketing copy instead of a record. "Your trusted partner for all financial needs" contains zero facts an engine can match to a query. Replace every adjective with a noun a prospect would search for.
2. Letting each listing be "a bit different". Staff update JustDial one way and LinkedIn another. The engine sees two entities. Consistency is the entire point; a slightly better paragraph on one surface is worse than the same paragraph on all of them.
3. Ignoring the name string. "Sharma & Associates" versus "Sharma and Associates" versus "Sharma & Associates LLP" are three names to a machine. Pick one legal form and use it everywhere, including in review replies.
4. Leaving the old address alive. A stale Google Business Profile or a forgotten directory entry keeps feeding the engine wrong data. Audit and correct every listing you have ever created.
5. Skipping the schema. The description property in your LocalBusiness or ProfessionalService JSON-LD is the most machine-readable copy of the brief you will ever publish. Put the canonical paragraph there, exactly.
6. Writing it once and forgetting it. A new partner joins, a service line is added, the office moves. If the brief is not updated and re-deployed, the entity fragments again within a quarter.

Is the Entity Brief compliant for CAs, lawyers and doctors?

Yes, and it is arguably the most compliant marketing asset a regulated professional can have.

ICAI's April 2026 guidance permits a digital presence and the dissemination of factual information about a practice. An Entity Brief is factual information by construction: name, credentials, services, location, contact. It contains no solicitation and no comparative claims, and the discipline of "no unverifiable adjectives" removes the exact language that creates compliance risk. For advocates, Bar Council Rule 36 restricts advertising; a factual entity description on a directory the Bar Council itself maintains is not advertising, but each surface should be reviewed with counsel. For clinics, the same factual discipline protects against the promotional health claims regulators watch for.

A firm described accurately and consistently everywhere is both more visible to AI and more defensible to a regulator. Those are the same property.

Frequently asked questions

What is an Entity Brief in generative engine optimization?

An Entity Brief is a canonical paragraph of 80 to 120 verified words stating a firm's exact name, credentials, client segments, services, location and contact details, deployed identically across every directory, profile and schema block the firm controls so that AI engines resolve it as one consistent entity.

How many places should a professional firm publish its Entity Brief?

Typically 15 to 20 surfaces: the firm's website and schema, Google Business Profile, Bing Places, the relevant professional directory (ICAI, Bar Council, Practo), LinkedIn, four to six Indian business directories, review platform profiles, Quora and Reddit bios, and press boilerplate.

How long does it take for AI engines to reflect a new Entity Brief?

Engines that browse live, such as Perplexity and ChatGPT with search, can reflect corrected listings within days of re-crawl; in Lumicite engagements, consistent and accurate descriptions of the firm across ChatGPT, Perplexity and Gemini typically appear within 30 to 60 days of full deployment.

Find out how AI describes your firm today

The Lumicite AI visibility audit tests 25 real client queries across five AI platforms, records how each engine currently describes your firm, including any wrong services or old addresses, and maps every surface where your entity is fragmented. You receive a 15-page report in 5 to 7 working days. Writing and deploying the Entity Brief is Month 2 of the 90-day sprint the audit credits toward.

Request the AI visibility audit →

₹9,500 · 5 to 7 working days · creditable

A note on the example: the before-and-after is a composite of audit findings with a fictional firm name, and no real firm is described. The 30 to 60 day timeline reflects Lumicite engagement observations rather than a controlled study. Statistics re-verified 17 September 2026; next review December 2026.

Sources

1. SEMrush, LinkedIn citation share across engines, from an analysis of 325,000 prompts (2026)
2. SE Ranking, roughly 3x source-selection likelihood for domains with listing and review profiles (2025)
3. SparkToro, under a 1 in 100 chance that two responses to the same prompt list the same brands (2026)
4. 5WPR AI Platform Citation Index, Reddit's share of AI citations across ChatGPT, Gemini and Claude (2026)
5. ICAI, April 2026 guidance on digital presence and factual information about a practice (2026)
6. Lumicite engagement observations, the before-and-after example is a composite of audit findings with a fictional firm name; the 30 to 60 day timeline reflects engagement observations, not a controlled study (2026)