ShortlistGTM Engineering

When your buyer asks an AI who to buy from, it answers with three names.

Shortlist is a GTM engineering firm. We build the systems that decide whether AI tools and agents can read your business, whether they cite you when your market asks a question, and whether they can transact with you. If you're not on the shortlist, a competitor is.

GET / accept: text/markdown
200 text/html (ignored)

This page was written for a human. Below is the request log a machine gets from a typical B2B site: the markdown request ignored, the discovery files missing. Every adjective you can see, it cannot use.

How your buyer sees you

“A trusted partner with fifteen years of experience, dedicated to delivering excellence and building lasting relationships with our clients…”

How the agent sees you
GET /            200 text/html
GET / accept: text/markdown
                 200 text/html (ignored)
GET /llms.txt    404
GET /.well-known/mcp/server-card.json
                 404
GET /.well-known/api-catalog
                 404

Same company, two readers. The left side is your homepage copy. The right side is the request log when a machine asks that same site for something it can use: no markdown, no llms.txt, no declared capabilities. It builds the shortlist from whoever answered. Use the toggle above to read this page the way it does.

Why now

The majority reader of your website is no longer a person.

In June, Cloudflare said bots had passed humans on their network: 57.5% of requests. A year ago, people arriving from AI tools converted at about half the rate of everyone else. Today they convert 60% better. They show up already decided.

{
  "bot_share_of_http_requests": 0.575,
  "source": "cloudflare",
  "measured": "2026-06-03",
  "caveat": "all_bots_not_only_ai_agents",

  "ai_referred_conversion_vs_baseline": 1.60,
  "ai_referred_traffic_growth_yoy_tech": 0.63,
  "source": "adobe_digital_insights_2026",

  "claim_absolute_channel_size": "small",
  "claim_urgency": "standards_forming_now"
}
The questions being asked right now

Your market is already asking. The answer is a short list of names.

Whatever you sell, your market's research now starts as a question to an AI. Not ten blue links. One answer, a few names, and a recommendation. These are the shapes that question takes:

{
  "query_class": "recommendation",
  "answer_format": "ranked_names",
  "names_returned_typ": 3,
  "decided_before_first_contact": true
}
“What is the best product to solve [my problem]?”
“Who are the best [service] providers for a company like ours?”
“Find me the best [product] under [budget] and order it.”
“Which vendors should we shortlist for [what you sell]?”
“What are the top alternatives to [the market leader]?”
Every one of them ends the same way: a shortlist of names. Our whole job is making sure yours is on it.
Services

Three layers, sold as one channel.

SEO gets you ranked. GEO gets you cited. Agent readiness gets you transacted with. Most firms sell one of the three and call it strategy; the buyer's journey now runs through all three in a single session, so we build them as one system.

{
  "layers": [
    { "seo": "ranked" },
    { "geo": "cited" },
    { "agent_readiness": "transacted_with" }
  ],
  "sold_as": "one_channel"
}
Benchmark · no charge

The AI Shortlist Audit

Cloudflare Agent Readiness scores for you and two competitors you name. Citation checks across the real questions your buyers ask, in fresh sessions, screenshotted. A prioritised fix list. Two pages, five business days, yours to keep whether or not we ever speak again.

Engagement · retainer

AI Channel Engineering

Six-month initial term. We make the business machine-legible with structured data, markdown negotiation, and agent-completable conversion paths, rewrite the pages that answer your buyers' questions, then run the channel: monthly re-scans, citation share tracked against named competitors, agent traffic from your logs, standards updates as they ship.

How engagements run

Benchmark, build, manage.

1

Benchmark

Establish where you actually stand: how machines read your site today, and which companies the major AI systems name when your buyers ask. Two numbers nobody in your category is currently tracking.

2

Build

Sixty days. Discoverability and bot-access declarations first because they are config, not construction. Then markdown negotiation, structured data, and the content rewrite that actually moves citations, because the scanner fixes alone never do.

3

Manage

The part that makes it a channel rather than a project. Monthly re-scans, citation share against competitors by name, agent traffic pulled from your logs, and adaptation as the standards move, which right now is monthly.

Principal

Engineer-minded, seller-led.

Shortlist is led by Tom Kiehn, who spent his career in enterprise B2B sales running complex, multi-stakeholder deals. GTM engineering is usually built by technicians who have never sold. Shortlist inverts that: systems built with a closer's understanding of how deals are actually won, delivered by the principal, with a deliberately limited client roster and no hand-offs.

The firm exists because of one observation from inside enterprise sales: buyers changed how they build their vendor list, and almost nobody selling to them has noticed yet.

{
  "principal": "Tom Kiehn",
  "background": "enterprise_b2b_sales",
  "delivery": "principal_led",
  "client_roster": "limited",
  "handoffs": 0
}
Before you engage

The questions we get asked first.

Is this SEO?

No. Different reader. SEO decides your position in a list of links a human scrolls. This decides whether an AI system can read your business at all, whether it names you when a buyer asks, and whether an agent can complete a transaction with you. Your SEO agency may be excellent; almost none of them measure any of the three.

How do you measure results?

Citation share against named competitors, agent-readiness score deltas, agent traffic in your server logs, and branded-search lift, reported monthly. We will say this plainly rather than bury it: these are proxies. AI-channel attribution is genuinely imperfect industry-wide, and we would rather show you honest proxies than invented ROI.

Isn't it too early?

For the revenue, arguably. For the build, no. The work is config-level and content-level, it takes about sixty days, and the scores are public. Your competitors can run the same scan on you tomorrow. The cost of being early here is a quarter of effort. The cost of being late is that someone else is the default answer by the time you start.

Who is this for?

Any business whose market starts its research inside an AI. The mechanics are the same whether that market is buying software, ordering products, choosing a local provider, or deciding where to work: can the machine read you, does it name you, can it transact with you. Send us the question your market asks and we will show you what the AI currently answers.

Start with the benchmark

Find out what the shortlist says about you.

Send your website and two competitors. You get Agent Readiness scores for all three, screenshots of what ChatGPT and Perplexity say when your buyers ask, and the three fixes that matter most, whether or not we work together.

Request the audit

Or write to tjkiehn12@gmail.com with your site and two competitors. We take a limited number of audits each month.

Sources & method · Bot traffic share: Cloudflare, 3 June 2026 (57.5% of HTTP requests, all bot categories). Conversion and traffic-growth figures: Adobe Digital Insights, April and 19 August 2026. Readiness scores in our audits come from Cloudflare's public Agent Readiness scanner (isitagentready.com); it measures what a site publicly declares through emerging agent standards, not how AI tools behave toward it, which is why every audit also includes live citation checks. AI answers vary by session and change over time; every figure we report is reproducible as of its capture date.