Lookalike companies
A three-route account-research app that ranks 20 real Trayo lookalikes of Vercel best-match-first, with the contacts Trayo found at each and an honest page on how the list was…
- Built in
- 11 min
- Capabilities
- 4
- Live API calls
- 15
The prompt
Paste it into any coding agent with the Trayo MCP connected.
- + any MCP client
Use Trayo to find 20 companies most like Vercel, best match first.
What this app asked Trayo for
Every name, number and email on the page came back from one of these. None of it was typed in.
- Reads your workspace
What you sell, who you sell to, and the buyer personas already saved in Trayo.
- Finds companies
Surfaces accounts that match your ICP, plus lookalikes of the ones already working.
- Finds the right people
Named stakeholders at an account, matched to the buying roles you care about.
- Builds your account list
Imports, updates and organises accounts so the work survives the session.
Take the tour
The same run as the video above, one stop at a time.
- 01
Before a first call, a rep gets told to go find more companies like one they already know. This is our lookalike shortlist for any account we work, shown on Vercel, and it starts with how the list was made.
- 02
Vercel alone returned 40 rows, mostly name matches like Vectara and Valory, so we threw that run out.
- 03
A sentence search named Heroku, Netlify, Render and Warp, but Trayo flagged it degraded: seven rows, not sixty.
- 04
Here's the result: twenty accounts, best match first, each with Trayo's similarity score drawn as a bar.
- 05
Netlify, Render and Heroku have no score because they were seeds. Ranks four down come straight from Trayo.
- 06
Open Replit, at 0.893, and you see how many of the four seeds it matched.
How it was built
The data behind each screen, and what's faked for the demo.
A salesperson who has just been told to “go find more companies like Vercel” opens this and gets twenty accounts ranked best match first. Each row shows Trayo’s own similarity score as a bar, plus headcount, HQ and the most senior contact found there. Clicking a row opens the company description Trayo holds, a badge for how many of the four seeds it matched, and every stakeholder found at it. A second page lays out all 56 contacts as cards you can filter by account. A third shows the API calls behind the list, including the two I threw away.
The manual version of this is running a lookalike search, working out which results are noise, looking up each company, then searching for people company by company, across several tabs. Here the shortlist, the firmographics and the contacts are already in one place, and the accounts are in the workspace. The list also comes with an account of how it was made, so the person using it can see what to trust.
I checked who I was connected as and what the workspace held. It was empty. My first pass was trayo_find_lookalikes seeded with Vercel alone. It returned 40 rows badly contaminated by name similarity (Vectara, Valory, Vespa.ai, Verint, Verifiable, Vegavid), so I discarded that run. Next I ran a sentence search through trayo_find_companies, which named Heroku, Netlify, Render and Warp as the real category peers. Trayo flagged that search as degraded because a data source was unavailable, and it returned 7 rows instead of the 60 I asked for. I re-seeded trayo_find_lookalikes with all four platforms, size matching off and the industry narrowed, and got 80 much cleaner candidates. Ranks 4 to 20 are ordered straight off that run’s similarityScore, from Replit at 0.893 down to GitLab at 0.856. A trayo_find_companies lookup by website gave me firmographics, logos and descriptions for all 21 companies. trayo_import_accounts then created the 20 accounts in the workspace. Two trayo_find_people calls returned 44 and 31 rows, which I deduplicated to 56 contacts covering every account.
Three judgment calls are mine, and the app says so. Netlify, Render and Heroku sit at ranks 1 to 3 with no score. I used them as seeds, so Trayo could not score them against themselves, and leaving out the three closest peers would have been the wrong answer. I also dropped off-category rows that Trayo scored highly (Verint, VAST Data, Netomi, C3 AI, NetSuite, Yellow.ai). That call is mine, not Trayo’s, and so is the “why it is on the list” line on each account. Both are labelled in the app. The contacts skew towards Field CTO at the larger accounts. That is a pre-sales title rather than a buyer, but it is what the title filter returned, so I left it uncorrected instead of quietly cleaning it up.
Everything except the one-line rationales came from the Trayo API against a real workspace. I built this as a demo for Snowflake. Push to Salesforce, Draft intro, Slack and Export are stubs. Each is tagged demo and opens a toast naming exactly what would have been sent, and nothing is sent. I enriched no email addresses, so the contact cards carry profile links only.