What you could build
Five workflows for Telescope's data stack.
Ordered by how fast Mike could run them with a key. Play 01 is the taxonomy problem. Companies and titles shown here; names come back in the pull.
01
Draw your own market lines, as narrow as you want
The taxonomy problemRun it this week
Preset categories break at the niche end. After the call we described four clinic-software markets in plain English and got back 41 on-target vendors. The ABA set alone is filed under 10 different industry labels, one of them "Advertising Services". No category list would have drawn that line. A sentence did.
How it runs
- Describe the market the way a partner would ("EMR and billing for outpatient PT clinics") and rank companies by what they actually do with
company_search in semantic mode.
- Pull each company's description and listed specialities with
company_enrich, plus their site text with web_fetch when the description is thin, and run your own classifier over it.
- Write your labels to the warehouse and rerun monthly. New entrants show up on their own: the vet query surfaced vendors founded this year.
You set the granularity. The category list never caps how narrow a market can go.
→ your warehousemonthly refreshnew thesis or market
Four narrow markets · plain-English queriesOct 8, 2026
25 ranked results per query. We dropped billing services, review sites and general clinic tools by hand. That cleanup is the step your classifier takes over.
02
Spot the first sales hire, before the round
Only Telescope runs this oneSignal · job postings
Telescope's entry point is the moment a sales process starts to repeat. That shows up in a job post months before it shows up in a press release. Since Jul 10, 192 US software companies with 10 to 100 people posted 367 roles for a founding AE or a head of sales, or described the role as their first sales hire. Plenty of them are capital-efficient: Haystack, Claimlane, Jugl and Breakout have each raised under $4M on record.
How it runs
- Search full posting text by exact phrase ("founding account executive", "first sales hire") with
job_search, scoped to software companies of 10 to 100 people.
- Cut it with company fields you already think in: funding on record, headcount growth, and the vertical labels from play 01.
- Set a watch so every new matching posting arrives by webhook, ready for a partner to look at the same week.
A funding feed tells you after the round. The posting tells you when the motion starts.
→ deal pipelinedailynew matching posting
A sample of the 192 · lightly funded or none on recordOct 8, 2026
"None" means no round in our data: bootstrapped or unreported. Some of these may already be on your radar.
03
Feed the ARR model with history, not snapshots
For the modelsBulk or API
Revenue isn't public at this stage, so the model leans on what is. Take Hellonote, a PT, OT and speech therapy EMR from the play 01 pull, with no outside round on record. Its profiles grew from 8 to 14 since late 2022, its site traffic rose 46% last quarter, and it posted a growth marketing role on Sep 16. Every one of those is a monthly series you can train on.
How it runs
- Pull history per company by domain or id with
company_enrich: headcount over time and by function, web traffic, followers, funding and an estimated revenue band.
- Add hiring velocity by function from
job_search on the company id.
- Join on domain to your CRM and to the real ARR and customer counts you have for portfolio companies, train on those, and score the rest of the market.
Straight about itRevenue comes as an estimated band, not reported ARR, and customer count isn't a field. Your portfolio's real numbers are the labels. We supply the inputs.
→ feature storemonthlymodel retrain
Hellonote · one row of featuresOct 8, 2026
A company you might never see in a funding feed, with nearly four years of headcount history behind it.
04
Enrich on add, then keep watching
Run it this weekAPI
The simple one. An investor adds a company to the CRM, one call fills the record, and a watch keeps it current. Here's AlohaABA, the ABA vendor from the hero, as it would arrive: 60 profiles, an estimated $2.5M to $5M in revenue, and a demand gen and GTM leader just posted.
How it runs
- Your CRM fires on a new company.
company_enrich by domain returns firmographics, funding, headcount by function, hiring and traffic.
- Write the fields you care about to the record. The rest stays in the warehouse for the models.
- A
watch on the company sends changes by webhook: a new round, a jump in sales headcount, a new sales leader.
Same data as the models, on the same key. The record a partner opens matches the row the model scored.
→ CRM recordon add, then on changeinvestor adds a company
New company in CRM · enrichedOct 8, 2026
CompanyAlohaABA · ABA practice managementalohaaba.com
Profiles60, up from 13 in mid-2023+71% 2-yr
Sales + BD11 of 60by function
Est. revenue$2.5M to $5Mestimate
Web visitors388,825 a month+83% QoQ
HiringDirector of Marketing, Demand Gen & GTMSep 11
FundingNo outside round on recordfunding
→ written to the CRM record · watch set for changes
05
Find the next sales hire for a portfolio company
Value-add for founders
Recruiting for the portfolio runs on the same people data. 465 people in the US used to work at the 41 clinic-software vendors from play 01, and 47 of them are in sales roles today. If a portfolio company sells into clinics, that's a shortlist of people who already know the buyer.
How it runs
- Take the founder's role and the vendor list from play 01.
person_search on past employer, current function and location.- Rank against the role and hand the founder a short list, with the reason each person is on it.
→ founderper open roleportfolio company hiring
In sales now · a sample of the 47Oct 8, 2026