Summer 2024

autarc

autarc is the agentic OS for energy installers.

Visit website

What they do

agentic OS for small energy installers via WhatsApp

How close to us

5, very close

one layer sold to many small firms of one type; agents run installer admin

Wedges

  • W5 One layer for one type of small firm direct
  • W4 Fit call to tailored proposal in 10 minutes adjacent
L1 Company brain (knows the organisation) 1, light
L2 Agent team (runs agents, tools, approvals) 2, clear
L3 Does and learns (finishes, checks, measures, updates the brain) 1, light

Who buys

Firm size
sme
Buyer type
business
Role
owner or operations lead of a heat pump or solar installer
UK
yes
Their own description (from YC)

The problem: Energy installers still waste dozens of hours per project on pen & paper processes. Their CRM, ERP, email, planning tools are completely detached from each other.

The solution: autarc lets installers automate their repetitive business tasks on a single platform. They manage everything from one interface: WhatsApp or iMessage. This combines their legacy tools into one system. We save 24+ hours on every home energy project, like heat pump, solar, wallbox or battery installations.

The technology: On top of our OS, autarc leverages LiDAR for spatial analysis, computer vision for automated assessment, and AI for system recommendations. This enables installers to confidently transition to and scale up sustainable home energy solutions.

Full teardown

autarc (YC Summer 2024): teardown vs XMarts

In one line: autarc sells one software platform to small heat pump and solar installers. It runs their office work, designs the heating system from a LiDAR scan, and adds AI helpers on the phone and on WhatsApp [site] [yc]. Threat: 1, because it is our W5 idea (one layer for one type of small firm) already built and selling, but for energy installers, not for the firms we would pick [site] [yc]. A YC partner can use it as proof that the W5 shape works, so expect it to come up.

1. Fact file

Item Finding Source
What they say "The agentic OS for energy installers." Older YC launch title: "The OS for heat pump installers." [site] [yc]
Founded, team, funding Founded 2023, Berlin. Team size is given three ways: 30 (YC page), 40+ (home page), 60+ (a job ad). The job ad is the newest. Pre-seed Dec 2023, "single-digit million" (euro, currency not stated), led by PropTech1 Ventures. A seed of about $8.9M, Feb 2025, is reported by data sites with Visionaries Club, YC and better ventures. No official announcement found, and one site dates it Dec 2024. Total raised: unknown. [site] [yc] [jobs] [press]
Founders Etienne-Noel Krause (CEO, ex-HVAC industry, part-time heating engineering studies). Thies Hansen (COO, ex-logistics process work, then startups and VC). Marius Seufzer (CTO, ex-Lead iOS Engineer at Sameday Health, bio says he built Passkeys for Apple). What YC likely saw: a founder from the trade plus a strong product engineer, and about 80 installer interviews before building (inference from the launch post). [yc] [press]
Customers They claim 2,000+ businesses in Europe. Named logos: Coppen, Daume, Hörmann, Enerix, Hückstädt, Aircona, Theodor Bergmann, iHeat, D5-Energie. At YC launch they claimed 380 customers and 7-digit ARR in Germany. [site] [yc]
Price Public. UK: Core AI £39 a month, Pro £99 (heating) or £69 (solar), Premium £199 or £139, All-in-One £349. Annual billing, 12-month minimum, £290 setup fee (waived if you start within 24 hours of the demo). Shared "AI credits" pool per plan. No free trial. [site]
Path Heat pump planning tool (pre-seed, Dec 2023) → "heat pump operating system" (seed coverage, 2025) → "agentic OS for energy installers", with solar, an AI phone agent and Scotty on WhatsApp. Date of the "agentic" rename: UNVERIFIED. [press] [site] [yc]

2. Their proposition, reverse engineered

Part What they do
Who buys Owners and office leads at small heating, solar and electrical installers. Also utilities and energy consultants listed as target groups [site].
The pain Dozens of hours per project on paper and detached tools: CRM, ERP, email and planning do not talk to each other [yc]. Heat pumps need about 30 times more data than a gas boiler and take three to four times longer to quote, they claim [press]. Installers spend almost half their time on admin, they claim [press].
The promise Save hours on every project. They claim "24+" hours per project (YC page), "up to 12" hours in planning (site), and up to 20 hours per installation (launch post). The numbers differ by page [yc] [site].
How it works One platform for quotes, customers, financing, LiDAR heat load and hydraulic balancing, PV design, a field app and subsidy forms [site]. On top sit AI helpers: Scotty (project Q&A, documents, WhatsApp), an AI phone agent, a sales AI, and "AI workflows" and "AI planning check" marked coming soon [site].
The output A finished design, a quote, subsidy and grid forms, call notes and tasks created in the CRM [site].
Proof style Hours saved per project, number of businesses, Trustpilot rating (4.5 to 4.7 stars, the pages differ). Named customer stories, no quotes in the voice agent page [site].
Who sells it Own sales team in Berlin (job titles seen: sales development rep, account executive, working students for Italy and the UK). Partners: financing firm Cloover, Oct 2025 [jobs] [press]. No self-serve trial (inference from "no free trial" and the setup fee) [site].
Price model Monthly seat-and-project subscription plus AI credits plus one-off setup fee. Voice agent also quoted per call minute (from $0.39, UNVERIFIED which plan carries it) [site].
The story Their enemy is paper and disconnected tools. Their belief: installers need one system so they can scale heat pumps. Public goal: plan 1 million heat pumps by 2030 [jobs] [press].

3. Positioning

  • Category they claim: "The agentic OS for energy installers" [site] [yc]. The UK page says "The AI software for energy installers" [site].
  • Category a YC partner would put them in: vertical software for the trades, with AI helpers on top (inference: the planning engine, quotes and CRM are the core, and the agents sit on them).
  • What they want to be known for: the one system a small installer runs the whole business in.
  • Where they sit on the map: layers L1/L2/L3 = 1 / 2 / 1, buyer = sme, business.

4. Three layers vs ours

Layer Them (score + evidence) Us (Marton's logs show)
L1 Company brain 1. One project record per property (they claim 800+ data points) that Scotty and the phone agent read [yc] [site]. It is a project database for one trade, not a memory of the company that learns from its work (inference). A vault of about 1,400 pages, 112 call transcripts and about 8,700 customer quotes, searchable by meaning.
L2 Agent orchestration 2. Several agents (Scotty, phone agent, sales AI) that create tasks, route calls and suggest engineers by availability [site]. No public claim of approvals, and "AI workflows" is marked coming soon [site]. About 5,600 jobs run by agents from three makers, each checked by a different AI, then automatic tests.
L3 Execution and learning 1. The software does real work (heat load, design, quotes, forms) [site]. The check, "AI planning check", is coming soon, and no learning loop is claimed [site]. About 160 failure records, each adding an automatic check. Offer feedback lines fell from 44 to 18 as rules were added. Only about 15% of jobs reached the main code, and merges were unreliable in October.

5. Same and different

Topic Same or different Them Us
Core belief Same Small firms in one trade need one layer that joins their tools and removes admin [yc]. Same (W5 in the brief).
Buyer size Same Small firms, tens of staff [site]. Target: SMEs and scale-ups (brief).
Buyer type Different One trade, energy installers, with hard physical work [site]. Not picked. W5 names accountancy firms as an example (brief).
Kind of work Different Physical-world design (LiDAR, heat load, site photos) plus admin [site]. Knowledge and back-office work. Marton's logs show code, content, offers and call filing.
Channel Different Their agent is reached through WhatsApp and the phone [site]. Marton's logs show agents take work from a job queue, and decisions go to one person on a review board. Built for one person, on Macs.
Proof Different Hours saved, businesses count, named stories [site]. 2,000+ businesses (they claim). No outside customer pays yet.
Price Different Public monthly tiers from £39, setup fee £290 [site]. Not set yet.
Delivery Different Self-serve software plus a sales team plus partners [site] [jobs]. The system plus the two founders.
Learning Different No claim that the agents learn from checked results [site] (UNVERIFIED that none exists). Marton's logs show rules added from failures and feedback.

6. Wedge overlap

Wedge Fit Why
W5 One layer for one type of small firm direct Same shape: one platform for one kind of small firm, agents running the repeat admin. A YC partner would call it the same idea in another vertical [yc] [site]. Different type of firm, so not a rival for an accountancy layer.
W4 Fit call to tailored proposal in about 10 minutes adjacent Their quote tool builds an offer from a scan and project data, and the sales AI compares rival quotes [site]. Same job (offer from a first meeting), different buyer (installers, not consultancies).

7. Whitespace (where they stop)

  • Layer gap: their agents answer, route and draft. They do not claim to check results and feed them back into rules [site]. To own it we need a rule-change log tied to checked runs, with rework falling on one customer workflow.
  • Buyer gap: one trade, installers [site]. Any other type of small firm (accountants, agencies, consultancies) is open. To own it we need one named firm of that type with the pain on file.
  • Number gap: they claim hours saved, and the figure moves between 12, 20 and 24+ on different pages [site] [yc]. They show no money result. To own it we need one before/after number a finance person would sign.
  • Price gap: none. Entry is £39 a month with a £290 setup fee, which a small firm can pay [site]. Do not claim this one.
  • Speed gap: none found. They claim setup in under 30 minutes and the phone agent "ready in a few minutes" [site]. Do not claim this one.

8. For the YC application

  • Proven today: Marton's logs show his system runs agents that do and check work: about 5,600 jobs, each checked by a different AI, with about 160 failure records that each added an automatic check. autarc's public pages describe agents that answer, route and draft, and a planning check that is coming soon [site].
  • Draft answer to "how are you different from autarc?": DRAFT. "autarc is the same idea for heat pump installers: one layer for one trade. We pick a different kind of small firm, and our agents check their own work and learn from it." Founders say it their own way.
  • Future claim (needs proof): "One layer, built once, sold to many firms of one type, with a number that moves." Needs one paying firm of the chosen type, one workflow run for 4+ weeks and a before/after number. A second firm of the same type would show it can be reused.
  • What we understand that they don't: the layer should learn from checked results, not only from the project data it holds. needs proof on a customer; Marton's logs show it on his own work only (offer feedback lines fell from 44 to 18).
  • Copyability (M3): medium. They could add checks and learning to their agents in months, since "AI planning check" and "AI workflows" are already on the roadmap [site]. Entering a new trade would pull them off their physical-design core and their installer sales team (inference from their product and job ads) [site] [jobs]. So the edge "learns from checked results" is medium; the edge "a different vertical" is hard for them, but a different vertical is not a defensible edge. Not hard, so M3 can not score above 3 against autarc.
  • Why now (M4): None found. Their new agents are general AI progress and do not count.
  • Do not claim: "nobody has built a vertical OS for small firms" (autarc has, and sells it). "Our agents learn from checked results for a customer" (no customer run yet). Any claim that their price or speed is a gap.

For the pitch, the strongest use of autarc is as proof for the form's market answer: a YC-funded vertical layer with 2,000+ small-firm customers (they claim) shows the W5 shape sells to small firms at £39 to £349 a month [site].

9. Ideas worth taking

  1. Reach the customer's tool they already use. Scotty runs on WhatsApp and the phone agent works by forwarding the existing number, so there is nothing new to learn [site].
  2. Make the first week free of friction: setup fee waived if you start within 24 hours of the demo. It pushes a buyer to decide while the demo is fresh [site].
  3. Show proof as hours saved per project with named customer stories. Same shape for us once we have a real number, but with money, not only hours.

10. Unknowns

  • Real ARR and total funding. The 7-digit ARR and 380 customers are from the YC launch post (undated here). The seed size is from data sites only. Check a German outlet or ask a customer.
  • Which team size is right (30, 40+ or 60+). The job ad is newest.
  • Whether the UK is a live market with UK customers or only a UK site and pricing. Capterra UK lists Germany only, while a rival's comparison page says autarc is active in the UK [press]. Look at their UK customer stories.
  • Whether the AI features (WhatsApp agent, workflows, planning check) are all live. Some are marked "coming soon".
  • Whether the agents ask for human approval before acting. Not stated on the pages read.

Sources

Technical detail
Slug
autarc
Lens hash
179096c479
Pass
deep
Stale
no