YC Root AccessBuilding AI for the $650B Home Services Market
CHAPTERS
- 0:05 – 1:57
Broccoli today: AI front office for home services, 500+ customers, rapid growth
Gustaf asks for a snapshot of what Broccoli is today. AJ explains that Broccoli automates the front office for trade businesses by answering calls and handling follow-ups and scheduling across voice and text channels, and shares growth and usage metrics.
- •Broccoli’s AI answers customer calls, follows up on leads, and schedules appointments across voice, text, and SMS
- •The company serves over 500 customers
- •AJ says their AI-booked jobs are at a run rate of over a billion dollars in booked revenue for customers
- •AJ says they will end the year at over $60M ARR after being at $0 revenue two years prior
- •Gustaf frames the company as having been under the radar despite strong traction
- 1:57 – 2:46
Early founder journey: college collaboration and the first (failed) campus delivery idea
Aditya explains how the founders met in college and began building projects early. They describe a simple local delivery app on campus that quickly ran into restrictions, but it set the tone for repeatedly building ideas together.
- •The founders met in college and started working on ideas from sophomore year
- •Their early campus delivery app was shut down/banned and they couldn’t take payments
- •They continued collaborating on each other’s projects over multiple years
- •Their experimentation mindset began well before YC
- •They studied at BITS Pilani in India
- 2:46 – 3:49
COVID-era edtech attempt: hard to scale and a reset
Aditya recounts building an edtech idea during COVID while he was stuck in Spain for a research internship. Gustaf recalls hinting during YC follow-up that the idea might not be right; the founders later found it difficult to scale and shut it down before taking a short break.
- •The founders launched an edtech idea during COVID and worked on it for years
- •Gustaf says they were strong founders but questioned whether the idea was the right one
- •Aditya says they couldn’t scale the edtech business and decided to shut it down
- •They took about three months off before restarting
- •They adopted a framework to try ideas for three months at a time and move on if needed
- 3:49 – 4:53
Telematica: EV APIs, getting into YC, and realizing India wasn’t the right initial market
The conversation shifts to Telematica, the idea that ultimately got them into YC. Aditya explains the EV-API intersection, while AJ says India was a good start but the market and use cases weren’t big enough, prompting a move out of India soon after YC.
- •Telematica was positioned as APIs for electric vehicles
- •They applied to YC within about 25 days of starting the idea and were accepted
- •AJ says India’s EV market and the targeted use cases weren’t large enough at the time
- •After getting into YC, they moved out—eventually to London
- •Gustaf probes whether India was the right geography; AJ answers no
- 4:53 – 7:11
Going where EV adoption is: Norway discovery and the demand-response charging use case
AJ explains how inbound interest from European energy companies pulled them toward Europe. They traveled to Norway, met many companies, and learned about grid balancing and demand-response needs that their APIs could support by controlling EV charging based on renewable supply.
- •European energy companies reached out asking to connect to vehicles like Tesla and BMW
- •After Demo Day and raising a seed round, they traveled to Norway for two months
- •They met with 20+ companies and explored more mature EV/energy use cases
- •Primary use case: demand response—managing consumption based on wind/solar supply
- •Their APIs could pause charging across many vehicles in a district when supply is constrained
- 7:11 – 8:08
Why Telematica didn’t work: weak consumer economics and low adoption
Gustaf asks why the Norway-focused concept didn’t take off. AJ says the consumer value was too small relative to the risk and inconvenience of remote control, so adoption didn’t materialize and utilities couldn’t get traction, which meant low usage for their APIs.
- •AJ says the idea sounded great but the economics didn’t compel consumers
- •He describes savings around $50/year as insufficient given the risk of a car not being charged
- •Low consumer adoption prevented utility programs from gaining traction
- •Because customers couldn’t drive adoption, Telematica didn’t see meaningful API usage
- •This became a key lesson about choosing high-priority customer problems
- 8:08 – 10:28
The pivot mindset: investor pressure, founder doubt, and a priority-based lesson
AJ describes the emotional difficulty of realizing a multi-year effort wasn’t the right path, including an investor suggesting they return capital and get jobs. He explains how they reframed investor capital as funding to learn and emphasized solving customers’ top pain, not a lower-ranked priority.
- •AJ describes the temptation to chase one big customer to ‘make it work’
- •He says raising capital can create pressure to keep pursuing the original idea
- •AJ reframes investment as backing founders to learn and decide next steps
- •An angel investor suggested returning capital and getting ‘real jobs,’ prompting self-doubt
- •Key takeaway: validate whether a problem is a top priority for the customer
- 10:28 – 11:39
Choosing contractors and a massive market: criteria for the next bet
After evaluating their existing customer segments, AJ explains why they chose contractors over utilities and fleet operators. He outlines two criteria drawn from Telematica: start with a big market now and focus on the customer’s single most painful problem.
- •They focused on existing customer types: utilities, contractors, and fleet operators
- •They chose contractors partly because AJ preferred selling to owners vs large-company product managers
- •Two selection criteria: target a massive market from day one; solve the most painful problem
- •They looked at trades/home services as a large opportunity and referenced ServiceTitan as a major player
- •They began customer discovery to find the top pain point for contractors
- 11:39 – 14:51
Door-to-door discovery: failing fast, then unlocking conversations by changing approach
AJ recounts early failures trying to talk to contractors and being treated like salespeople—especially when leading with ‘AI.’ They switched to introducing themselves as students learning about family-owned businesses, which opened doors and led to a large volume of in-person conversations.
- •Gustaf challenged them on doing discovery via LinkedIn and suggested going door to door
- •AJ describes getting rejected and even cursed out when pitching as ‘AI startup founders’
- •He notes AI was a taboo word in the service industry at the time
- •They changed tactics: no laptops, introduce themselves as students there to learn
- •After the change, they averaged about eight conversations per day and kept it up for two months
- 14:51 – 16:22
Finding the real pain: after-hours calls, expensive answering services, and lost leads
From dozens of interviews and shadowing technicians, they learned a recurring, high-stakes problem: missing calls after office hours loses revenue, but outsourcing answering is both costly and poor quality. AJ explains how this problem was urgent because customers quickly call the next provider if unanswered.
- •They spoke with roughly 140 contractors around the Bay Area and observed day-to-day operations
- •AJ asked owners what they think about after 5 PM to uncover real anxieties
- •Core issue: after-hours coverage—emergencies happen 24/7 but the office closes around 4–5 PM
- •Businesses relied on third-party answering services that were expensive and often bad
- •If calls aren’t answered, leads are lost because customers call the next provider
- 16:22 – 19:00
First customers without ‘selling AI’: demos, low-risk use cases, and embedding with a Sacramento plumber
AJ explains they avoided pitching AI directly and instead introduced a ‘magical’ voice demo after establishing trust and learning context. Their first committed customer was a plumbing company owner in Sacramento, where the founders embedded onsite and wrote the first code from the shop rather than from San Francisco.
- •They intentionally avoided presenting AI as the solution upfront due to industry skepticism
- •They asked whether owners had tried AI tools like ChatGPT to gauge openness
- •They played a recording of an AI voice, which many described as ‘magical’
- •Initial low-risk idea from owners: AI calling to confirm appointments
- •Their first customer was a Sacramento plumbing company owner; they embedded there for months and wrote the first code onsite
- 19:00 – 20:39
First Broccoli launch: ServiceTitan integration, limited rollout, and outperforming the old answering service
AJ describes building and launching quickly by learning the customer’s scripts and integrating with ServiceTitan. They started with a narrow time window, monitored every call, and quickly booked more appointments than the prior human/outsourced setup, convincing the customer to replace the answering service.
- •They built an initial version in about two weeks, including integrating with ServiceTitan
- •They learned the scripts and processes used by the customer’s call center staff
- •Initial rollout answered calls only from 5–7 PM while they monitored every call
- •They booked three appointments in two hours; AJ says the prior maximum overnight had been two
- •They later ran a full weekend and booked about 19 appointments vs a previous max of five, prompting the customer to go fully live
- 20:39 – 21:42
Referrals and what product-market fit felt like: being ‘pulled’ into more use cases
Gustaf asks about the moment product-market fit became obvious. AJ says that after meeting their first customer, they never paused to find growth because demand and use cases kept pulling them forward, and their first customer’s referrals helped them expand quickly.
- •Their first customer drove the next wave of growth through referrals and word of mouth
- •AJ says they didn’t need to keep going door to door after early traction
- •AJ describes PMF as being continuously pulled into conversations and bigger use cases
- •Gustaf frames this as the first real ‘taste’ of PMF after years of searching
- •AJ adds that PMF must be re-found at each stage as the business evolves
- 21:42 – 24:08
Scaling operations: early team, round-the-clock support, and the market size and spend dynamics
They discuss the small early team that supported customers intensely, including a founding engineer working from India to maintain a near 24-hour response cycle. The conversation also covers how Broccoli fits into contractors’ software and labor spend, and AJ cites the size of the trades and home services markets.
- •Aditya says a founding engineer joined during Telematica and stayed through pivots
- •They coordinated across time zones so issues could be fixed overnight while calls ran in Pacific Time
- •At around $2M ARR, they had roughly 15 people
- •AJ explains contractors spend on systems like ServiceTitan and QuickBooks and significant office labor, while Broccoli automates part of the front office
- •AJ cites market sizing: trades broadly and home services specifically, describing it as a very large market
- 24:08 – 26:20
From fast startup to scaling company: leadership hiring and engineering culture at 75 people
Gustaf asks what’s top of mind as they head toward the stated revenue scale. AJ focuses on building the leadership team, while Aditya explains the engineering challenge of maintaining velocity and culture as headcount and customer responsibilities grow.
- •AJ says his top focus is building the company’s leadership team as they scale
- •AJ says the company is around 75 people at the time of recording
- •Aditya contrasts small-team ‘willpower’ speed with the need for structure as you grow
- •He emphasizes maintaining engineering/product velocity while retaining culture
- •They aim to build principles, mentorship, and customer-centric thinking as the team expands
- 26:20 – 29:10
How AI may change the trades: reducing experience gaps and making technicians more effective
Gustaf asks how the technician’s job might change beyond front-office automation. AJ predicts AI can compress the time needed to become highly competent by providing guidance during diagnosis, and Aditya adds that much of the work is operational—finding the right documentation and steps—where AI can improve efficiency.
- •AJ describes Broccoli as ‘automating the work around the work’ today
- •AJ predicts AI will help technicians diagnose issues faster and reduce the years needed to become highly skilled
- •He says AI could provide expert-level guidance ‘in your ear’ or ‘in your eyes’ while on the job
- •Aditya says many repairs involve identifying the appliance and looking up correct documentation and procedures
- •He describes current inefficiencies where technicians rely on back office support, and argues AI can put that knowledge in technicians’ hands while they still perform the physical work
- 29:10 – 31:40
Founder reflections: move toward customers, intrinsic motivation, and building with friends
Gustaf asks what they would tell their younger selves and what motivates them through hard periods. AJ emphasizes moving to where customers are and credits being in San Francisco as key, while Aditya highlights the satisfaction of solving real-world problems and the motivation of building alongside friends.
- •AJ says the serious framing is to move toward your customers—go where they are to get answers
- •He gives examples from their experience of traveling to customer-heavy regions
- •Aditya says motivation comes from seeing real-world impact on technicians quickly
- •Aditya also cites building with friends and a long-term belief in improving outcomes as core drivers
- •AJ says he doesn’t see himself doing anything else and rejects returning to traditional jobs as a path for him
