Level AI has raised about $73 million (roughly ₹702 crore) to sell contact-centre software to names like Affirm, Penske and Carta, yet the most important part of the company does not sit in Silicon Valley at all. Its headquarters is a Mountain View address in California; its engine room, the engineering and machine-learning teams that actually build the product, runs largely out of India, from Delhi NCR and Bengaluru.
That split is the story. Level AI is an American software company by cap table and go-to-market, and an India-built one by code. It was started in 2019 by Ashish Nagar, an IIT Delhi engineer who had already sold one startup to Amazon and then worked on Alexa before deciding that the least glamorous corner of enterprise software, the customer-support call, was where language AI would pay off first.
Quick facts
| Company | Level AI, Inc. (US-headquartered; core engineering in India) |
| Founded | 2019; launched publicly out of stealth in August 2021 |
| Founder | Ashish Nagar (Chief Executive Officer) |
| Businesses | AI-native contact-centre / conversation-intelligence SaaS: automated quality assurance, real-time agent assist, Voice of the Customer analytics, coaching |
| Total raised | About $73.1 million across seed, Series A, B and C (company-stated at the July 2024 round); Tracxn lists ~$74.4 million |
| Profit / loss | Not disclosed; the company is private and files no audited public accounts |
| Listed | Private (no IPO) |
| Last valuation | Not publicly disclosed in absolute terms; the March 2022 Series B was described as roughly a four-fold jump on the Series A valuation |
| Key backers | Battery Ventures, Adams Street Partners, Eniac Ventures, Village Global, Cross Creek, Brightloop |
What Level AI does
Level AI sells software to the people who run customer-support operations: the contact centres behind e-commerce refunds, fintech disputes and travel bookings. Its platform listens to the conversations those centres have with customers across voice, email, chat and text, and turns that raw talk into scores, alerts and coaching. The company describes it as an AI-native platform built on contact-centre-specific large language models rather than a generic chatbot.
- Auto-QA: automated quality assurance that scores support interactions instead of a manager sampling a handful by hand. Level AI says one customer pushed automated quality review coverage to more than 80% (company-stated case study).
- Agent Assist / CX Copilot: real-time hints and answers surfaced to an agent mid-call, plus a copilot for managers.
- Voice of the Customer: analytics that mine the whole conversation stream for why customers are calling and what is going wrong.
- Coaching and screen recording: tools that flag issues such as “dead air” and feed manager-led performance improvement; agent screen monitoring and recording were added around the 2022 Series B.
The origin: from Amazon to the contact centre
Ashish Nagar is not a first-time founder. He studied Engineering Physics at IIT Delhi, then took a master’s degree at Stanford and an MBA from Stanford’s Graduate School of Business. In 2014 he started Relcy, a next-generation mobile search company that contemporaries compared to what Perplexity does today. Amazon acquired it, and Nagar moved inside Amazon to work on conversational AI, including the Alexa Prize, a competition to build a machine that could hold an open conversation.
The founding insight came out of that Alexa work. Nagar has said that the big technology companies were pouring effort into AI models but not into the end-to-end workflows where those models would actually earn money, and that frontline workers, customer-service staff above all, were among the largest and least-tooled pools of human labour in the world. Level AI was the bet that ambient, always-listening language AI could augment those workers rather than replace them.
The struggle years: a pivot away from the first idea
The first version of the company was not the contact-centre product it sells today. With about $2 million in seed funding from Eniac Ventures and Village Global, the team first built a voice assistant for frontline workers. It was the customers, not the founders, who redirected the company. Feedback kept pointing at one specific, painful, budget-holding buyer: the contact centre, where the incumbent quality-assurance and analytics tools were dated and where managers were still sampling a tiny fraction of calls by hand.
So Level AI pivoted into contact centres. That is the uncomfortable part of the story that founder decks usually smooth over: the company that later raised tens of millions did so only after abandoning the product it was originally funded to build. Two things had to survive the pivot, and did: the core language-understanding technology, and the discipline to move fast on it. Nagar has framed the early years as a bet on shipping quickly and finding product-market fit rather than perfecting a demo.
The turning point: coming out of stealth
The single turning point was August 2021, when Level AI stepped out of stealth with a $13 million Series A led by Battery Ventures, taking total funding to $15 million. On one side of that event was a small, pre-revenue-scale company still proving its pivot. On the other was a funded contact-centre platform with a marquee lead investor and real usage: the company reported that the volume of customer-service conversations it processed had grown roughly six-fold in the prior six months, and that it was already working with more than 30 brands through enterprise partners, including the large outsourcer Infocision.
Battery’s Neeraj Agrawal, who led the round, put the market thesis bluntly, saying that “half a trillion words are exchanged between companies and customers in contact centers every day” and that those words had until then been “mostly dark data.” That framing, untapped conversational data at enormous scale, is what carried Level AI from a stealth pivot to a fundable category bet, and Agrawal backed it again at the next round.
The money behind it
Level AI’s funding is entirely venture capital, raised in four rounds. The first big figure, about $73.1 million total, is roughly ₹702 crore at $1 ≈ ₹96.0. The shape of the raise:
- Seed, ~$2 million: from Eniac Ventures and Village Global, used to build the original frontline voice assistant.
- Series A, $13 million (25 August 2021): led by Battery Ventures (Neeraj Agrawal), with seed backers Eniac and Village Global and angels; total funding reached $15 million.
- Series B, $20 million (3 March 2022): again led by Battery Ventures, with Eniac and Village Global. Level AI described the round as roughly a four-fold increase on its Series A valuation, struck only about six months earlier, and named new customers including OpenTable, Toast and Carta.
- Series C, $39.4 million (23 July 2024): led by Adams Street Partners (partner Robin Murray), with Cross Creek and Brightloop and continued backing from Battery Ventures and Eniac Ventures. This took total funding to a company-stated $73.1 million.
Two honest caveats. First, Level AI has never published an absolute valuation; the only public marker is that relative “four-fold” jump at the Series B, so any headline unicorn-style number would be invented. Second, sources differ slightly on the totals: the company and TechCrunch cite $73.1 million and a 23 July 2024 close, while the tracker Tracxn lists about $74.4 million across four rounds and dates the Series C to August 2024. Both are within rounding of each other.
How it makes money
Level AI is a business-to-business SaaS company, so the money comes in as recurring subscriptions sold to enterprise contact centres, typically priced against the number of agents and the volume of conversations analysed. There is no consumer product and no advertising. The economics work like most enterprise AI software:
- Money in: annual or multi-year software contracts with mid-market and enterprise support organisations; expansion revenue as a customer adds seats, channels or modules (Auto-QA, Agent Assist, Voice of the Customer).
- Costs out: engineering and machine-learning talent (the India teams), cloud compute and, increasingly, the cost of running large language models over high volumes of conversation, plus enterprise sales and customer-success staff in the US.
- Where the margin sits: in software gross margins once a customer is live, offset in the early years by heavy R&D and the compute bill for processing conversations at scale.
- The part people get wrong: the pitch is not “replace agents.” Level AI sells measurable operational savings on an existing cost centre, faster and broader quality review, shorter handle times and fewer repeat contacts, which is an easier enterprise sale than headcount replacement.
The numbers
Level AI is private and publishes no audited revenue or profit figures, so an honest deep dive cannot show a multi-year profit-and-loss table; anyone who prints one is guessing. What is verifiable is the funding trajectory and the company’s own forward guidance. Amounts are in US dollars, as raised.
| Round | Date | Amount | Lead | Cumulative |
| Seed | ~2019–2020 | ~$2.0M | Eniac Ventures | ~$2M |
| Series A | Aug 2021 | $13.0M | Battery Ventures | $15M |
| Series B | Mar 2022 | $20.0M | Battery Ventures | $35M |
| Series C | Jul 2024 | $39.4M | Adams Street Partners | ~$73.1M |
- Revenue guidance: at the Series C, Level AI told TechCrunch it expected to exceed $50 million in annual recurring revenue within two years. This is company-stated forward guidance, not booked or audited revenue.
- Headcount: about 135 employees at the July 2024 Series C (TechCrunch), with plans to add at least a dozen more within six months; Tracxn lists roughly 194 employees in its 2026 profile.
- Market size (company-stated): Level AI frames a total addressable market of about $35 billion, a serviceable market of about $7 billion and an obtainable market of about $3.4 billion.
Where the demand comes from
Level AI does not break out revenue by segment, but the public customer list points clearly at where the demand sits, and it is not one industry. The named customers span fintech, e-commerce, travel and outsourced support:
- Fintech and financial services: Affirm and Carta.
- Marketplaces, food and travel tech: OpenTable and Toast (added around the Series B).
- Mobility and diversified enterprises: Penske and Vista.
- Outsourced contact centres: Infocision, which serves Fortune 500 brands, and “more than 30 brands” reached through enterprise partners as of the 2021 Series A.
The surprise is geographic rather than commercial. Level AI reads as a US enterprise-software story, but the labour that builds it is Indian: the company lists its base as Mountain View and Delhi, and most of its engineering, machine-learning and quality roles are advertised in Noida, Delhi NCR and Bengaluru. In other words, the value capture (US enterprise contracts, US and global VC) and the value creation (India R&D) sit on opposite sides of the world, a structure that keeps engineering costs efficient while selling into the world’s highest-paying support budgets.
The risks
- A crowded, better-funded field. Contact-centre AI is not an empty market. Level AI competes with the fellow Indian-founded Observe.AI, with Cresta, and with established analytics and workforce vendors such as CallMiner, Verint and NICE, several of which have raised or earn far more. Differentiation and pricing power are the constant pressure.
- Platform and model risk. Level AI’s product runs on top of large language models it does not own. If foundation-model providers, or the big contact-centre-as-a-service platforms such as Genesys, Five9 and Zendesk, fold automated QA and agent-assist into their own stacks, the standalone category could be squeezed. The moat has to be the contact-centre-specific data and workflows, not the underlying model.
- Disclosure and financing risk. The headline $50 million ARR figure is a company-stated projection, not audited revenue, and no valuation is public. A private, VC-funded company at this stage depends on continued fundraising and on converting long enterprise sales cycles; a tighter capital market or slower conversion would bite.
- Privacy and compliance exposure. The product works by recording and analysing customer conversations, which means it processes conversational personal data at scale. That runs straight into consent and privacy regimes, two-party-consent states in the US, GDPR in Europe and India’s Digital Personal Data Protection Act, 2023, and any tightening raises the compliance cost of the core product.
The takeaway
The transferable lesson from Level AI is not “use AI.” It is that the founders were willing to kill their first product. They raised seed money to build a frontline voice assistant, listened when customers kept pointing somewhere else, and moved into contact centres, an unglamorous cost centre with real budgets and dated tools. The pivot, not the original idea, is what attracted Battery Ventures and, later, Adams Street. Paired with that is a quieter structural lesson for Indian technologists: you can build a credible category contender with its commercial head in California and its engineering heart in Delhi and Bengaluru, and treat that split as an advantage rather than an accident.
Frequently asked questions
Is Level AI an Indian company?
It is US-headquartered and India-built. The legal parent, Level AI, Inc., is based in Mountain View, California, but the company was founded by IIT Delhi engineer Ashish Nagar and runs most of its engineering and machine-learning work from India, with roles concentrated in Noida, Delhi NCR and Bengaluru.
What does Level AI actually sell?
Software for contact centres. It automatically scores and analyses customer-support conversations across voice, email, chat and text, gives agents real-time help during calls, surfaces “Voice of the Customer” insights and supports coaching, all built on contact-centre-specific large language models.
How much money has Level AI raised, and from whom?
About $73.1 million (roughly ₹702 crore) across seed, Series A, B and C as stated at its July 2024 round; the tracker Tracxn lists about $74.4 million. Lead investors were Battery Ventures (Series A and B) and Adams Street Partners (Series C), with Eniac Ventures and Village Global among early backers.
Who is Ashish Nagar?
Level AI’s founder and CEO. He has a B.Tech in Engineering Physics from IIT Delhi and graduate degrees from Stanford, founded the mobile search startup Relcy in 2014 which Amazon acquired, and then worked on Amazon’s Alexa conversational AI, including the Alexa Prize, before starting Level AI in 2019.
Is Level AI profitable or publicly listed?
No. Level AI is a private, venture-funded company with no stock-market listing and no audited public accounts, so its profit or loss is not disclosed. Its only public financial guidance is a company-stated target of exceeding $50 million in annual recurring revenue within two years of the 2024 round.
Sources
Figures are as of September 2026. Currency converted at $1 ≈ ₹96.0 as of 18 September 2026 (Trading Economics).
- Level AI blog, “Level AI secures $39.4 million Series C funding” — July 2024
- Level AI blog, “Level AI Launches With $15 Million in Funding” (Series A press release) — August 2021
- TechCrunch, “Level AI applies algorithms to contact center pain points” — July 2024
- Pulse 2.0, “Interview With Founder & CEO Ashish Nagar” — 2024
- FINSMES / MarTech Series / Venture Capital Journal, “Level AI Raises $20M in Series B Funding” — March 2022
- Crunchbase, Level AI company and Ashish Nagar person profiles — accessed September 2026
- Tracxn and PitchBook, Level AI company profiles (funding total, employee count) — 2026
- Battery Ventures, Level AI company page and Neeraj Agrawal commentary — 2021–2022
- Adams Street Partners, “Why We Invested in Level AI” — 2024
- Level AI careers (Lever) and LinkedIn / Glassdoor listings (India engineering locations) — accessed September 2026
Found an error? Write to us and we’ll correct it in the open, dated, on the piece.