Startup Deep Dive : Tars — built on customer revenue, not the venture rounds its records suggest

Tars has run for close to a decade without a single institutional venture round changing hands in public records — or so its co-founder has said in interview after interview. Yet Tracxn and Crunchbase both list a $100,000 seed cheque from a Bengaluru accelerator in 2018, credited to a company that calls itself bootstrapped.

That contradiction is the story of Tars: a Bengaluru-built, US-incorporated conversational-AI company that turned a failed WhatsApp concierge experiment into a no-code chatbot builder now used by 800-plus brands, grown almost entirely on customer revenue rather than investor capital, with the one accelerator cheque it did take barely registering next to what rivals in the category have raised.

Quick facts

What they do

Tars sells a drag-and-drop platform for building AI-driven chatbots and conversational agents that businesses deploy on their websites, on WhatsApp and over SMS, in place of static forms or standard support widgets. The company states it has automated more than 60 million customer conversations for over 800 global brands across finance, healthcare, retail, government and education, with agents live in more than 10 countries (hellotars.com/about, accessed September 2026). Its customers are mostly marketing and customer-support teams at mid-sized and enterprise businesses that run long, form-heavy sign-up or lead-capture flows — insurance, real estate, mortgages, legal services and auto dealerships were the verticals Ish Jindal named as the earliest fit (Mixergy, July 2019).

The origin

Before Tars, Ish Jindal had built Padhaaro, a travel-experiences marketplace. Running Padhaaro’s customer support over WhatsApp groups, he kept seeing the same handful of questions from travellers, asked over and over, each one answered by a human greeter (Mixergy, July 2019). That repetition was the seed of the idea: if a bot could hold the conversation and still get the traveller what they needed, nobody had to retype the same answer for the hundredth time. Jindal paired with technical co-founder Vinit Agrawal to test it, and the two decided the insight was bigger than travel — it was about how anyone collects information from anyone else on a small screen.

The struggle years

The first version of that idea was not a chatbot builder at all. In June 2015, Jindal and Agrawal launched a WhatsApp-based concierge service: four interns sitting behind a phone number, answering whatever a user asked, from ordering pizza to finding an electrician to buying a SIM card (Mixergy, July 2019). It worked, in the sense that people used it — but it was a three-month lesson in what the founders did not want to build. Agrawal wanted to write software, not manage a shift of interns fielding random requests, and the operational load of a general concierge service could not scale the way a product could. They shut it down.

The second attempt, from January 2016, came from a narrower and more deliberate observation. Studying the websites of home-services companies, the founders zeroed in on one specific, unglamorous problem: long web forms on mobile screens had terrible completion rates. Their fix was to turn the form into a conversation — one question at a time, in a chat interface, instead of a page of fields (Mixergy, July 2019). It took most of 2016 to find a paying customer willing to bet on it. For six months, from January to June 2016, Jindal cold-emailed roughly 3,000 recently funded startups and spoke with 500 to 600 founders directly, trying to find anyone who would pay for a converted form (Mixergy, July 2019). For close to two years he was effectively the company’s only salesperson, working from roughly 2pm to 4am India time so his calls would land inside a US business day (Mixergy, July 2019).

The turning point

The first paying customer arrived in December 2016: Timesaverz, an Indian home-services company, signed on at $99 a month (Mixergy, July 2019). Before that signature, Tars had a hand-built demo and no revenue. What followed over the next few months changed the shape of the company. A Product Hunt launch on 21 February 2017 pulled in more than 950 upvotes (Mixergy, July 2019), turning a niche tool into something a wider audience of marketers had heard of. Then, in April 2017, Tars ran a lifetime-deal campaign through AppSumo that added roughly 2,000 accounts in one push, against a base that had started at around 100 accounts when the self-serve builder first went live (Mixergy, July 2019). Going from one $99-a-month customer to thousands of signed-up accounts inside five months is the pivot point the rest of the company’s growth sits on: it is when Tars stopped being two founders’ bet on a hunch and became a product with a funnel, a price and a queue of self-serve users to convert.

The money behind it

This is the part where the record gets genuinely contested. Two things are both true and both sourced, and they do not fully agree:

• Tracxn’s company profile records Tars raising a total of $100,000 in a seed round in 2018, with Upekkha named as the investor (Tracxn, accessed September 2026).
• Tars has been part of the Upekkha Catalyst Accelerator since 2017, a Bengaluru-based accelerator built around the “ValueSaaS” philosophy of capital-efficient SaaS companies that stay independent of large venture rounds (Tars blog, “How To Build A Profitable B2B SaaS Business Without Giving Away All Your Equity”, August 2020).
• In two separate founder interviews — Mixergy (July 2019) and Just Go Grind episode 132 (August 2020, titled “Bootstrapping from $0 to $1M in ARR”) — Ish Jindal frames Tars as self-funded, growing on customer revenue rather than outside capital.

Read together, the most defensible account is that Tars took a small accelerator-style cheque from Upekkha as part of joining its 2017 cohort, and has otherwise run on its own revenue — which is consistent with both the $100,000 figure on Tracxn and a founder who describes the company as bootstrapped rather than venture-backed. What is not in the record anywhere: a Series A, a named venture fund beyond the accelerator, or a priced valuation. Reporting elsewhere that a firm called Hat-trick Capital also invested could not be independently confirmed from a primary source and is left out here.

Total raised, on the only figure that is actually sourced: $100,000. Latest valuation: not disclosed, and not found in any filing or press report as of September 2026.

How it makes money

Tars is a subscription SaaS business — customers pay to build and run chatbots on the platform, rather than paying per conversation or per lead.

• Pricing point of entry: the first paying customer, Timesaverz, signed at $99 a month in December 2016 (Mixergy, July 2019); the company has since layered on tiered and enterprise plans as its customer base moved from small self-serve accounts toward larger brands.
• Customer profile: businesses with long, form-heavy conversion funnels — real estate, insurance, mortgage, legal services and auto dealerships were the verticals the founders targeted first, on the thesis that a conversational form lifts completion rates more where the original form is longest (Mixergy, July 2019).
• Distribution: growth ran through a self-serve builder plus one-time promotional pushes — a Product Hunt launch (February 2017) and an AppSumo lifetime-deal campaign (April 2017) that together took the account base from about 100 to roughly 2,000-plus almost overnight (Mixergy, July 2019) — rather than a large outbound sales team.
• Cost structure: the founders have pointed to Bengaluru-based operations as the reason the model could work on small revenue — monthly operating costs in the $300–670 range in the early years, against thousands of dollars for equivalent US overheads, and engineering hires at roughly $10,000 a year versus $100,000–150,000 for comparable US roles (Mixergy, July 2019). That cost gap is the mechanism behind the “ValueSaaS” positioning Tars later adopted through Upekkha: profitability on modest revenue is only possible because the cost side is this much lower.
• What people get wrong: the headline growth story (“Product Hunt hit, AppSumo deal”) suggests a viral consumer product. The actual mechanics are a B2B lead-generation tool sold to marketing and support teams — the promotional spikes brought in accounts, but the durable revenue came from renewed monthly subscriptions from businesses using it for its stated job, form conversion, not from one-off deal-hunters.

The numbers

Tars has never filed for an IPO and does not publish audited financials, so the figures below are the disclosed or third-party-estimated data points that exist, not a continuous filed series. Profit or loss has not been disclosed for any period.

The gap between the GetLatka data point for 2020 (about $240,000 annualised) and the Just Go Grind episode title for the same year (implying roughly $1 million in ARR) is a real conflict in the public record, not a rounding difference. Both are named here because neither could be independently verified against the other, and because inventing a reconciled number would be worse than showing the disagreement.

What can be said with more confidence is headcount, which moved consistently upward across every data point found: 4 employees in July 2018, 6 in December 2022, 28 in December 2023, 34 in October 2024 (GetLatka, accessed September 2026), and 49 as of August 2026 (Tracxn, accessed September 2026).

Where the money comes from

Tars does not publish a revenue split by geography or industry, so there is no percentage breakdown to report. What is documented, from the company’s own published case studies, is which kinds of customers it points to as proof of the product working, and the numbers each one claims:

• State of Indiana (US public sector): the company states its deployment is “saving an average of 4,000-plus calls a month” for the state (hellotars.com/about, accessed September 2026).
• UCI Merage School of Business (US higher education): more than 5,000 interactions logged from website visitors (hellotars.com/about, accessed September 2026).
• Unitel, a telecom operator in Cabo Verde: a reported 5% reduction in contact-centre call volume within four months of deployment, with 20% month-on-month growth in agent usage (hellotars.com/about, accessed September 2026).

The surprise, given the company’s Bengaluru roots and early Indian home-services customer base, is how far its published case studies now sit from that starting point — a US state government, a US business school and a West African telecom operator, rather than Indian SMBs. That is consistent with the platform’s own claim of deployments in 10-plus countries (hellotars.com/about, accessed September 2026), but it also means the Indian SMB origin story and the current published customer logos describe two different markets, several years apart, with no disclosed figure for how much revenue still comes from India versus elsewhere.

The risks

• Category commoditisation from general-purpose AI. Tars’s original edge — turning a web form into a guided, one-question-at-a-time chat — is a pattern that large language model providers and no-code tool vendors can now replicate inside broader products. A company built around a single conversational-form mechanic is exposed if that mechanic becomes a default feature elsewhere rather than a reason to buy a separate platform.
• Thin, self-reported financial disclosure. Every revenue figure available for Tars — from GetLatka, from a podcast title, from Tracxn’s estimate — is either self-reported by the founders or a third-party estimate, and two of those figures for the same year (2020) disagree by roughly 4x. That level of disclosure makes it hard for outside partners, acquirers or larger enterprise customers to underwrite the company’s scale with confidence.
• Capital-light growth caps how fast it can out-spend better-funded rivals. The founders’ own account of the business — low-cost Bengaluru operations, a single accelerator cheque, no priced venture round — explains its survival on modest revenue, but the same structure limits how much it can spend on sales, partnerships or enterprise features against venture-funded conversational-AI competitors with far larger war chests.

The takeaway

Tars’s most transferable lesson is not the Product Hunt spike or the AppSumo bump — those were distribution tactics, useful once and not repeatable as a strategy. It is that the founders let the market tell them what to build twice, and were willing to shut down something that was already working, in a small way, to build the second thing. The WhatsApp concierge service was not a failure by the usual definition — people used it — but it was answering a demand the founders did not want to keep serving. Only when they narrowed from “any question, answered by a human” to one specific, boring, well-defined problem — mobile forms have bad completion rates — did the company find something a self-serve customer would pay $99 a month for without a salesperson pushing them. The insight that scales is usually smaller and duller than the one you start with.

Frequently asked questions

Who founded Tars and when?

Ish Jindal and Vinit Agrawal founded Tars, starting with a WhatsApp concierge experiment in June 2015 and pivoting to a conversational-forms product from January 2016 (Mixergy, July 2019).

How much funding has Tars raised?

Tracxn records $100,000 raised in a 2018 seed round involving Upekkha, the Bengaluru accelerator Tars joined in 2017 (Tracxn, accessed September 2026). Co-founder Ish Jindal has separately described the company as bootstrapped in interviews (Mixergy, July 2019; Just Go Grind, August 2020), and no larger venture round or priced valuation is on the public record.

What does Tars actually sell?

A no-code platform for building AI chatbots and conversational agents deployed on websites, WhatsApp and SMS, used mainly for lead-generation forms and customer support (hellotars.com/about, accessed September 2026).

Is Tars profitable?

Tars has not published a profit or loss figure for any year. Its founder-led positioning through the Upekkha accelerator’s “ValueSaaS” framing emphasises capital efficiency and low overhead rather than venture-scale losses, but that is a stated philosophy, not an audited number (Tars blog, August 2020).

How big is Tars today?

Employee count has grown from 4 in July 2018 to 49 as of August 2026 across the sources checked (GetLatka and Tracxn, both accessed September 2026), and the company states it serves more than 800 global brands with deployments in 10-plus countries (hellotars.com/about, accessed September 2026). It remains a private company with no confirmed valuation.

Sources

Figures are as of September 2026. Currency converted at $1 ≈ Rs 96.0 as of 18 September 2026 (Trading Economics).

• Mixergy — “Solving the problem of form abandonment (with a more modern experience),” interview with Ish Jindal, July 2019
• Just Go Grind — Episode 132, “Ish Jindal, Co-Founder and CEO of Tars, on Bootstrapping from $0 to $1M in ARR with a SaaS Company for Marketers,” August 2020
• Tars blog (hellotars.com) — “How To Build A Profitable B2B SaaS Business Without Giving Away All Your Equity,” August 2020
• Tars (hellotars.com/about) — company history and case-study page, accessed September 2026
• Tars (hellotars.com/team) — team page, accessed September 2026
• Tracxn — Tars company profile (funding, employees, revenue estimate), accessed September 2026
• GetLatka — Hellotars financial and headcount data points, accessed September 2026

Found an error? Write to us and we’ll correct it in the open, dated, on the piece.