Bolna AI

Bolna AI

Voice AI to enable India's next billion.

Bolna AI Product Overview[1]

500+

Enterprise customers

BFSI, e-commerce, BPO, recruiting

>1M

Monthly call minutes

50–60% margins · 10+ Indian languages

<500ms

Conversational latency

Sub-500ms on real phone calls

Thesis

Bolna is a developer-first, India-native voice AI platform that automates phone calls for enterprises with sub-500ms, human-like, multilingual agents (Hindi/English + regional languages), orchestrating best-of-breed ASR/LLM/TTS behind a simple API.[1] The longer-term bet: every Indian enterprise's phone channel — sales, support, collections, recruiting — runs through a Bolna agent rather than a human BPO seat.
  1. 01

    AI tailwinds are uniquely strong in India. OpenAI usage in India has tripled YoY; India is OpenAI's 2nd-largest market with a local office and low-cost plan — accelerating enterprise readiness for AI.[4] [5] [6]

  2. 02

    Call centers are the canonical wedge for voice AI. 2M+ India call center agents and multi-billion annual labor spend create an immediate labor-replacement ROI surface.[2] a16z highlights call centers/BPO as the earliest high-WTP category for production voice agents.[3]

  3. 03

    India-first multilingual orchestration is the defensible moat. Real calls require seamless code-switching (Hindi↔English↔regional), barge-in handling, and noise robustness; Bolna routes per-turn to the optimal ASR/LLM/TTS stack for language, accent, and channel conditions.[1]

  4. 04

    Cost structure and deployment model fit India's constraints. Global infra prices voice minutes at US-centric rates; Bolna's supplier arbitrage across ASR/TTS/LLMs keeps per-minute costs low while preserving 50%+ gross margins at scale.[8] [1]

  5. 05

    Sovereign data, local telco/routing, and compliance are required to win regulated India logos. BFSI and telecom sensitivities favor local hosting and multi-carrier routing; Bolna's India-resident stack reduces latency and compliance risk vs US-hosted competitors.[1]

Problem

India runs on phone calls. No global voice-AI platform was built for them.

Indian enterprises run tens of thousands of calls every day across sales, support, collections, and recruitment. The phone is still the dominant customer touchpoint — in a country with 20+ official languages and 200+ dialects, and a market where every rupee matters.[2]

Real calls don't fit the US-centric mold. They code-switch mid-sentence between Hindi, English, and a regional tongue. They happen over poor connections, in noisy environments, with heavily accented speech that breaks generic ASR. And the unit economics don't tolerate the per-minute rates global infra platforms charge.[8]

The result: India enterprises have the largest voice surface area on the planet — millions of agents, billions in labor spend — and the smallest set of viable tools to automate it. Generic chatbots and US-trained voice APIs ship a worse product than the human BPO seat they're trying to replace.[7]

20+

Official languages

200+ dialects in active use

2M+

India call center agents

Multi-billion annual labor spend

$3.86B → $9B

Outsourcing market

15% CAGR through 2030

Why Now

India is now OpenAI's second-largest market — and voice is the first production AI wedge.

Three converging tailwinds make F25 the right year for an India-native voice infra company.

The macro AI tailwind is hitting India harder than anywhere else.

OpenAI's footprint. Sam Altman has publicly said India is now OpenAI's second-largest market — and could become the largest — with ChatGPT usage tripling year-over-year. OpenAI is opening a New Delhi office and has launched a low-cost India plan, both accelerating enterprise readiness for generative AI.[4] [5] [6]

Voice as the canonical first wedge. a16z's 2025 voice-agents update names contact centers, BPO, and recruiting as the highest-WTP categories — the segments where companies will pay the most, fastest, for production-quality voice agents.[3]

A large, fast-growing market. India's call and contact center outsourcing market is ~$3.86B in 2024 and on track for ~$9B by 2030 (15% CAGR), with voice the dominant channel. Even single-digit-percent automation translates to hundreds of millions of AI call minutes annually.[7]

Call centers and BPO are the earliest high-WTP categories for production voice agents — the place where labor-replacement ROI is most obvious and AI deployment ramps fastest.
a16z, AI Voice Agents: 2025 Update[3]

How It Works

Telephony → ASR → LLM → TTS, routed per turn for India's languages and channels.

A self-serve, developer-first orchestration platform. Provision a number, point it at your CRM, deploy in minutes.[1]

Step 01

Build from transcripts and FAQs

Enterprises stand up a Voice AI agent directly from existing call transcripts and knowledge bases. Full control over agent intelligence, tone, and workflow.

Step 02

Provision a number, plug into your stack

One-click phone provisioning. CRM, OTP, and analytics integrations out of the box. Non-technical builders and developers can both ship.

Step 03

Route per-turn to the best model

Parallel ASR/LLM/TTS pipeline with interruption handling. Real-time routing picks the best-fit model for English vs Tamil vs Hinglish, noisy vs quiet channel.

The orchestration layer is the product — not any single model.

Latency and fidelity. Sub-500ms conversational responses on real phone calls. Barge-in and crosstalk handling for natural dialogues — not the stilted half-duplex feel of generic IVR bots retrofitted with an LLM.[1]

Model-agnostic by design. Bolna mixes best-of-breed ASR, TTS, and LLMs and routes per turn. Analogous to global voice infra (Vapi) but tuned for India's languages, telcos, and price points — and the routing layer compounds: every new dialect or accent makes the next call cheaper and better.[9] [8]

Enterprise controls. India data residency. Configurable policies. Full logs and analytics. Integrations across support, sales, collections, and recruiting — so a single enterprise can deploy Bolna across multiple use cases without changing platforms.[1]

Bolna handles the complexity of voice infra so enterprises can automate calls faster, better, and cheaper, at scale.
Bolna — product overview[1]

Traction

500+ customers, 1M+ monthly minutes, and named logos across verticals.

Cart recovery — GoKwik

~12–14% re-conversion

Bolna agents engaged 400k+ customers; re-conversion approached human benchmarks for cart abandonment outreach.[1]

BPO outreach — Futwork

10k+ daily calls

A BPO partner runs more than ten thousand daily outbound and support calls through Bolna.[1]

Recruiting — Awign

Higher completion rates

Automated interview screening meaningfully improved completion rates vs human-led phone screens.[1]

Platform totals

500+ companies · 1M+ min/mo

E-commerce, fintech, education, BPO. Thousands of concurrent calls in 10+ Indian languages. 50–60% gross margins per minute.[1]

Market

Voice is the first production AI wedge — and India is the densest version of the wedge.

A $9B contact-center outsourcing market on top of a 2M-agent labor base.

Contact center outsourcing. India's call and contact center outsourcing market is ~$3.86B in 2024 and on track for ~$9B by 2030 (15% CAGR). Voice remains the largest channel.[7]

Labor surface. 2M+ India call center agents. Even single-digit-percent automation translates to hundreds of millions of AI call minutes annually — and a sustained tailwind from labor cost replacement at every BPO logo.[2]

Voice as the first AI wedge. a16z names contact centers, BPO, recruiting, and coaching as the top WTP verticals for voice agents — supporting rapid expansion from a single wedge per enterprise into adjacent use cases.[3]

Macro AI tailwind. OpenAI's India expansion and localized pricing accelerate enterprise adoption, energizing vernacular use cases — and the phone is still the dominant customer touchpoint in India.[4] [5] [6]

India's contact center outsourcing market is ~$3.86B in 2024 and on track for ~$9B by 2030 at a 15% CAGR — with voice the dominant channel.
Grand View Research — India Call & Contact Center Outsourcing[7]

Competitive landscape

US-centric infra, premium enterprise suites, omnichannel platforms — none of them are India-native.

Each competitor category has a structural limitation in India. Bolna's positioning maps onto every one of them.

Vapi

Global API-first voice infra; developer-led

Series A · $20M (2024)

Gap vs. Bolna

Not India-tuned for languages, telco peering, or compliance; US-centric per-minute pricing puts a hard ceiling on India adoption.[8]

Uniphore

Enterprise contact center AI platform

Series F · $260M at ~$2.5B (2025)

Gap vs. Bolna

Heavy professional services motion; not developer self-serve; premium pricing misaligned with India budgets. Strong on integrations, slow to deploy.[10]

Yellow.ai

Omnichannel conversational AI suite (chat + voice)

$100M+ raised; India enterprise footprint

Gap vs. Bolna

Voice is one module of many; not optimized for India voice latency, cost, or per-turn model routing.[11]

PolyAI

Premium enterprise voice assistants

Series C · $50M (2024)

Gap vs. Bolna

Top-tier conversational quality, but high-touch and premium-priced — less India focus and not infra-first.[15]

ElevenLabs

TTS / voice cloning (supplier)

Series B $80M (2024); Series C $180M (2025)

Gap vs. Bolna

Best-in-class TTS, but not a full phone-agent stack. Complements Bolna as a routable supplier rather than a competitor.[13]

Bland AI

Horizontal AI phone agents (US)

Series B · $65M (2025)

Gap vs. Bolna

US pricing and US-centric language support; not localized for India cost structure or compliance regime.[17]

Retell AI

Conversational AI for call centers

Seed · $4.6M (2024)

Gap vs. Bolna

Production-oriented features, but early-stage and no India specialization. Smaller customer footprint than Bolna's 500+.[18]

India-first multilingual orchestration is the defensible moat. Every dialect, telco integration, and BFSI deployment compounds — and a US-centric incumbent localizing for India ships a worse product on day one than Bolna's already-live stack.
Investment thesis

Strategic advantages & gaps

The moat is real today. The gaps are the next 12–18 months of execution.

Advantages

  • India-native multilingual orchestration. Code-switching, accents, and noise — with sub-500ms latency and measurable lift in containment and CX.[1]
  • Cost-efficient infra and supplier routing. Tuned to India budgets vs US-centric competitors — preserves 50%+ margins at scale.[8]
  • Developer-first platform. Fast time-to-value vs heavy professional-services suites. Compounding integrations and usage data per logo.
  • Data residency and local routing. Required for regulated industries (BFSI, telecom, government).

Gaps

  • Content and vertical coverage. Must scale across collections, sales, service, and recruiting with reusable flows and guardrails.
  • Enterprise feature depth. RBAC, audit, and SLA tooling need to harden as larger BFSI logos ramp.
  • Internationalization. SEA and MENA expansion require telco peering, language coverage, and compliance buildouts.

Founders

Prateek Sachan

Prateek Sachan

Co-Founder & CTO

Co-founded and leads the technical product at Bolna, building a self-serve enterprise voice-AI platform and orchestration layer that routes calls to best-fit speech/LLM models. Previously built and scaled engineering at Zomato, Tata 1MG, BrowserStack, and Atlassian. B.Tech in Civil Engineering from IIT Delhi (2014). Hands-on operator who took Bolna from initial deployments in 2025 to hundreds of thousands of calls per day, with a seed round led by General Catalyst.

Maitreya Wagh

Maitreya Wagh

Co-Founder

Engineer turned consultant turned AI founder. Co-founded Bolna to build the total voice AI orchestration layer for India — owning go-to-market, customer success, and the commercial motion that took the platform from zero to 500+ enterprises in a year. Loves sports and travel.

Prateek's path to voice infra. Technical founder focused on production-grade voice AI for India's complex, multilingual market. Previously built and scaled engineering systems at Zomato, Tata 1MG, BrowserStack, and Atlassian — all India-scale operations with hard latency and reliability constraints. B.Tech in Civil Engineering from IIT Delhi (2014), where he also co-founded a public street-art project. Builds orchestration and infrastructure rather than a single model — emphasizing routing calls to best-fit models for cost, latency, language, and realism trade-offs.

The early-2025 inflection. Hands-on operator who scaled early commercial deployments rapidly — Bolna moved from initial deployments in 2025 to handling hundreds of thousands of calls per day, and secured a seed round led by General Catalyst with participation from Y Combinator. Led product and technical strategy through the seed raise and YC participation.

Maitreya on the commercial side. Engineer turned consultant turned AI founder. Owns go-to-market, customer success, and the commercial motion that took Bolna from zero to 500+ enterprises in a year, including marquee logos in BFSI, e-commerce, and recruiting. The classic technical-plus-commercial split, with both founders having shipped production systems before.

Why this team for this market. India voice AI is a stack problem — telephony, ASR/TTS routing, LLM workflows, BPO integration patterns, BFSI compliance. It rewards founders who have shipped production systems against India-scale operational constraints and who can carry both the technical product and the enterprise sales motion. Prateek and Maitreya are the rare team with both halves.

Risks & mitigations

Risk

Global infra entrants (Vapi, Bland, Retell) localize for India — pricing, hosting, models.

Mitigation

Maintain a 12–18 month lead via language and dialect tuning, telco peering, and India-resident deployments. Deepen integrations and the data advantage from per-logo usage. Bolna's developer-first GTM and dialect coverage compound with every new enterprise — a US-centric incumbent localizing for India ships a worse product on day one than Bolna's already-live stack.

Risk

Model commoditization erodes the value of orchestration over time.

Mitigation

Differentiate above the model layer: latency and interruption handling, tool use, workflow templates, and analytics. Move toward outcome-based SLAs and vertical packs (collections, recruiting, support) where the value is in the workflow shape, not the underlying model.

Risk

Enterprise trust and accuracy hurdles in regulated BFSI and telecom verticals.

Mitigation

Human-in-the-loop fallback, strict guardrails, full audit trails, and on-prem options where required. Reference wins and KPI-measured deployments in collections and lead qualification de-risk the next BFSI logo.

Risk

Scaling GTM and support beyond founder-led inbound — and telecom reliability at scale.

Mitigation

Build a solutions-engineering function plus partner channels (BPOs and SIs) to compress pilots. Operationally, multi-carrier routing, call-quality monitoring, and automated failover across providers manage the long tail of telecom and compliance complexity.

What we're watching

  • Expansion from BFSI/recruiting/e-commerce wedges into new verticals (telecom, healthcare, government) — and per-vertical reusable flows and guardrails.
  • Internationalization — first SEA/MENA market launch (telco peering, language coverage, residency).
  • Enterprise feature depth: RBAC, audit, SLA tooling, and on-prem options as larger BFSI logos ramp.
  • Per-minute unit economics as model providers and TTS vendors continue to compete on price.

References

  1. [1]Bolna — Features, Pricing & Alternatives (MyBesh)
  2. [2]Tomato.ai — Top 10 Call Centers in India
  3. [3]a16z — AI Voice Agents: 2025 Update
  4. [4]Reuters — India now OpenAI's second largest market, Altman says
  5. [5]Reuters — OpenAI to launch first India office in New Delhi
  6. [6]NDTV — Sam Altman: India could become OpenAI's largest market
  7. [7]Grand View Research — India Call and Contact Center Outsourcing Market (2030)
  8. [8]Vapi — Pricing (indicative US-centric per-minute rates)
  9. [9]TechCrunch — YC-backed Superpowered pivots to voice API platform (Vapi)
  10. [10]Uniphore — Series F press release (NVIDIA / AMD / Snowflake / Databricks)
  11. [11]Tracxn — Yellow.ai company profile (funding / scale)
  12. [12]Yellow.ai — Voice Bots for Call Centers (product page)
  13. [13]Sifted — ElevenLabs raises $180M at a $3B valuation (2025)
  14. [14]ElevenLabs — $80M Series B (blog)
  15. [15]PR Newswire — PolyAI closes $50M Series C
  16. [16]Vapi — $20M Series A announcement (blog)
  17. [17]AI Magazine — Bland raises $65M Series B
  18. [18]Y Combinator (LinkedIn) — Retell AI raises $4.6M seed