Bolna Is Doing 200K Calls a Day. Here's Why Infra ≠ Screening Product.

Bolna AI just hit 200,000 daily voice calls. They raised $6.3M from General Catalyst and Y Combinator in January 2026. They scaled from 1,500 to 200,000 daily calls in 12 months. Voice AI infrastructure is clearly working at scale.

So why is HireQwik, a platform built on voice AI, telling you that Bolna’s milestone is irrelevant to your hiring decision?

Because infrastructure is not a product. And the distinction matters significantly when you’re evaluating whether to build your own AI screening solution or buy a purpose-built one.

Section 1: The Voice AI Infrastructure Wave

Bolna is not alone. ElevenLabs, Retell, and a growing number of voice AI infrastructure providers are scaling rapidly. The global voice AI market is projected to hit $17.97 billion in 2026. Inference costs are dropping. API availability is increasing. The underlying technology is becoming a commodity.

This is good news. It means the technology works. Voice AI can conduct coherent, natural conversations at scale. The infrastructure problem — can we do this at all? — is solved.

But infrastructure commoditisation creates a new problem: differentiation moves up the stack. When everyone can access the same voice AI APIs at similar prices, the question shifts from “can we do voice AI?” to “what does a voice AI screening product actually need to do to be useful for HR?”

That’s a very different question. And the answer to it is expensive to build, time-consuming to get right, and almost entirely separate from the underlying voice infrastructure.

Section 2: The Hidden Costs of DIY Voice AI for Hiring

Here’s what Bolna’s cost figures actually represent: Rs.56–90 per call in raw infrastructure costs, before you’ve built anything.

That’s the cost of the voice call itself. Before:

A conservative engineering estimate for building this on top of Bolna or a similar voice API: 6–9 months for an MVP, 12–18 months for a reliable production product. That assumes you have ML engineers, full-stack developers, UX designers, and HR domain expertise on staff.

Most HR teams don’t have any of that. And the ones that do have engineering capacity are not going to prioritise a niche internal tool over their core product roadmap.

Section 3: The Screening Product Layer

What separates a voice AI infrastructure play from a voice AI screening product is the layer that makes it useful for the specific job of evaluating candidates.

HireQwik’s pilot COGS: Rs.27.1 API-only, Rs.47.3 all-in — delivered as a product at Rs.39–59 per interview.

That Rs.39–59 includes everything:

• Structured 15–20 minute voice screening with real-time conversational probing
• Anti-scripting logic — questions adapt based on candidate responses
• Automated campaign setup — launch a drive in under 2 hours
• Candidate scheduling and reminders — candidates self-schedule within your configured window
• Transcript generation — full conversation captured and searchable
• AI scoring rubric — Strong Go / Go / On Hold / No Go with documented reasoning
• HR Decision UI — recruiter reviews classifications with supporting transcript
• Compliance documentation — audit trail, prompt versioning, human-in-the-loop review layer

The product layer is where 90% of the value lives — and 90% of the build time. The infrastructure cost is a small fraction of what it takes to make voice AI useful for an HR team that needs to run a drive next Monday.

Section 4: The Build-vs-Buy Decision Matrix for HR

If you’ve been pitched “build it yourself on voice AI infrastructure,” this is the framework to apply:

Buy if:

• You need to run a campaign in the next 30–90 days
• You don’t have ML engineering and HR product expertise in-house
• Your HR team will be the primary users (not your engineering team)
• You need compliance documentation out of the box
• The value of accuracy matters more than the value of customisation

Build if:

• Voice AI screening is a core competitive differentiator for your business model
• You have 12–18 months and a dedicated product engineering team
• You need deep customisation that no existing product can provide
• You are a voice AI infrastructure company or HR tech startup yourself

For Indian enterprises, GCCs, and IT services firms running campus hiring — buy. Your HR team does not have the bandwidth to manage an infrastructure project on top of a hiring season. And the “customisation” you think you need almost certainly already exists in a purpose-built product.

The economics confirm this: building on Bolna at Rs.56–90/call before UX, workflow, and scoring means you’re starting above what HireQwik charges all-in, before spending a rupee on product development.

The Commoditisation Argument, Flipped

Voice AI infrastructure commoditising is actually an argument for buying a purpose-built product rather than building your own. Here’s why:

When infrastructure commoditises, it becomes table stakes. Everyone has access to the same calls at similar prices. The competitive moat shifts entirely to the product layer — the scoring logic, the anti-scripting prompts, the UX, the HR-specific workflows, the compliance architecture.

A purpose-built product has been investing in that product layer for years. A company that just got access to Bolna’s API has been investing in it for zero days.

Bolna scaling to 200K daily calls means the underlying technology is validated. It does not mean that any company can now build a production-quality HR screening product in a few weeks. The infrastructure problem and the product problem are different problems. Only one of them is solved by Bolna’s milestone.

For Indian HR leaders who’ve been told “we can build this internally on a voice AI platform” — this is the analysis to pressure-test that proposal against.

The Rs.39 all-in question: what is your engineering team’s time worth, and how many months of that time are you willing to spend building a screening product that already exists?

Start a pilot at app.hireqwik.in — and see the full product layer in action before your next campus drive.