Bolna
Bolna runs multilingual voice AI agents for inbound and outbound phone calls in Hindi, Hinglish, Tamil and Telugu
If your callers pick up in Hinglish or Tamil, Bolna is one of the few stacks where that is the default rather than a plugin. Graph Agents and the October 2026 SIP trunk release give you real control over routing and telephony, and per-minute economics are legible: 6.00¢ standard, down to about 4.51¢ on larger top-ups and about 4.2¢ on the $500 pilot pack. The $5 signup credit won't tell you much. Budget a Pilot if you need a fixed per-minute rate, but note it caps concurrency at 25 — GoKwik and Futwork run 250+ concurrent calls, which is volume-plan territory. Compare against VAPI or Retell AI only if your calls are English-first.
Verified 1h ago · liveness 73/100 · cite: rightaichoice.com/tools/bolna
- Indian e-commerce teams running COD confirmation, cart recovery and collections at scale
- Recruitment firms screening candidates in Hindi, Hinglish or regional languages
- BFSI teams handling EMI reminders, collections and vernacular inbound triage
- Enterprises with existing SIP carrier contracts that want outbound AI calls plus human escalation
- Teams that need a flat unlimited monthly seat price — Bolna meters minutes and tokens
- English-only, US-centric operations better served by VAPI or Retell AI
- Pilots that need more than 25 concurrent calls before committing to a volume plan
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Bolna if your callers are English-only and US-based, or if you need a flat unlimited monthly seat price rather than per-minute and per-token metering.
Your agents stop making calls the moment credits hit zero — set automatic top-ups or low-balance notifications before a campaign starts.
Pay-as-you-go suits a team running a first campaign off $10 to $5,000 of credits; the $300 or $500 Pilot buys a fixed ~4.6¢ or ~4.2¢ per-minute rate for a one-off proof of concept, capped at 25 concurrent calls. Volume plans start at 6.00¢/minute and fall to about 4.51¢ at the $3K top-up tier. Enterprise is committed-volume and custom. Compared with Retell AI or VAPI, Bolna's floor is competitive for Indian-language calls, but the seat-free model means cost tracks minutes — bill at high volume
In short
Bolna — Bolna runs multilingual voice AI agents for inbound and outbound phone calls in Hindi, Hinglish, Tamil and Telugu. Best for Indian e-commerce teams running COD confirmation, cart recovery and collections at scale, Recruitment firms screening candidates in Hindi, Hinglish or regional languages, BFSI teams handling EMI reminders, collections and vernacular inbound triage. Plans from $5.
What's new in Bolna
Checked 2 days agoAcross the latest 8 updates: 4 feature updates, 1 community discussion and 3 news mentions.
Static graph nodes can end the call
A static node with no transitions can hang up after its message plays, instead of leaving the caller in silence until timeout.
SIP trunks: bring your own carrier to Bolna
Connect your own SIP carrier and run agents on existing numbers. Carriers: Plivo, Telnyx, Twilio, Exotel, Vobiz, Vonage. Legacy SIP trunks deprecated, no sunset set.
Taking the Second Call with Workflows
Bolna Workflows give each call outcome a next step: retries, waits, extraction, API calls and WhatsApp follow-ups.
How to get started with Graph Agents on Bolna
Walkthrough of Graph Agents: nodes, transitions, routing order and per-node control, plus a step-by-step build guide.
Graph Agents: Where it all started?
Why Graph Agents split a call into nodes with one job each instead of one long prompt carrying every step.
BigQuery to ClickHouse at 15M call minutes a day
Bolna moved call analytics from Datastream/BigQuery to ClickHouse Cloud, making call data real time and cutting costs roughly 6x.
Testing Jev on real phone calls
TypeSafe's Jev benchmarked on three real jobs against 17 models: 1.7-3.3x faster, level with the best on judging and extraction, not ready for routing.
Shipping browser calls in under two weeks
Bolna shipped browser-based voice agents in about three weeks by reusing its existing Kamailio and FreeSWITCH telephony stack.
What people actually say about Bolna — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
5 mentions across 2 sources (Hacker News, Lemmy) · researched Jul 3, 2026.
Average across the 2 sources that answered — each source counts once, not each post.
- +Open-source platform allows full customization and self-hosting.
- +Supports pluggable LLMs, TTS, and ASR models for flexibility.
- +Designed specifically for Indian languages including Hinglish, Tamil, Telugu.
- +Sub-300ms latency for natural conversation flow.
- +No-code playground enables non-developers to build agents quickly.
- −Almost no community feedback or independent reviews available.
- −Recent buzz is absent — only one relevant post from June 2025.
- −No real-world performance or reliability data from users.
- −Hidden reliability issues may surface at scale without warnings.
- −Support responsiveness is unknown due to lack of user reports.
- • Usage overage charges for calls and API calls
- • Additional cost for premium support or SLAs
Viability Score
How well maintained and how widely used is Bolna? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Multilingual voice agents in 10+ Indian and foreign languages including Hinglish, Hindi, Tamil and Telugu (docs claim 40+ languages)
- Graph Agents: node-based call flows with transitions and per-node control
- Static graph nodes can end the call after their message plays, with optional interruptible goodbyes
- Per-value definitions and post-call values for graph agent extraction
- Workflows that act on call outcomes: retries, waits, extraction, API calls and WhatsApp follow-ups
- Inbound and outbound calling from one agent configuration
- Bulk campaigns at thousands of simultaneous calls
- Custom API triggers fired in real time during a live conversation
- Human-in-the-loop transfer to a live agent mid-call
- SIP trunks: bring your own carrier (Plivo, Telnyx, Twilio, Exotel, Vobiz, Vonage or standards-based SIP)
- Sarvam Bulbul v4 Flash text-to-speech: 222 voices across 9 languages, including 13 Hindi-English code-mixed voices
- Bring-your-own OpenAI-compatible TTS server, with the key stored encrypted
- Sarvam saaras:v4 speech-to-text with keyword biasing up to 50 terms
- Keyword and context biasing across seven transcriber providers
- Model switching per call across 20+ ASR, LLM and TTS providers
About Bolna
Bolna is a voice AI platform for teams whose callers speak Hinglish, Hindi, Tamil or Telugu before English. You pick a pre-built agent (Reminders, Customer Support, Lead Qualification, Announcements) or build one in Agent Studio, attach a phone number or your own SIP trunk, and run inbound and outbound calls off the same configuration. Bulk campaigns push thousands of concurrent calls, and Bolna reports 1,000+ companies on the platform with published case studies from Awign, Hyreo, GoKwik, Hypothesis and Futwork. Speech, telephony and orchestration sit in one place. Agents handle interruptions with sub-300ms replies, transfer to a human mid-call, and fire custom API calls while the conversation is still running. Graph Agents split a call into nodes with one job each rather than one long prompt, so routing stays readable, and as of 15 October 2026 a static node with nowhere to go can hang up instead of leaving the caller in silence. You can now bring your own SIP carrier — Plivo, Telnyx, Twilio, Exotel, Vobiz, Vonage or any standards-based carrier — on numbers you already own, and speak through your own OpenAI-compatible TTS server. Workflows give each call outcome a next step: retries, waits, extraction, API calls, WhatsApp follow-ups. Model switching is per call across 20+ ASR, LLM and TTS providers including Sarvam AI (now with Bulbul v4 Flash, 222 voices across 9 languages), Deepgram, Soniox and AssemblyAI; BYOK covers LLM, TTS and ASR keys if you want to hold provider spend yourself. Since 9 October 2026 Bolna is also listed in Claude's connector directory, so you can manage agents from a Claude chat. The docs claim 40+ languages. Compared with English-first stacks like VAPI or Retell AI, the Indian language mix is the default here rather than a plugin, and billing is per minute and per token rather than per seat.
Behind the Verdict
Bolna's centre of gravity is the Indian language mix, and that shows in the component choices rather than the marketing. Sarvam saaras:v4 handles speech-to-text with keyword biasing up to 50 terms, Sarvam Bulbul v4 Flash (added 15 October 2026) gives you 222 voices across 9 languages including 13 Hindi-English code-mixed ones, and Soniox tts-rt-v2 covers 63 languages when you need to step outside India. Keyword and context biasing works across seven transcriber providers, which matters most for exactly the thing English-first stacks handle badly: names, product SKUs and city names in code-mixed speech. The October 2026 SIP trunk release is the biggest structural change in this refresh. You can now point Bolna at your own carrier — Plivo, Telnyx, Twilio, Exotel, Vobiz, Vonage or any standards-based SIP carrier — add your existing numbers, and route inbound to an assigned agent or the trunk's default. Outbound leaves through your carrier's gateways. Each trunk has a channel count shared across inbound calls, outbound calls and transfers, and echo tests let you confirm audio before an agent is involved. Legacy SIP trunks are deprecated with no sunset date set. If you already have carrier contracts, this removes the main reason to keep a second voice vendor around. Where Bolna is weaker is the operator experience around money. Credits are the unit of account: pay-as-you-go top-ups run $10 to $5,000, every API call is billed on tokens processed, and agents stop making calls when the balance hits zero unless you set automatic top-ups or low-balance alerts. The Pilot plan is a one-time purchase — $300 for 6,500 minutes (~4.6¢/min) or $500 for 12,000 minutes (~4.2¢/min), billed in 30-second pulses — and it hard-caps concurrency at 25. That is fine for a proof of concept and not fine for a national campaign; the published case studies cite 250+ concurrent calls. After pilot minutes are used the account reverts to standard pricing from 6¢/minute. Graph Agents and Workflows are the two features that separate Bolna from a prompt-and-hope voice bot. Graph Agents break a call into nodes with one job each and explicit transitions, and the October release added per-value definitions and post-call values for graph-agent extraction so the structured output you get back is defined up front. Static nodes can now end the call once their message plays, and interruptible goodbyes keep a call alive if the caller talks over the sign-off. Workflows pick up where the call ends: retries, waits, extraction, API calls, WhatsApp follow-ups. Custom API triggers fire mid-conversation, so your backend can act before the call is over. The data side got a real engineering pass: analytics moved from Datastream/BigQuery to ClickHouse at roughly 15M call minutes a day, which the vendor says cut costs about 6x and made call data real time. Browser-based voice agents shipped in about three weeks by reusing the existing Kamailio and FreeSWITCH stack. That reuse is also the honest
Researching Bolna? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Bolna actually fits — and what changes day-one when you adopt it.
You connect your own Twilio SIP trunk on 15 October's release, add your existing COD numbers, and assign the Reminders agent template to the trunk's default inbound agent. Bulk campaigns dial cash-on-delivery orders in Tamil and Telugu, and a static node with End call after this message hangs up after the confirmation instead of leaving silence.
Outcome: Fewer failed deliveries because confirmations happen in the caller's language, and no dead air at the end of each call — while your carrier contract and number inventory stay where they are.
You build a Graph Agent for EMI reminders with one node for identity check, one for balance disclosure and one for payment link delivery, then attach a Workflow that retries unanswered calls after two hours and sends a WhatsApp follow-up when the outcome is a promise-to-pay.
Outcome: Per-node routing keeps the compliance-sensitive script readable, and every call outcome has a defined next step rather than dropping into a spreadsheet.
You POST to the call API from your Python service when a signup stalls, passing user_data variables into a Lead Qualification agent, and listen for the post-call extraction values the graph agent returns.
Outcome: A voice touchpoint fires without your team opening the dashboard, and the structured extraction lands back in your own system for scoring.
Use Cases
- Answer inbound support calls 24/7 in Hindi and English with FAQ handling and triage.
- Recover abandoned carts by calling shoppers in their preferred Indian language with WhatsApp links.
- Confirm cash-on-delivery orders in Tamil or Telugu to cut failed deliveries and returns.
- Screen and interview job candidates at scale with an agent that speaks Hinglish.
- Send EMI, payment and collections reminders for BFSI portfolios in regional languages.
- Run national outbound campaigns at 250+ concurrent calls across multiple languages.
- Trigger a call from your own backend mid-workflow using the call API and custom API triggers.
- Route live calls through your existing carrier by connecting a SIP trunk to your numbers.
Models Under the Hood
as of 2026-10-15
Limitations
- Pricing is usage-based rather than a flat seat licence: pay-as-you-go credits range from $10 to $5,000, and volume pricing starts at 6.00¢ (₹5.52) per minute, dropping to about 4.51¢ with larger top-ups.
- Speech-to-text, LLM and text-to-speech are bundled into the per-minute rate only when your agent uses the bundled models — telephony and the platform fee are added per minute.
- The Pilot plan is a one-time purchase ($300 for 6,500 minutes or $500 for 12,000 minutes) that caps concurrency at 25, so it will not suit high-concurrency tests; once those minutes are spent the account reverts to standard pricing from 6¢/minute.
- Customised integrations and deployment, priority support and a dedicated account manager sit on committed-volume or enterprise plans.
- Agents stop making calls when credits run out unless you configure automatic top-ups or low-balance notifications.
as of 2026-10-18
Verification history
We have re-verified Bolna 9 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 9 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Bolna tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Free Credits
$5
Ideal for
A developer or founder who wants to trigger a handful of live calls and hear the voice quality before spending anything.
What this tier adds
Starting tier: $5 in signup credits, enough to test Agent Studio and a template rather than run a campaign.
Pay-as-you-go
$10-$5,000 credit top-ups
Ideal for
A small team running ad-hoc outbound batches or an internal proof of concept where call volume is unpredictable month to month.
What this tier adds
Adds flexible credit top-ups from $10 to $5,000 with no commitment, token-based metering and BYOK for LLM, TTS and ASR.
Pilot Plan
$300 for 6,500 min or $500 for 12,000 min
Ideal for
A team that needs a fixed per-minute rate and a defined budget for a one-month evaluation at up to 25 concurrent calls.
What this tier adds
Locks in about 4.2¢-4.6¢ per minute for 6,500 or 12,000 minutes, adds enterprise features, WhatsApp priority support and four 1:1 POC calls.
Volume-Based Pricing
From 6.00¢ (₹5.52) per minute
Ideal for
An operations team past evaluation that is running recurring campaigns and wants the per-minute rate to fall as committed top-ups rise.
What this tier adds
Moves from the 6.00¢/minute standard rate down to about 4.51¢ at the $3K tier and removes the Pilot plan's 25-call concurrency cap.
Enterprise Plan
Custom
Ideal for
A large organisation committing to sustained call volume that needs customised integrations, deployment choices and a named account manager.
What this tier adds
Adds committed-volume custom pricing, customised integrations and deployment, priority support, a dedicated account manager and India/USA data residency with on-prem options.
Where the pricing makes sense
The company stage and team size where Bolna's pricing actually pencils out — and where peers do it cheaper.
Pay-as-you-go suits a team running a first campaign off $10 to $5,000 of credits; the $300 or $500 Pilot buys a fixed ~4.6¢ or ~4.2¢ per-minute rate for a one-off proof of concept, capped at 25 concurrent calls. Volume plans start at 6.00¢/minute and fall to about 4.51¢ at the $3K top-up tier. Enterprise is committed-volume and custom. Compared with Retell AI or VAPI, Bolna's floor is competitive for Indian-language calls, but the seat-free model means cost tracks minutes — bill at high volume
Setup time & first value
How long it actually takes to get something useful out of Bolna — broken out by persona, not the marketing-page minute.
A no-code builder can sign in, clone a pre-built template and trigger a first call in under 30 minutes. A developer wiring the call API needs a token, an agent_id and a from number — roughly an hour to first successful call. Bringing your own SIP carrier is the longest path: create the trunk, hand your carrier Bolna's IPs, add numbers, then run echo tests in both directions before attaching an
Switching to or from Bolna
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From VAPI or Retell AI: rebuild the prompt as a Graph Agent so each step is its own node, then point your existing numbers at a Bolna SIP trunk.
- →From a legacy Bolna SIP trunk: follow the migrate-to-SIP-trunks guide — legacy trunks keep working with no sunset date set, so you can move trunk by trunk.
- →From a scripted IVR: map each IVR menu branch to a graph node with a transition, then recreate the DTMF fallback using trunk keypad input.
- →From a spreadsheet-driven outbound process: import the list into a bulk campaign and let Workflows handle retries and WhatsApp follow-ups.
- ↗To VAPI or Retell AI: export your agent prompts and re-implement the call flow in their builder; expect to rebuild graph routing logic by hand.
- ↗To an in-house Kamailio/FreeSWITCH stack: keep your carrier trunk and move only the agent layer, re-pointing trunk.bolna.ai traffic to your own gateway.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “Bolna”, and we withheld 6: 6 could not be judged, because “Bolna” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Bolna.
Official links
Tools that pair well with Bolna
Common stack mates teams adopt alongside Bolna, with the specific reason each pairing earns its keep.
Awaz.ai
No-code AI voice agents that run outbound and inbound phone calls, book meetings, and take call-time actions for you.
Synthflow AI
Synthflow AI runs enterprise voice AI agents on its own telephony network for automated phone calls at scale.
Vapi
Vapi is an API-first platform for building voice AI agents that run real inbound and outbound phone calls at sub-500ms latency.
Featured Head-to-Head Comparisons
Bolna vs Locus Robotics
These tools occupy completely different niches: Locus Robotics automates physical warehouse labor, while Bolna automates phone conversations in Indian languages. Buyers should decide based on operational domain, not compare them head-to-head. If you need to move boxes, choose Locus; if you need to handle calls, choose Bolna.
Bolna vs Truleo
Truleo and Bolna serve completely different markets. Truleo is a specialized AI intelligence platform for law enforcement, automating lead generation from siloed data. Bolna is a conversational voice AI platform for Indian businesses, offering multilingual support and flexible model integration. Choose Truleo if you are a police department wanting to cut case research time; choose Bolna if you need vernacular voice bots for customer calls.
Bolna vs Presto Voice
If you run a QSR drive-thru chain in the US, Presto Voice is the clear choice due to its specialized upselling engine and proven 95% non-intervention rate. For any Indian-market business needing multilingual voice bots with flexible model choice, Bolna wins with its 10+ language support and BYOK model integration. Pick the one that matches your geography and use case.
Alternatives to Bolna
View all
Awaz.ai
No-code AI voice agents that run outbound and inbound phone calls, book meetings, and take call-time actions for you.
Synthflow AI
Synthflow AI runs enterprise voice AI agents on its own telephony network for automated phone calls at scale.
Frequently Asked Questions
Categories
Best-of guides
Used Bolna? Help shape our editorial sentiment research.