Back to vendors
G

Gnani.ai

Also known as: Gnani AI, Gnani Innovations, Inya

Visit site
Entry priceNot public. Pricing is quoted through enterprise sales.Full pricing detail

Voice AI platform with foundational speech models and agents that run production workflows for enterprises.

Gnani.ai builds its own speech models and the enterprise voice agents that run on them. The company was founded in Bengaluru in 2016 and sells to banks, lenders, insurers, telecom operators and healthcare providers. The model lineup covers speech to text (Prisma), text to speech (Timbre), speech to speech (Warp) and a domain calibrated language model (Evon), trained on more than fourteen million hours of real telephone audio. The same models are sold as APIs, with a Python SDK and LiveKit and Pipecat plugins for developers building their own voice stacks.

On top of the models sits the Inya agent platform. Teams build agents on a knowledge base of uploaded files and web pages, choose each agent's language model from Gnani's own Pampa models, OpenAI or Anthropic, and connect actions to CRM, ticketing, SMS, email and any HTTP endpoint. Workforce links specialized agents on a canvas with conditional handoffs that carry the transcript and extracted data forward, and a transfer step passes a call to a human agent.

Action logs show each call the agent made to an outside system, with the payload, response and status. Ready made agents cover restaurant ordering, hotel concierge, banking collections and service booking, and a Platform API creates and updates agents and pulls conversation logs. Alongside the agents Gnani sells agent assist for human staff, conversation analytics with QA scoring, and passive voice biometrics.

Deployment runs from cloud and private cloud to on premise and air gapped installation. Gnani.ai is SOC 2 and ISO 27001 certified, with GDPR, HIPAA and PCI-DSS compliance, and it publishes no trust center or report. Access control uses three fixed roles, with sharing set for each agent. There is no single sign on, and audit logs are marked coming soon. Testing is manual, through chat, a browser voice session and test calls, and pricing is quoted through sales.

Vendor details

Canonical URL

https://gnani.ai

Category

Voice agent

Subcategory

Enterprise voice AI platform

Funding status

Independent, founded in Bengaluru in 2016 by Ganesh Gopalan and Ananth Nagaraj. Has raised roughly 17.7 million dollars in total, including a 10 million dollar first tranche of a Series B led by Aavishkaar Capital in March 2026, with InfoEdge Ventures participating. Reported EBITDA profitability in fiscal 2025 with revenue more than doubling, and was selected under the IndiaAI Mission to build a 14 billion parameter voice foundation model.

Company status

independent

Use cases & customers

Primary use cases

enterprise voice agentscontact center automationagent assist and speech analyticsvoice biometrics and authentication

Target customers

enterprise

Deployment options

SaaSprivate cloudon-premair-gapped

Integrations

Gnani.ai runs voice and digital agents across telephony, chat, email and messaging channels. The agents connect to existing enterprise systems, and a low code builder comes with prebuilt templates. The full model stack of speech to text, text to speech and speech to speech is also available through APIs for developers, enterprises and OEMs to build on.

In practice

Your contact center handles millions of calls in a dozen languages with heavy accents and code switching. Gnani's voice agents run on speech models trained on real telephone audio, with its background noise and regional accents, and automate the routine calls.

A bank cannot let customer voice data leave its own environment. Gnani runs the voice agents on premise or air gapped inside the bank's environment, with SOC 2 and PCI DSS compliance and passive voice biometrics.

Human agents need help in the moment. Gnani's agent assist delivers hints, compliance prompts and suggested next actions inside the live call, while speech analytics transcribes and scores every call for compliance, performance and coaching.

Agentic Index coverage score

11.5 / 14 capabilities · 82%

Integrations & Tool Calling Full

Agents act in outside systems through actions, including prebuilt CRM, ticketing, SMS and email actions, a Webex Contact Center integration, and custom actions that call any HTTP endpoint with GET, POST, PUT or DELETE and an Authorization header, so the agent can write to the customer's systems during a call.

SourceGnani.ai, docs.gnani.ai Custom Integrations and Actionsread 2026-09-28

Workflow Orchestration Full

Workforce builds multi agent voice workflows on a canvas, where specialized agents are joined by conditional edges from a trigger node and the transcript, extracted data and session context pass to the next agent. Agent Chaining adds decision nodes that branch on conditions and event nodes that transfer or end the call.

SourceGnani.ai, docs.gnani.ai Creating a Workforce and Agent Chainingread 2026-09-28

Knowledge Grounding & RAG Full

Knowledge bases hold uploaded PDF, Word, spreadsheet, CSV and text files and imported public web pages, are processed once content is added, and then serve any agent linked to them during calls. Content is refreshed by uploading new files or removing old ones, and the Platform API manages per agent FAQ entries. Gnani.ai does not describe the retrieval mechanism itself.

SourceGnani.ai, docs.gnani.ai Creating a Knowledge Baseread 2026-09-28

Human Oversight & Guardrails Partial

A transfer event hands the call to a human agent when conditions set in the flow are met, and a QA role must mark an agent ready for production before it ships, which gates the configuration, not individual agent actions. No person approves an agent action before it commits. Agent assist is a separate product that suggests responses to human staff.

SourceGnani.ai, docs.gnani.ai Agent Chaining and Organizationsread 2026-09-28

Security, Identity & Governance Full

Access uses three fixed roles (Org Admin, Developer, QA), with agents private by default and shared read only or with edit rights, which only the creating developer can grant. Gnani.ai states it is SOC 2 certified and ISO 27001 certified, with GDPR, HIPAA and PCI-DSS compliance. There is no SSO or SAML, audit logs are marked coming soon, and no trust center or report is published.

SourceGnani.ai, docs.gnani.ai Organizations and gnani.ai homepageread 2026-09-28

Observability & Auditability Full

Action Logs record each action the agent triggers during a conversation (CRM, email, SMS and custom API calls) with the trigger time, the payload sent, the response, the status code and the parameters, so a customer can see what the agent did on each call. The Platform API returns conversation logs and analytics.

SourceGnani.ai, docs.gnani.ai Action Logs and Platform API introductionread 2026-09-28

Memory & State Persistence Partial

Within a call, a workforce passes the transcript, structured data extracted during the call and session context from one agent to the next. The agent has no store it reads across calls with a set scope and lifetime.

SourceGnani.ai, docs.gnani.ai Creating a Workforceread 2026-09-28

Deployment & Data Residency Full

Deployment options are Cloud, Private Cloud, On-Premise and Air-Gapped, so a customer can run the stack in its own environment, and the speech to text API offers on premise deployment for data sovereignty. The privacy policy names no hosting region.

SourceGnani.ai, gnani.ai homepage, Speech-to-Text API page and privacy policyread 2026-09-28

Prebuilt Agents, Templates & Packs Full

Gnani.ai offers four ready made agents for separate jobs, a Restaurant Ordering Agent, a Hotel Concierge Agent, a Banking Collections Agent and the Neo Service Booking Agent, and prompt templates are offered as a starting point when an agent is created.

SourceGnani.ai, gnani.ai Agents product page and docs.gnani.ai Creating Your First Agentread 2026-09-28

Triggers & Channel Coverage Full

Agents handle inbound and outbound calls, and the Inya Workforce product runs across voice, chat, SMS and email. An inbound call from a customer's own callers starts the agent with no one prompting it, and each workforce starts from a trigger node. Gnani.ai does not describe scheduled campaigns or webhook starts.

SourceGnani.ai, gnani.ai homepage, Inya Workforce page and docs.gnani.ai Creating a Workforceread 2026-09-28

Model Flexibility & Routing Full

Each agent's LLM is a provider and model setting, and the Get LLM Models endpoint lists Anthropic (Claude 3.5 Sonnet), OpenAI (GPT-4.1 mini, GPT-4o mini) and Gnani's own Pampa models. Customers choose among these third party providers, while Gnani's own speech models are fixed.

SourceGnani.ai, docs.gnani.ai Get LLM Models and Creating Your First Agentread 2026-09-28

APIs, SDKs & MCP Extensibility Full

The Agent Builder Platform API at api.inya.ai/platform, authenticated with an x-api-key header, covers agents, agent configuration, per agent FAQ knowledge and conversations (call logs and analytics, by polling or webhook callbacks). The speech models are also sold as APIs with a Python SDK.

SourceGnani.ai, docs.gnani.ai Platform API introductionread 2026-09-28

Testing, Debugging & Optimization Partial

Agents are tested by hand in a chat window, a browser voice session and live test calls from whitelisted numbers, then tuned on temperature, prompt, transcriber and voice. There are no scored test cases, simulations or quality gates. The Inya Insights analytics product scores human agents' calls for coaching, not the voice agent.

SourceGnani.ai, docs.gnani.ai Testing Your Agent and gnani.ai Inya Insights pageread 2026-09-28

Browser & Computer Use Not documented

Agents work over telephony, chat, SMS and email and act through API actions, and they do not control a browser, desktop or computer.

SourceGnani.ai, gnani.ai homepage and docs.gnani.ai Agent Builder pagesread 2026-09-28

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Product updates

·Workflow orchestrationVerified

The Prompt Engine in Gnani.ai's Agent Builder splits a voice agent's conversation into scenarios, each a self contained piece of the flow, joined by transitions set up in advance. On every turn the agent reads only the scenario it is in, not the whole script. Global transitions catch moments such as a callback request, a competitor mention or a request for a human, wherever the call happens to be.

Bears on: Workflow orchestration

View source
·Deployment / data residencyVerified

The agentic conversational platform from Gnani.ai is now live as a fully colocated stack in the US, Singapore, Japan and the UAE, with production voice conversations already running. In each region, orchestration, inference, conversation data, tool calls and observability run in the country across voice, chat, text and email, offered as a managed in region deployment or dedicated single tenant infrastructure.

Bears on: Security / enterprise

View source
·Agent capabilityPartially Verified

Gnani.ai launched Artha, a sovereign enterprise AI stack that combines its new 30-billion-parameter open-weight foundational model, Evon v3.3, with the Plexus platform for AI agents.

Bears on: Agent capability

View source
View all 4 Gnani.ai updates →Tracked since Aug 2026 · Verified from public vendor sources

Pricing

Not public. Pricing is quoted through enterprise sales.

The billing unit is not published.

What is public

No public rate. Gnani publishes no pricing page, plan or per minute figure. Model APIs and enterprise deployments are quoted through sales.

Billing mechanics

Not published. Commercial terms for agents, agent assist, analytics, biometrics and the speech model APIs are quoted through sales. Deployment options run from cloud and private cloud to on premises and air gapped.

Cost watchouts

On premises and air gapped deployments and multiple product modules can each affect pricing. Which of voice agents, agent assist, analytics and biometrics a quote includes, and whether interaction volume is metered, is not published.

Variable cost rationale

No billing unit is published, so what the bill meters is not stated. Each quote covers some mix of voice agents, agent assist, analytics, biometrics and the speech model APIs, so more modules and a heavier deployment raise the price. No rate, cap, minimum or commitment structure is published, so nothing public limits the bill as usage grows.

Additional watchouts

No billing unit is published, so cost cannot be tied to call or minute volume from public material. How on premises or air gapped deployment changes the commercial terms is not published either.

Sales call required

Yes, required for paid access

Free / trial

None published, quoted through sales

Lowest paid plan

Not published

Key ambiguities

No entry rate, per minute or per interaction figure is published, and the billing unit is not stated. All pricing is quoted through sales.

Agentic Index verified 2026-09-28

Alternatives to Gnani.ai

The closest documented capability profiles to Gnani.ai among voice agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

  • Smallest AI11.0 / 14A lighter documented profile than Gnani.ai
  • Autocalls10.5 / 14A lighter documented profile than Gnani.ai
  • Floatbot11.5 / 14Fuller documented coverage on Testing, Debugging & Optimization
  • Phonely11.5 / 14Fuller documented coverage on Testing, Debugging & Optimization
  • Synthflow12.5 / 14Fuller documented coverage on Memory & State Persistence and Testing, Debugging & Optimization
  • Vapi11.5 / 14Fuller documented coverage on Testing, Debugging & Optimization

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.