Giga
Also known as: Giga ML, GigaML, giga.ai
Giga offers enterprise AI support agents for voice, chat, email and messaging that resolve requests in business systems, with a browser agent, a CLI and SDK, and an agent that proposes and tests its own improvements.
Giga (Giga AI, Inc., San Francisco, formerly Giga ML) sells AI agents that resolve customer support work across voice, chat, email and messaging in 99 languages. The agents act inside the conversation, issuing refunds, updating accounts, applying discounts and changing subscriptions. A Browser Agent handles systems with no API. It launches a cloud browser, logs in and completes the workflow, or, with the customer's approval, takes control of the cursor on the company's website while every step stays visible and can be stopped. Context carries across channels, so a follow up email picks up where yesterday's call left off.
Teams build agents in Agent Canvas by writing policies and workflows in plain language, starting from transcripts, recordings or SOPs, and test them with simulated conversations before a staged rollout that starts on 1 percent of traffic. Scout, Giga's own analysis agent, drafts agents from policies, reads resolved and failed conversations, tests safe changes on a slice of traffic and holds everything else for a person.
Nothing goes live until the team approves it. Insights records the intent, the actions taken and the outcome of each ticket and turns patterns into recommended changes. A CLI and Python SDK (giga-sdk, MIT licensed on PyPI) drive agents, tests, tickets, KPIs, the knowledge base and deploys over a REST API, and the developer docs sit behind a customer login.
Giga holds SOC 2 Type II, ISO 27001 and ISO 42001 certification and runs a trust center. AWS, Microsoft Azure and Google Cloud are its hosting subprocessors, and there is no customer hosted option or region choice. Pricing is quoted through sales, and Giga publishes terms for an eight week enterprise pilot.
Vendor details
Canonical URL
https://giga.ai
Category
Voice agent
Subcategory
Enterprise voice and chat support agents
Funding status
Giga is independent and headquartered in San Francisco, and was founded in 2023 as Giga ML by Varun Vummadi and Esha Manideep Dinne. It has raised about sixty five million dollars across three rounds, including a three point six million dollar seed in 2023 led by Nexus Venture Partners and a sixty one million dollar Series A in November 2025 led by Redpoint Ventures, with Y Combinator and Nexus participating, at an undisclosed valuation. It runs a lean team and serves large business to consumer enterprises including DoorDash and Zepto, with reported production resolution rates above ninety percent.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Agents act in the business's systems during the conversation (refunds, account updates, discounts, subscriptions) and reach legacy tools without APIs through a cloud browser. A CLI and Python SDK over a REST API manage agents, tests, tickets, the knowledge base and deploys. The connector list and API reference sit in login gated docs.
In practice
Your call center is buried in repetitive calls about order status and account changes, and hold times are long. Giga's voice agents answer in the customer's language and resolve most issues end to end, including refunds and account updates, without a human.
Some of your support work lives in legacy tools with no API. Giga's Browser Agent logs in through a cloud browser and completes the workflow itself, with every step visible.
You keep discovering too late that a policy is confusing customers. Giga's Insights and its Scout agent read across conversations, pinpoint the root cause, test a fix on a slice of traffic and wait for your approval before it goes live.
Sources & related URLs
Agentic Index coverage score
10.5 / 14 capabilities · 75%
| Integrations & Tool Calling | Full |
|---|---|
|
Agents carry out transactions in the business's systems inside the conversation, issuing refunds, updating accounts, applying discounts and managing subscriptions, and send follow up emails. Giga does not publish the list of supported systems. SourceGiga, giga.ai Omnichannel pageread 2026-09-28 |
|
| Workflow Orchestration | Full |
|
Agent Canvas defines support rules and multi step workflows in plain language for voice, chat or multimodal agents, deploys them through versioned, git synced releases, and Scout, a second agent, reads conversations and proposes and tests changes to the first. SourceGiga, giga.ai Agent Canvas and Scout pages and CLI post (14 Jul 2026)read 2026-09-28 |
|
| Knowledge Grounding & RAG | Full |
|
Each agent has a knowledge base that the CLI and SDK manage alongside agents and deploys, and Scout drafts agents from the business's policies and knowledge. How retrieval works is not public. A managed, updatable knowledge base points to a maintained retrieval layer, though Giga does not say so. SourceGiga, giga.ai CLI post (14 Jul 2026) and Scout pageread 2026-09-28 |
|
| Human Oversight & Guardrails | Full |
|
Scout tests only safe changes on a slice of traffic and routes risky ones to human review, and nothing it proposes goes live until the team approves it. The Browser Agent takes control of a website only with the user's approval, and the user can stop it at any time. Teams also set when the agent answers, asks a question or hands off to a person. SourceGiga, giga.ai Scout and Browser Agent pages and homepageread 2026-09-28 |
|
| Security, Identity & Governance | Partial |
|
Giga reports SOC 2 Type II, ISO 27001 and ISO 42001 and runs a trust center at trust.giga.ai, and the CLI uses API keys and rotates credentials. There are no roles, SSO or other controls over who in the customer's organization can change an agent, and the developer docs sit behind a login. SourceGiga, giga.ai homepage, CLI page and trust.giga.airead 2026-09-28 |
|
| Observability & Auditability | Full |
|
Each ticket records the customer's intent, the actions the agent took and how the exchange ended (resolved, transferred or abandoned). Full transcripts with intents and tags are pulled from the CLI, Insights explains patterns across conversations, and every Browser Agent step is visible as it runs. SourceGiga, giga.ai Introducing Insights post (19 Oct 2025) and CLI pageread 2026-09-28 |
|
| Memory & State Persistence | Partial |
|
Context carries across channels and conversations for each customer, so a customer who called yesterday and chatted this morning gets an email that picks up where they left off. That context is kept per customer, though Giga gives no lifetime or delete path for it. SourceGiga, giga.ai Omnichannel pageread 2026-09-28 |
|
| Deployment & Data Residency | Not documented |
|
There is no region choice and no customer hosted or customer cloud option. Hosting runs on AWS, Microsoft Azure and Google Cloud, which Giga names as its hosting subprocessors. SourceGiga, giga.ai Data Processing Addendum and homepageread 2026-09-28 |
|
| Prebuilt Agents, Templates & Packs | Partial |
|
Agents are drafted for each business, by Scout from its policies and knowledge or in Agent Canvas from transcripts, recordings or SOPs. Each customer assembles its own agents this way, and Giga offers no named prebuilt agents or template catalog. SourceGiga, giga.ai Scout and Agent Canvas pages and sitemapread 2026-09-28 |
|
| Triggers & Channel Coverage | Full |
|
Inbound calls, chats, emails and messages (SMS, WhatsApp, iMessage) from the business's customers start the agent without anyone at the business prompting it, and the agent sends follow up emails that carry the conversation's context. SourceGiga, giga.ai Omnichannel page and homepageread 2026-09-28 |
|
| Model Flexibility & Routing | Not documented |
|
Giga names no language or voice model provider and offers customers no choice of model. Its subprocessor list is kept in the trust center. SourceGiga, giga.ai Voice Experience page, homepage and Data Processing Addendumread 2026-09-28 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
The Giga CLI and Python SDK (giga-sdk, MIT licensed on PyPI) run over a REST API authenticated with API keys and cover agents, tickets, tests, KPIs, secrets, the knowledge base and deploys, with JSON output for scripting. Its API reference at docs.giga.ai sits behind a customer login. SourceGiga, giga.ai CLI page and CLI post (14 Jul 2026)read 2026-09-28 |
|
| Testing, Debugging & Optimization | Full |
|
Agent Canvas runs simulated test cases, the CLI runs test suites with CI gates before deploy, rollouts start on 1 percent of traffic and scale on KPI results, and Scout tests changes on a slice of live traffic against the KPI the team sets. SourceGiga, giga.ai Agent Canvas and Scout pages and CLI post (14 Jul 2026)read 2026-09-28 |
|
| Browser & Computer Use | Full |
|
The Browser Agent drives a real browser. It launches a secure cloud browser, logs in to browser based and legacy tools without APIs, and clicks, selects and validates inputs through the workflow. With the user's approval it can also take control of the cursor on the business's website, with every action visible. SourceGiga, giga.ai Browser Agent pageread 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
Event Telemetry records each step a Giga agent takes, including calls, actions, code runs and API requests, as ordered events that teams pull as JSON with the Giga CLI. Events carry the agent version that handled a conversation, say why a transfer failed, and can be read back for up to 30 days. Scout, the assistant in the Giga console, reads the same events to explain what went wrong in a conversation.
Bears on: Observability / auditability
View sourceGiga launched real-time synthetic voice detection to screen live inbound calls for AI-generated voices. The system dynamically analyzes conversations to flag machine speakers, leveraging training pipelines optimized for telephony conditions and adversarial robustness.
Bears on: Security / enterprise
View sourceGiga voice AI agents can now independently schedule and place outbound callbacks based on customer requests. The platform preserves the conversational context and memory from the initial call, allowing the agent to resume the interaction at the scheduled time.
Bears on: Workflow orchestration
View sourcePricing
Not published. Pricing is quoted through enterprise sales.
The billing unit is not published.
What is public
There is no list pricing and no pricing page. The enterprise pilot terms are public, but rates are not.
Billing mechanics
Billing is not published. Pilots run eight weeks at the cost in a pilot order form. Production needs a new order form and a master services agreement. The billing unit is not stated.
Cost watchouts
The pilot is priced in its own order form and does not roll into production automatically. Pilot fees and production contract terms are set separately.
Variable cost rationale
No billing unit is published, so nothing public says whether the bill grows with interactions, resolutions, minutes or seats, or what limits it. The eight week pilot is priced in its own order form, and production is priced separately under a new one.
Additional watchouts
No public rate is published, and nothing says whether pricing is per resolution, per minute or per seat.
Sales call required
Yes, required for paid access
Free / trial
Eight week enterprise pilot, priced in a pilot order form. No free tier.
Lowest paid plan
Not published
Key ambiguities
There is no public rate and no stated billing unit (per resolution, per minute or per seat). Pricing is sized through a sales engagement.
Cancellation / refund
The eight week pilot does not convert to a paid subscription automatically, and Giga may end it early. Production needs a new order form and a master services agreement.
Related vendors
- AgentVoice — AgentVoice builds and runs inbound and outbound AI voice agents, and…
- aiola — Voice agents for field teams on Salesforce
- Aircall — Aircall's AI Voice Agents answer and place calls on Aircall's own…
- Allo — Allo is a business phone system for small teams with an AI…
- Autocalls — Autocalls is a no code platform for AI voice agents that place and…
- AviaryAI — Outbound AI voice agents for credit unions, banks and insurers,…
Alternatives to Giga
The closest documented capability profiles to Giga among voice agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Regal10.5 / 14Adds documented Model Flexibility & Routing
- Salient9.5 / 14Fuller documented coverage on Security, Identity & Governance and Prebuilt Agents, Templates & Packs
- Telli8.5 / 14Adds documented Model Flexibility & Routing
- Callers8.0 / 14Fuller documented coverage on Security, Identity & Governance
- careCycle9.0 / 14Fuller documented coverage on Security, Identity & Governance and Prebuilt Agents, Templates & Packs
- Consio7.0 / 14A lighter documented profile than Giga
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded