Ada (company)
Last reviewed
Jun 4, 2026
Sources
20 citations
Review status
Source-backed
Revision
v1 ยท 2,367 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 4, 2026
Sources
20 citations
Review status
Source-backed
Revision
v1 ยท 2,367 words
Add missing citations, update stale details, or suggest a clearer explanation.
Ada (legally Ada Support Inc., and branded at the domain ada.cx) is a Toronto-based software company that builds an AI-powered platform for automating customer service. The product is an AI agent that resolves customer inquiries across channels such as chat, email, voice, and SMS, drawing on a company's own knowledge and policies to answer questions and complete tasks without a human agent. Ada was founded in 2016 by Mike Murchison (chief executive officer) and David Hariri, and is named after Ada Lovelace, the 19th-century mathematician often described as the first computer programmer. The company coined the term "Automated Customer Experience" (ACX) to describe its category, later reframing it around "Agentic Customer Experience" as the product shifted from rule-based chatbots to large language model-powered agents. Backed by investors including Accel, Spark Capital, Bessemer Venture Partners, FirstMark Capital, and Tiger Global, Ada reached a 1.2 billion US dollar valuation in 2021, making it one of Canada's AI unicorns.
Ada grew out of a failed first venture. In 2014, Mike Murchison and David Hariri launched Volley, a social network for connecting founders, developers, designers, and other people in technology. Volley raised more than 500,000 US dollars from Version One Ventures and Fastbreak Ventures but struggled to scale, in part because of the volume of customer service inquiries the small team had to handle manually.
Concluding that customer support was a universal and underserved problem, the two pivoted. By their account they spent roughly a year working as front-line support agents inside seven different companies to understand the work first-hand. They found that around 30 percent of inquiries were repetitive and that existing support tooling was poor and did not scale. They quietly tested an early version of Ada inside one of those companies and observed that customers could not reliably tell the automated answers apart from human ones, which they took as validation. The company was named after Ada Lovelace.
Murchison studied cognitive science, psychology, and human-computer interaction at the University of Toronto, and was selected for the Canadian entrepreneurship program The Next 36. Hariri, a designer and developer, had previously worked at the Toronto studio Teehan+Lax, which was later acquired by Facebook (now Meta); at Ada he has led research and development as co-founder.
Ada emerged from stealth in July 2017 with a 2.5 million US dollar seed round led by Bessemer Venture Partners, with participation from Version One Ventures. At the time the company said it had gone from answering tickets manually to automating more than 2 million customer questions a month, for early customers including Shopify, Coinbase, Telus, and Medium. The platform was positioned as a proprietary machine learning engine that could deflect support volume and let customers find answers on their own.
Ada raised five named equity rounds between 2017 and 2021. The figures below are in US dollars unless noted.
| Round | Date announced | Amount | Lead investor(s) | Valuation | Selected other investors |
|---|---|---|---|---|---|
| Seed | July 2017 | $2.5M | Bessemer Venture Partners | not disclosed | Version One Ventures |
| Series A | December 2018 | $19M | FirstMark Capital | not disclosed | Bessemer, Version One, Leaders Fund, Burst Capital, Barney Pell |
| Series B | March 2020 | $44M | Accel | not disclosed | Bessemer, FirstMark, Version One, Leaders Fund, Burst Capital |
| Series C | May 2021 | $130M | Spark Capital | $1.2B | Tiger Global, Accel, Bessemer, FirstMark, Burst Capital |
The 19 million US dollar Series A in December 2018 was led by FirstMark Capital (with FirstMark's Matt Turck joining the effort) and was earmarked for international expansion and for moving into verticals such as travel and financial services. At the time Ada said it had about 70 employees and aimed to double that in 2019, and that more than 30 million customers worldwide had received automated support through the platform. The company described its system as able to automate up to about 70 percent of customer interactions.
The 44 million US dollar Series B, announced on March 19, 2020 and led by Accel, came as Ada emphasized its no-code, drag-and-drop builder for non-technical teams and support for many languages. Ada described handling tens of millions of personalized conversations a year for customers including AirAsia, Mailchimp, Shopify, Telus, Upwork, and Zoom.
The 130 million US dollar Series C, announced on May 7, 2021 and led by Spark Capital with Tiger Global participating, valued Ada at 1.2 billion US dollars and gave the company unicorn status. Ada said the round brought its total funding to roughly 200 million US dollars (databases such as Crunchbase later put the cumulative total at about 190 to 192 million US dollars). The company reported it had automated more than 1.5 billion brand interactions in 2020, served more than 350 businesses including Facebook (Meta) and Square (Block), and had grown revenue nearly sixfold in under three years. Spark Capital's Yasmin Razavi joined in connection with the round, and Ada said it would use the capital to expand globally, grow to more than 500 employees, and invest in research and development.
Ada's platform began as a chatbot built on traditional natural language processing and natural-language understanding, where teams authored answer flows in a no-code interface. The company moved aggressively into generative AI once large language models matured.
Ada integrated GPT-3 and other large language models into its product as early as January 2023. On April 18, 2023 it announced what it called the first omnichannel generative AI suite for customer service, letting companies ground an agent in their existing knowledge base so that "build time for the bot is essentially zero," in Murchison's framing, and deploy the same automation across messaging, voice, and other channels. The company stressed a proprietary pipeline layered on top of foundation models to keep answers accurate, safe, and relevant, and to mitigate hallucination, which Murchison called one of the hardest problems in applying generative AI at runtime.
Through the rest of 2023 Ada layered on agentic capabilities. In October 2023 it introduced Generative Actions, letting teams build third-party integrations in plain language so the agent could take actions and pull from multiple sources without code, and around the same time moved to outcome- or resolution-based pricing, charging for conversations the agent actually resolves rather than per interaction. In November 2023 it launched a new AI Agent powered by the Ada Reasoning Engine, positioning the product around "automated resolutions" and the ability to solve more complex, action-oriented inquiries.
In February 2026 Ada announced a unified Reasoning Engine, described as a patent-pending, single intelligence layer (a single "brain") that an enterprise can configure once and have replicated across voice, messaging, social, and email in many languages, drawing on the same knowledge, policies, and brand standards in every channel. Ada described the design as a dual-reasoning architecture that gives immediate, empathetic replies while working through more complex tasks in the background, built on a multi-model (multi-LLM) foundation. At that announcement the company said it had powered more than 5.5 billion interactions since 2016, had 550 or more AI agents deployed, and operated across 85 or more countries.
Ada sells a single AI agent for customer service that a company configures with its own knowledge, policies, and tone, then deploys across channels. The platform is intended to be operated by non-technical staff (support, marketing, and operations teams) rather than engineers.
The company frames all of this with the language of "Automated Customer Experience" (ACX), and more recently an "Agentic Customer Experience" operating model that it describes as combining technology, methodology, and services to build and maintain high-performing agents. Ada has also historically offered a more traditional scripted, rule-based chatbot alongside the generative AI agent.
Ada moved to outcome- or resolution-based pricing in 2023, charging customers for conversations the AI agent resolves rather than for every interaction or per seat. Conversations that fall back to a human are generally not billed as resolutions. The company has positioned itself toward higher-volume operations; third-party comparisons describe Ada as a fit for organizations with very large annual conversation volumes and as carrying a longer, professional-services-assisted implementation than some lighter-weight competitors. Ada does not publish standard list pricing.
Ada reports its scale primarily through volume of automated interactions and number of customers, and through customer case studies.
These adoption and revenue figures come largely from Ada's own statements and third-party databases and should be read as company-reported.
Ada competes in a crowded market for AI customer service. Its closest comparisons include Intercom's Fin AI Agent, and a wave of newer agent-focused startups such as Decagon, Sierra (co-founded by former Salesforce co-CEO Bret Taylor), and Forethought, as well as incumbents and platforms like Zendesk, Salesforce (Einstein and later Agentforce), and IBM watsonx Assistant.
Independent comparisons tend to position Ada as an enterprise-grade, platform-agnostic agent that sits on top of an existing help desk, with deeper automation and resolution-based pricing, versus Intercom's Fin, which is native to Intercom's own suite, priced per resolved outcome, and generally faster and lighter to deploy. Ada often emphasizes its early move into conversational AI and generative AI (it was building automated support from 2016 and integrated large language models in early 2023, ahead of some incumbents that launched AI agents only after GPT-4) and its focus on a single, channel-spanning reasoning layer. As with all of these products, the underlying value proposition rests on reliably resolving customer issues, and analyses note the persistent risks of LLM hallucination and the commoditization pressure from general-purpose AI assistants.