How Perplexity went to market with an answer engine and a founder on X

The X anglePerplexity used Twitter threads to show early products such as Bird SQL in December 2022, and later put its answer engine inside X as the Ask Perplexity bot.

Launching a product one week after ChatGPT should have been a go to market disaster. Perplexity's answer engine went live in December 2022, just as the entire technology audience on X was busy posting screenshots of OpenAI's chatbot. Instead of being drowned out, Perplexity used the moment to define itself by contrast: not a chatbot, but a way to get sourced answers. Its go to market strategy since then has relied on sharp positioning, a relentless shipping cadence and a founder who treats X as a product channel as much as a marketing one.

The product and the bet

Perplexity was founded in August 2022 by Aravind Srinivas, Denis Yarats, Johnny Ho and Andy Konwinski. The core product took a query, searched the web, and returned a written answer with numbered citations pointing to its sources. The interface looked like a search box, not a chat window, and the output looked like a short research note rather than a conversation.

The bet was that a large share of search behavior is really question answering, and that people would trade a page of blue links for a direct answer if they could check where it came from. Citations were the product's trust mechanism. They addressed the single biggest objection to language model answers at the time, which was that they could be confidently wrong.

Positioning against the obvious comparison

Every AI product launched in late 2022 was compared to ChatGPT. Perplexity leaned into the comparison instead of avoiding it. Srinivas told Decrypt that the company was not trying to sell another chatbot or a ChatGPT alternative. The message was that Perplexity was a different category: an answer engine, closer to search than to an assistant.

That framing did two things. It gave journalists and users a simple way to file the product, and it set up the competitive frame Perplexity actually wanted, which was against traditional search rather than against OpenAI. For any go to market strategy for AI startups, this is the central positioning question: which incumbent should a buyer mentally replace?

Who it was for

The early audience was information-heavy knowledge workers: researchers, students, analysts, developers and journalists who searched many times a day and cared whether an answer could be verified. These users were also disproportionately active on X, which made them an efficient first segment. They had real search pain, they could judge citation quality, and they talked about their tools in public.

The launch

The December 2022 launch was modest in format. The product went live on the web, and the team showed it off through posts and examples rather than a staged event. The timing, roughly seven days after ChatGPT, meant there was no shortage of attention on AI. The challenge was to capture a slice of it.

Perplexity's answer was to ship more than one thing. In the same month, the company introduced Bird SQL in a thread on Twitter. Bird SQL was a search interface over Twitter data, using OpenAI's Codex model to translate plain language questions into SQL queries. Example queries in the thread asked for things like the most liked tweets about the World Cup, or for prominent astrophysicists to follow. Analytics India Magazine reported at the time that Jack Dorsey, Twitter's former chief executive, praised the tool.

Bird SQL was clever distribution. A tool that searched Twitter was naturally interesting to people on Twitter. Its outputs were posts and accounts, which users could share straight back into the feed. And it demonstrated Perplexity's underlying capability, turning natural language into structured search, in a context the audience already understood.

How X carried it

Perplexity's relationship with X has moved through three distinct phases, each of which offers a different lesson.

Phase one: demos as threads

In the early months, the company and its founders used threads to show the product working on real questions. This was the classic developer-tool playbook adapted for consumers: show the input, show the cited output, let readers try the same query. The posts were the product demo. Because answers came with sources, a screenshot carried its own credibility.

Phase two: the founder as a shipping log

Srinivas built a reputation for posting frequently about product updates, competitive moves and the company's view of search. Secondary profiles of the founder describe him using X and LinkedIn to share product news and AI explainers. The effect, in analysis, was to make Perplexity's cadence visible. Each new feature, model integration or platform launch became an X moment, and the accumulation of those moments signaled momentum to users, partners and the press.

This is a pattern worth naming for any go to market team: the founder's account becomes the changelog that people actually read. It works when the shipping is real and frequent. It fails when posts outrun the product.

Phase three: the product inside the feed

The most distinctive move came later. Perplexity put its answer engine directly inside X through the Ask Perplexity account. Users could tag the account in a post or reply and get an answer back. In June 2025, reporting from outlets including the Digital Watch Observatory described a new feature that let the bot generate short AI videos on X in response to tagged prompts, available to X users who tagged the account.

This turned X from a marketing channel into a distribution surface. Every answer the bot posted was public, attributed to Perplexity and seen by the followers of whoever asked. It is the logic of Bird SQL scaled up: meet users where they already are, and make every interaction a visible demonstration.

What came next

Perplexity expanded steadily from its web launch:

• Mobile. The company released its iPhone app in late March 2023, a free download that brought the answer engine to the phone.
• A paid tier. Perplexity added a subscription for heavier users, selling access to more capable models and higher usage limits on top of a free tier. Pricing remained freemium in shape: the core answer engine stayed free, and power features sat behind the subscription.
• New surfaces. In July 2025, Perplexity launched Comet, an AI-powered browser built on Chromium. TechCrunch reported that it went first to subscribers of Perplexity's top paid plan and to a limited group of people from the waitlist, a gated rollout that used scarcity to reward the most engaged users.

Across these moves, the consistent thread was surface expansion. Perplexity started as a website, then became an app, a bot inside X, and a browser. Each new surface was a new place where an answer engine could intercept a question before a traditional search engine did.

The go to market lessons

Perplexity is a useful case precisely because it did not have the novelty advantage ChatGPT had. It launched into a crowded moment and still built a distinct position. The lessons below are analysis drawn from that path.

1. Name the category, and name the incumbent

Perplexity's answer engine framing told users what it replaced. It was search, improved, rather than a chatbot, repeated. A clear go-to-market definition starts here: positioning is a decision about which alternative the customer will stop using.

2. Make trust visible in the output

Citations were not only a feature. They were a marketing asset, because every shared answer included its own evidence. Founders building AI products in categories where accuracy is the main objection should ask how the output itself can carry proof.

3. Build tools for the channel you launch on

Bird SQL searched Twitter and launched on Twitter. The Ask Perplexity bot answers inside X. Building something native to the launch channel gives the audience a reason to engage that goes beyond curiosity about a new company. A small, channel-native tool can be a better launch asset than a general announcement.

4. Treat shipping cadence as a marketing system

Perplexity's frequent releases and the founder's habit of posting them created a steady drumbeat. For a b2b go to market strategy or a consumer one, a reliable rhythm of visible updates compounds in a way that a single big launch does not. The prerequisite is an engineering team that can actually sustain the pace.

5. Gate new surfaces to reward the best users

Comet's rollout to top-tier subscribers and waitlisted users is a reminder that access itself can be a benefit. Gating a new product to paying customers first both rewards loyalty and gives the paid tier a reason that is not just a feature list.

6. Launching late is fine if the angle is sharp

The most transferable lesson is the timing. Perplexity launched a week after the most talked about AI product of the decade and still found room. The analysis here is that it did so by refusing to compete on the same axis. A startup entering a hot category should spend less time worrying about being first and more time on being clearly different.

Three ways a company can use X

Perplexity's history shows three distinct uses of X, each of which a founder can choose deliberately. The table below is an analytical summary.

Most startups use only the first. The more ambitious move is the third, where the product itself lives on the platform and every interaction is visible to an audience.

Where it fits among go to market strategy examples

Perplexity represents a middle path between ChatGPT's pure product pull and the heavily orchestrated launches of later AI agents. It combined a sharp category claim, a trust mechanic built into the product, and a founder who used X as a running product log. For founders drafting their own go to market plan, it is a reminder that positioning and cadence can carry a company through a crowded launch window. Compare it with ChatGPT, which it launched alongside, and with DeepSeek, which reshaped the same market two years later.