A new startup called Poke is trying to make AI agents feel a lot less technical and a lot more familiar.
Instead of asking users to install complex software, manage tools through a terminal, or learn a new interface, Poke lets people access an AI agent through messaging platforms like iMessage, SMS and Telegram. In some markets, it can also work through WhatsApp. TechCrunch says the idea is to make using an AI agent feel as simple as sending a text.
What Poke Actually Does
Poke launched publicly in March and is pitched as a personal assistant that can take action on a user’s behalf through text messaging. According to TechCrunch, it can help with things like daily planning, calendar management, health and fitness tracking, smart home control, photo editing, sports scores, umbrella reminders, medication reminders, and email alerts. Users can also write their own automations in plain text and share them with friends.
The startup also offers pre-made “recipes,” which are basically ready-to-use automations. These cover areas like health and wellness, productivity, finance, scheduling, travel, home, school, email, community and developer tools. TechCrunch reports that these recipes can connect with familiar services including Gmail, Google Calendar, Outlook, Notion, Linear, Strava, Oura, Philips Hue, Sonos, GitHub, Supabase and Vercel.
Why This Startup Stands Out
What makes Poke interesting isn’t that it’s building an AI agent, it’s that it’s trying to remove the technical barrier that still makes agentic AI feel intimidating for most people.
TechCrunch frames Poke as a more consumer-friendly alternative to more technical agent systems, noting that tools like OpenClaw can still feel out of reach for less technical users because they often require software installation, dependency management and deeper system access. Poke’s pitch is the opposite: visit the website, enter your phone number, and start using the assistant over text.
That could matter a lot if the broader AI agent category is going to move beyond developers and early adopters. Many people may like the idea of an AI assistant that can actually do things for them, but not if getting started feels like a mini engineering project. Poke is clearly betting that convenience and familiarity will matter as much as model quality.
A Bigger Push Into Everyday Automation
Under the hood, TechCrunch says Poke chooses the model that best fits the task, using either large AI providers or open-source models depending on the need. The company positions that model-flexible approach as a long-term advantage over rivals tied to a single provider.
The startup is also leaning heavily into growth. TechCrunch reports that Poke has raised an additional $10 million on top of a $15 million seed round from last year, and is now valued at $300 million post-money. The company says it is free to start, with personalized pricing depending on how expensive the requested automations are to run in real time.
That combination of text-first access, integrations, shareable automations and flexible pricing makes Poke feel less like a chatbot and more like an attempt to build a consumer operating layer for everyday AI tasks. Whether it succeeds is another question, but the product direction is clear: bring AI agents out of the lab and into ordinary daily life.
Why this matters for Australia
For Australian readers, the interesting part isn’t just the startup itself, it’s the direction it points to. AI agents are often discussed as if they are only for coders, founders or people happy to tinker with complex tools. Poke suggests the next wave may be delivered through much more familiar channels, including plain old text messaging.
That could make agent-style AI more relevant to everyday users here, especially if the biggest winners in this category turn out not to be the most technically impressive tools, but the ones that feel easiest to use. Scheduling, reminders, email filtering, fitness tracking and smart home control are all much easier to imagine as mainstream features than abstract “agentic workflows.”
The broader takeaway is simple: the race to build AI agents is no longer just about raw power, it’s also about who can make them feel normal enough for everyday people to actually use.
Source links
Source: TechCrunch
