Text-First AI Agents Gain Momentum Across Consumer and Workplace Messaging
AI assistants are increasingly moving into messaging platforms like iMessage and WhatsApp, offering autonomous task execution without standalone apps.
Instead of requiring users to download and navigate dedicated applications, a growing ecosystem of artificial intelligence agents is operating directly within everyday messaging channels. By embedding into standard communication protocols such as iMessage, RCS, SMS, Telegram, and WhatsApp, these systems handle scheduling, research, communications, and purchases through standard text threads while retaining memory and connecting to everyday digital services.
The sector has seen rapid investor interest, led by Instinct, which reached a $10 billion valuation after securing $1 billion in funding. However, a wide range of platforms has emerged alongside it, targeting specific niches ranging from household coordination to enterprise automation and travel.
Broadening Approaches to Everyday Assistance
Among the general-purpose assistants, interaction styles and operational scopes vary significantly. Poke, which launched in March 2026, gained early traction by becoming the first AI agent approved on the Apple Messages for Business platform before its parent company, The Interaction Company of California, was acquired by coding startup Cognition in a deal valued in the low nine figures. Another general assistant, Tab, emerged from stealth with a $300 million valuation backed by SV Angel, Valar Ventures, and American Spirit.
Other platforms are designing agents with their own operational identities. Wajo introduced Fo, an agent equipped with its own phone number, email address, and payment card, enabling it to contact vendors or complete transactions without using personal credentials. If Fo encounters an unresolvable obstacle, Wajo routes the issue to a human assistant. Similarly, Instinct has begun assigning dedicated email addresses to assistants so they can handle service registrations and inquiries directly, alongside newly introduced support for placing voice calls.
At the enterprise level, Town raised a $55 million Series A round led by Andreessen Horowitz and Forerunner to deploy "Townie," an assistant integrated with Slack, documents, and calendars that automates recurring work tasks. Open-source options have also appeared; Comma runs across computers, browsers, and messaging apps like Signal and WeChat, providing daily briefings and proactive task completion for free.
Specialized use cases are also seeing text-native adoption. In family organization, tools like Fambot, Ohai, and Ollie aggregate schedules and school reminders into text digests. Ollie, which launched in June 2026, stands out by securing SOC 2 compliance. In travel, Miso utilizes iMessage to pair AI route planning with human travel team assistance, while Stanley offers creator-focused post scheduling and ideation via iMessage and Telegram.
Infrastructure and Architectural Trade-Offs
Building agents that live in chat threads requires distinct architectural decisions around runtime environments and user privacy. While many services depend entirely on cloud infrastructure, Underdog entered an invite-only beta in October 2026 with a local-first approach. By running its models directly on a user's Mac or Windows hardware, it aims to eliminate data transmission to remote servers.
In contrast, Folk provides an isolated cloud computer for each user, allowing the agent to write code and execute complex multi-step workflows rather than simply returning conversational text. Autonomous agents that hold their own credentials and email addresses, such as Instinct, have similarly triggered discussions around security and data privacy as their scope of independent action expands.
What it means for developers
For software engineers and system architects, the expansion of text-native agents represents a departure from traditional mobile app development. Instead of maintaining dedicated frontend interfaces for iOS and Android, developers are shifting focus toward building backend orchestrators that handle natural language via SMS, RCS, and messaging APIs.
Key architectural challenges include:
- Persistent Context and Memory: Services like Caddy, Iris, and Rene distinguish themselves by tracking ongoing conversational history across multiple touchpoints to follow up autonomously.
- Tool Calling and Multi-Step Execution: Modern agents are expected to do more than summarize text; they must interact with external tools, make phone calls, query calendars, and execute code.
- Model Orchestration and Cost Management: Because agentic workflows involve continuous loops of planning, verification, and API calls, token consumption scales quickly. Developers experimenting with these workflows can try top AI models cheaply through one API at https://apixoai.online to evaluate latency, tool-use reliability, and costs across different model providers.
As text-based agents expand from basic reminders to autonomous software operators, developer focus is gravitating toward security boundaries, multi-modal integration, and reliable execution within the messaging apps users already open daily.
Source: Here are the top AI agents that can live in your text messages — TechCrunch AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
