Poke launches AI agents across iMessage, SMS and Telegram

Poke raised an additional $10M on top of $15M previously raised, valuing the 10-person startup at about $300M after the latest financing; the texting-based AI assistant publicly launched in March. The service works via iMessage, SMS and Telegram (WhatsApp limited), integrates through Linq, and offers ready-made 'recipes' with integrations across Gmail, Google Calendar, Outlook, Notion and developer tools (PostHog, Webflow, Supabase, Vercel, Sentry, GitHub). Security is multi-layered with penetration testing and permission controls (tokens hidden by default). Pricing starts free, monetization is secondary, and the company plans to catalog thousands of user-created automations while paying creators for referrals.

Analysis

Reducing distribution friction by running autonomous assistants through existing messaging channels materially lowers CAC and accelerates organic growth curves relative to app-install models. If a startup can convert 1-2% of a mature messaging user base into active agent users within 12 months, unit economics flip: marginal CAC falls below $10 while LTV (via paid automations/data access) can exceed $100 within 2 years, creating a steep winner-take-most dynamic for the orchestration layer.

The single biggest structural risk is dependency on third‑party messaging endpoints and their policy cadence: a platform policy reversal or new carrier/telecom regulation can remove critical routes to market in weeks and force either on‑device models or expensive API agreements. Expect two distinct regulatory/corporate reaction windows — immediate (0–6 months) where access is volatile and tactical, and strategic (12–36 months) where incumbents either integrate similar primitives or impose revenue/permissioning models.

Second‑order winners will be providers of secure connector tooling, consented token vaults, and audited execution layers (security vendors, identity platforms, and hyperscalers hosting low-latency inference). Conversely, pure orchestration players without scale or defensible data/control planes will see margins compressed within 24–36 months as platform owners and cloud providers internalize the stack and commoditize routing/recipes.

Key Decisions for Investors

• Long MSFT (6–18 months): buy a modest position in Microsoft or 12–18 month call spreads (e.g., Jan 2028 vertical) sized 1–2% AUM — thesis: hyperscalers capture the hosting and inference demand as agents scale; upside 20–35% if agent usage grows, downside capped by core SaaS revenue; hedge with broad tech put if sentiment deteriorates.
• Long AAPL (6–12 months): accumulate AAPL stock or buy 9–12 month calls (size 1% AUM) ahead of potential API/partnering moves that favor iMessage distribution; R/R: asymmetric upside if Apple tolerates/adopts agent workflows, limited downside versus equity on policy slippage.
• Long security/identity vendors (CRWD or OKTA, 6–12 months): buy CRWD or OKTA outright (0.5–1% AUM each) or 9–12 month call spreads — rationale: demand for token vaulting, permission auditing, and secure connectors rises; expected re-rating if enterprise contracts follow consumer adoption, payoff 30–50% vs sell‑off risk if agent adoption stalls.
• Tactical pair: short META / long MSFT (3–9 months): initiate a small pair (net market neutral sizing) to profit from platform gating risk — if messaging platforms tighten access or monetize, Meta faces growth drag while hyperscalers win; set stop if Meta announces open partner program or a new monetization share that favors platform access.