Tasmela AI Agent Setup: 10 Minutes From Sign-Up to First Message, No Code (2026)

Per the Stanford AI Index 2024 report, the top barrier to enterprise AI agent adoption remains perceived technical effort: 47% of surveyed decision-makers cite integration complexity as the primary blocker. Yet most 2026 AI agent platforms no longer require code. This guide walks through the full Tasmela setup, screenshot by screenshot, from “create account” to a first message to your agent connected to Gmail, LinkedIn, and your CRM.

The goal: remove the technical excuse for not trying. Plan on about 10 minutes hands-on, sometimes more if your Google Workspace admin requires re-consent.

#TL;DR

The Tasmela flow fits five steps: a 3-minute quiz, a 1-minute plan choice, automatic instance provisioning in 2 minutes, connecting key integrations in 3-4 minutes, and a first chat message in 1 minute. No code, no API keys to fiddle with. Everything runs through OAuth and the Tasmela web interface.

#Prerequisites

Before starting, gather the following so the flow doesn’t break mid-way. This saves you from stalling because an admin access is missing.

A professional email address (Google Workspace recommended for smoother Gmail OAuth). The accounts you want to plug in during week one: Google Workspace (Gmail, Calendar, Drive), personal LinkedIn, HubSpot or Notion or Slack per your stack, Telegram for mobile pings. Thirty minutes of focused time to do the setup in one go. A credit card to start the 14-day trial (no charge during trial).

If your Google Workspace is administered by IT, check ahead that third-party apps are authorized for your domain. Otherwise OAuth will stall on a “your administrator must approve this app” message.

#Step 1: the qualification quiz (3 minutes)

You start with the Tasmela quiz, accessible without account creation. Five short steps that determine your recommended plan, your default LLM model, and your suggested integrations.

Quiz step 1: your professional profile (solopreneur, SMB team, agency, other). This shapes the workflow examples the agent suggests on first contact. Step 2: your current stack (CRM used, primary client communication channel, prospecting tools). Step 3: your activity volume (leads per month, team size, frequency of repetitive actions). Step 4: your primary use case (lead qualification, customer support, prospecting, admin ops). Step 5: your email to receive the recommendation.

The quiz isn’t a disguised marketing gate. It genuinely sizes the instance and pre-configures the agent with the right models and suggested integrations.

#Step 2: pick a plan (1 minute)

After the quiz, you land on the recommendation page with a suggested plan. You can override manually.

Starter at €29/month: for a light-volume solo start, 2-3 integrations, €20 of initial LLM credit included. 14-day trial.

Essentiel at €49/month: for active solos or small SMB setups, €30 of initial LLM credit included. 14-day trial.

Pro at €200/month (recommended for teams): cpx32 server (4 vCPU, 8 GB), €100/month of recurring LLM credits included. The sweet spot for most serious B2B setups. 14-day trial.

Business+ at €1,000/month (contact-sales): dedicated cpx42 server (8 vCPU, 16 GB), uncapped LLM credits, dedicated support. For high volume or enterprise requirements.

Honest framing on credits: the plan pays for your server, the agent, and the integrations. LLM credits cover model calls (Claude, GPT, Gemini). If you go past the included quota, you top up in USD. The pricing page details the full tier breakdown.

#Step 3: instance provisioning (2 minutes, automatic)

Once the plan is chosen and payment validated (card saved, no charge during trial), Tasmela launches your dedicated instance provisioning in the background. Here’s what actually happens, with no input needed from you.

A Hetzner cloud server provisions in the Falkenstein region (Germany, EU). The OpenClaw stack installs, your agent receives its initial configuration. The subdomain {your-slug}.agents.tasmela.com is created automatically and points to your server. The OpenRouter key dedicated to your instance is generated for LLM calls. The chat interface readies.

You see a “Booting” state on screen for 60 to 120 seconds, then “Ready”. At that point, the agent is technically alive but knows nothing about your stack until you connect your integrations.

#Step 4: connect your key integrations (3-4 minutes)

You land on the Integrations page. The 22 registered integrations are available, but five are worth turning on first for most setups.

#Google Workspace (Gmail, Calendar, Drive)

Click “Enable”. A Google OAuth flow opens in a new tab, you grant the permissions (mail read, calendar, drive depending on what you want to expose to the agent). If your Workspace admin blocks, you get a clear message: get the Tasmela app validated by IT, then retry.

#LinkedIn (via Tasmela’s integration)

OAuth through Tasmela’s LinkedIn integration. You sign in with your personal LinkedIn account, you grant access to your LinkedIn messaging and feed. The agent can read and draft messages but stays within LinkedIn’s behavioral quotas to protect your account.

#Slack

OAuth per Slack workspace. You pick the channels the agent can read and write in. Best practice: start by giving access to a single test channel, then expand progressively.

#Notion

OAuth Notion with explicit page selection. Limit access to useful pages (product knowledge base, personal CRM, KB) to reduce noise in the agent’s context.

#Telegram

Pair via QR code or bot token per your preferred setup. Useful for mobile reminders, sending receipt photos, or simply chatting with the agent from your phone without opening the browser.

#Step 5: your first message (1 minute)

You arrive at the chat interface at {your-slug}.agents.tasmela.com/chat. That’s where you talk to your agent. The top bar shows the current LLM model (default Claude or GPT per your setup) and a button to switch at runtime.

A useful first message to test calibration: “Summarize my last 10 Gmail messages and flag the 3 most urgent”. The agent reads your inbox, analyzes, and replies in a few seconds. This tests both the Gmail integration and the contextual analysis capability.

To switch the LLM model at runtime, click the selector at the top. You can toggle to Claude for nuanced reasoning, GPT for quick production, or Gemini for multimodal tasks. The switch takes a few seconds (the agent restarts its internal process).

#Week two: calibration

The first message worked, but real value lands with two to three weeks of calibration. Here are the key steps.

Give feedback on the agent’s first actions. When it drafts a miscalibrated email, tell it why. When it tags a record into the wrong segment, correct it with context. The agent internalizes your preferences through feedback.

Tune permissions case by case. If you notice the agent acting too fast on some actions (sending emails), switch back to “propose, don’t execute” mode. You keep the autonomy toggle on each action type.

Add a skill via the skill manager. The platform exposes a skill manager that lets you upload custom skills (specific workflow, source-of-truth access). You can create new ones or download from the catalog.

Create a sub-agent for a dedicated task. For workflows requiring specialized memory (a sub-agent dedicated to LinkedIn prospecting, another to customer support), create them in the interface. Each has its own prompt and memory.

#Common pitfalls

Google Workspace OAuth blocked by admin. Get the Tasmela app validated by IT before the setup. Otherwise, you stall at Gmail connection time. Validation usually takes 24-48 hours on the IT side.

Instance status stuck in “Booting”. This is rare (under 1% of provisionings). If the state lasts more than 5 minutes, email [email protected] with your account email. Manual support intervenes quickly.

LLM credits burning too fast. If you default to a premium model (Claude Opus, GPT-4) on high-volume workflows, credits drain fast. Switch to a more economical model (Claude Sonnet, GPT-4 mini) for repetitive tasks, keep premium for complex reasoning.

Sensitive conversations left on autonomy too early. During the first 7 to 14 days, keep the agent in “propose, don’t execute” mode on outbound emails and LinkedIn messages. Human supervision prevents off-brand outputs.

#FAQ

#How long does setup really take?

Plan 10 to 20 minutes hands-on for a standard setup. Quiz and plan choice take 5 minutes. Instance provisioning takes 2-3 minutes in the background. OAuth for each integration takes 1 to 2 minutes. First message arrives in under a minute. If your Google Workspace admin blocks, plan 1-2 extra days for IT validation.

#Can I try Tasmela without a credit card?

Quiz registration and consulting the recommendation are free. Starting the 14-day trial requires saving a card (no charge during trial). You can cancel before the trial ends to avoid any charge.

#Does Tasmela support multi-user accounts?

Yes. Each Tasmela instance supports multiple users with roles (admin, member). The admin manages billing and integrations, members use the chat and their sub-agents. Configure from the /manage interface once the instance is provisioned.

#Where is my data stored?

Your dedicated instance runs on Hetzner Falkenstein (Germany, EU). Your conversations, internal CRM records, and logs live on that isolated server. LLM calls (Claude, GPT, Gemini) route through OpenRouter per the model provider you pick. Check data residency with your model provider for specific GDPR or regional requirements.

#How do I change plans later?

From /settings/billing, you change plans in a few clicks. Upgrade takes effect immediately (your instance resizes to a larger server). Downgrade takes effect at the next renewal. No lock-in: you can cancel any time from the Stripe portal.

#Conclusion

Setting up a Tasmela AI agent requires no technical skill: quiz, plan, automatic provisioning, OAuth for your integrations, first message. Ten minutes for most cases, no code, no API keys.

Real value lands with two to three weeks of calibration, not on day one. Give feedback on the agent’s first actions, tune permissions, and progressively switch stable workflows to autonomy. To start, the Tasmela quiz takes three minutes. To compare tiers, the pricing page details what’s included.

To go deeper, read our guides on the AI agent replacing a sales employee, the HubSpot AI agent setup, the social media AI agent, and the AI agent for solopreneurs.