What is BotTrottr? A Tour of Your Agent Command Center

BotTrottr Team · · Product

Most AI tools are conversational. You open a chat, you type, it answers, you close the tab — and nothing happens again until you come back. That works for asking questions. It doesn’t work for the things you actually want handled on a recurring basis: the morning brief, the weekly digest, the inbox you keep meaning to triage, the number you keep forgetting to check.

BotTrottr is built for that second category. It’s a dashboard where AI agents run on a schedule, in the background, and report back to you — whether or not you’re watching.

The idea in one sentence

You describe what an agent should do and how often it should do it. BotTrottr runs it, handles the model calls and tools, and delivers the result to your inbox.

What an agent actually is

In BotTrottr, an agent is a small bundle of decisions:

  • A task — the instruction, written in plain language. “Summarize new support emails and flag anything urgent.”
  • A model — which AI provider does the thinking. Bring your own API key, or use the platform default.
  • A schedule — how often it runs. Hourly, daily, weekly, or on a custom interval.
  • Tools & data — what it’s allowed to reach: web search, connected services, files in its own storage.
  • A destination — where the output lands. Email is the default.

Configure it once, and it keeps working. You can pause it, edit it, or run it on demand whenever you want to test a change.

The anatomy of a run

Here’s what happens on every scheduled execution:

  1. The scheduler wakes up. A ticker checks which active agents are due based on their interval — no cron syntax, no server wrangling on your side.
  2. The agent assembles its context. Your task, its stored files, its connected tools, and its skills get pulled together into the run.
  3. The model executes. The agent works the task, calling tools as needed — searching, reading, computing.
  4. The result gets delivered. Output is formatted into a clean, readable email and sent. Media results — a generated image or video — come through with a preview and a direct link.
  5. The run is logged. You can see what ran, when, and what it produced.

If something fails, you find out. Errors surface in the dashboard and, for the platform owner, as alerts — a silent failure is worse than a noisy one.

Skills: teaching agents to do specific things

A skill is a packaged procedure — a SKILL.md with instructions, plus optional reference files and scripts. Shell out a repeatable workflow once as a skill, and every agent that loads it follows the same steps every time.

Skills are organized in tiers, and the most specific one wins when names collide:

  • Platform Skills — trusted defaults from BotTrottr.
  • Swarm Skills — shared across every agent in a Swarm, installed by a maintainer.
  • User Skills — your own, attached only when you explicitly choose them.

Only a skill’s name and description load up front. The full instructions are pulled in on demand when a task actually matches — so agents stay fast even with a deep library. For the full picture of this system, see our post on Teams, Swarms & Skills.

A Swarm is the environment a group of agents belongs to. It owns the shared skills, memory, and connectors. Bind an agent to a Swarm and it inherits that capability set — so five agents working the same domain stay consistent instead of drifting apart.

This matters because the most confusing failure mode in multi-agent setups is an agent behaving differently in production than it did in testing. Shared, explicit configuration kills that problem.

Bring your own key, or don’t

BotTrottr works with your own provider API keys — the free plan includes support for bringing your own. Prefer not to manage keys? Use the platform default instead.

Either way, agents are capped by plan rather than nickel-and-dimed per run. The free plan covers up to 5 agents; paid plans raise the ceiling from there.

Where agents tend to earn their keep

A few patterns that fit the model well:

  • Recurring briefings — a daily digest of what changed in a market, a repo, a news beat, or an inbox.
  • Monitoring — watch a source, and only speak up when something crosses a threshold.
  • Summarization — turn a firehose of long content into a short read on a fixed cadence.
  • Generated media on a schedule — have an agent produce and deliver an image or video on a recurring basis.

The common thread: the work is repetitive, the schedule is predictable, and you don’t want to be the one remembering to do it.

Getting started

  1. Create an account at bottrottr.app.
  2. Make an agent — write the task in plain language and pick a model.
  3. Set a schedule — start with daily; short intervals are easy to add later.
  4. Run it once on demand to check the output looks right.
  5. Let it run. Results start landing in your inbox.

That’s the whole loop. Describe the work, set the cadence, and let it run in the background — which is the point. The best automation is the kind you stop thinking about.

Follow along via the RSS feed for more on skills, Swarms, and what’s shipping next.

← Back to all posts