Autron has moved to the agent

Autron has always been two things. One of them just got a voice.

The ad engine runs your advertising. It adjusts bids, budgets, keywords and negatives against the target you set, around the clock, whether or not you log in. It is relentless and it does not need you.

The agent is how you see what is happening, ask why, change your mind, and do the things the engine is not doing. It is the part that needs you, and until now it did not exist. You will find it at agent.autron.ai.

Your plan, your price and your engine are unchanged. This explains what the agent is for.

Why we changed it

The dashboards did one job well. They told you what happened.

They could not do anything about it. Every useful session ended the same way — you read a chart, worked out what it meant, and then went somewhere else to act on it. Amazon, a spreadsheet, a note to yourself for next week, or a message to us asking what the engine was doing on a product and whether it could work differently there.

Whichever door you took, the distance between noticing something and changing it was measured in days. That distance was the real cost of using Autron, and more charts were never going to close it.

The engine is now steerable, in your own words

The engine still does the relentless part. What has changed is how you direct it.

You can set or change the ACoS or growth target on any product by saying so. You can stop the engine on a product that has gone out of stock and start it again when it returns. You can ask what the engine has been doing and get a straight answer, with the actual changes it made and why.

What used to take a few days now takes a sentence.

And you can work alongside it

Not everything belongs to the engine. A launch you want to run your own way, a competitor you want to go after this month, an experiment you want to try on three ASINs — the agent can do all of that directly on your Amazon Ads account, in parallel with the engine, without touching what the engine manages.

And when you do want to overrule the engine on something it manages, you can. A bid or a placement the agent sets for you is claimed for seven days, and the engine leaves claimed entities alone rather than moving them back. Ask for it to be permanent and it stays yours until you hand it back.

So you get both: a machine that never stops optimising toward your target, and a way to pursue your own strategy at the same time.

Wherever you happen to be

You can chat with it in the app. It has its own email address, so you can write to it directly or copy it into a thread the way you would a colleague. You can mention it or DM it in Slack. Routine digests land in your inbox or a Slack channel and the conversation carries on from there. You can dictate instead of typing. And you can connect Autron to Claude or ChatGPT and ask from whichever assistant you already have open.

Same agent, same data, same permissions, wherever you reach it.

Better answers, faster, with something to do about them

The agent reads your live account rather than a report built last night, so it can answer questions no dashboard anticipated and compare any two periods you like. Ask for a chart and it draws one.

Then, when the answer suggests something worth doing, it can do it — in the same conversation, and never without your approval. Four separate jobs — find out, work out, act, remember to do it again — collapse into one.

And the things that only happen if someone remembers: hand them to a routine. It runs on a schedule, reports back, and you reply to the report.

What you get today

Included with your plan, at no extra cost:

• The agent itself, with a monthly allowance of 150 credits.
• Email and Slack — your agent's own email address, and the same agent in Slack. Both free on every plan; reading a thread costs nothing.
• Routines — saved instructions that run unattended and report by email or Slack.
• Playbooks — proven routines installed in one click.
• Skills — your margins, your seasonality, your rules, written once and applied every time.
• Guardrails — limits the agent cannot cross, whatever it is asked.
• Your choice of model, and Autron inside Claude and ChatGPT.

What is not changing

• Your dashboards stay where they are. Nothing has been switched off.
• The engine runs as it always has. Same bidding, same goals. The agent directs it; it does not replace it.
• Your plan and your price. Nothing about your billing changes because of this.

Where this is going

The direction is simple to state: you should not have to learn a tool to run your advertising well. You should be able to say what you want, have the engine do the relentless part, and have the agent handle the judgement, the questions and the exceptions.

Two things we are building toward that now:

• Walmart. The same engine and the same agent, running your Walmart Connect advertising alongside Amazon.
• Memory. An agent that carries your strategy across every conversation, instead of being told again each time.

These are in flight rather than finished, but they are where the work is going.