Automation & Auto-Remediation
Auto-remediation lets routine problems fix themselves, keeps a person in the loop for everything else, and puts every fix on the record.
Auto-remediation: fix it, check it, page only if it failed New
Two ways. A runbook says: when this happens, run these steps in order, check whether the problem cleared, and page a person only if it did not. For a simple case, attach one approved script to a monitor or a device and it runs by itself when that keeps failing.
Chain the steps
Each step is one of your approved scripts, with the device’s drive, service or certificate filled in. After the last step TechSavve checks whether the alert cleared, and pages a person only if it did not.
Four fixes ready to add
A shared library covers a full disk, a stopped service, a stuck print spooler and an expired certificate (renewed with certbot or win-acme). Each is added switched off.
Safe by design
Nothing is switched on until you say where it applies. Warning alerts never page, down devices are never touched, and a daily cap limits how many fixes may run.
On the record
Every run shows its steps on the alert, the incident and the ticket, and counts in the customer’s business review as a problem solved before they noticed.
New. Runbooks were tested end to end with simulated certificate tools and in our own demo, and have not yet run on a customer’s devices. Start with one customer and one runbook.
A script library you can schedule
Nearly 200 built-in scripts, and your own beside them: 111 PowerShell, 71 shell and 14 Ansible playbook checks that run from Linux collectors.
What is in it
Diagnostics, storage, networking, security, services, updates and reboots, Active Directory, SQL Server and other databases, web servers, containers, VMware, Hyper-V, users and groups, Microsoft 365, and VPN and identity.
Run on a schedule
Run a script daily, weekly or monthly across a customer, a site, devices with a tag, or a mix of customers, with a preview of what will be hit first.
Results can alert
A scheduled script that keeps failing raises an alert and, past your threshold, a device incident, through the same tickets and on-call as everything else.
Credentials stay in the vault
A script asks for a credential by name. The collector fills it in when the script runs and masks it in the output.
Approvals
Hold scripts for an administrator’s approval before they run. A request that is not answered expires after a day.
Every version kept
Edit a script and the earlier versions stay, with a side-by-side compare and a restore button.
Ask the NOC New
An optional assistant for your technicians, on the AI engine you choose and your own account with it: Claude, ChatGPT, Gemini, Grok, DeepSeek, Mistral AI, GitHub Models, Azure OpenAI or Amazon Bedrock, so that company bills you by use and there is no fee from us, or Ollama on a server of your own, so nothing leaves your network.
What’s wrong?
On an alert, an incident or a device, one click explains it in plain words from the facts TechSavve already holds.
Draft the ticket
A reply, a summary or a resolution note, drafted for a person to edit and send.
Ask the NOC
A chat that answers questions about your customers, devices, alerts and tickets using read-only tools, and sees only what the person asking is allowed to see.
Under your control
Off until you choose an engine and add its key. Choose the model, set a monthly limit, and switch each feature on or off. Credentials, vault entries and monitor settings are never sent, and calls are logged by size, not content.
A fallback engine
Pick a primary engine and, if you like, a second one that answers when the first is down. Send everything to the primary first, let the platform choose by what the task needs (the chat’s look-ups need an engine that can use tools), or give each feature its own engine. An engine on your own network is never quietly replaced by one on the internet unless you allow it.
Explain a situation
A sparkle button on each situation of the dashboard’s Incident correlation card asks the AI to explain it, from the platform’s own summary of that one situation.
Talk instead of type
A microphone button beside the AI boxes turns speech into text in browsers that have speech recognition, such as Chrome and Edge. The words land in the box for you to read and fix. Your browser’s speech service hears the audio, never TechSavve, and an administrator can switch the button off.
Advice only
No AI feature changes anything. They read, explain and draft; a person decides and acts, so there is no approval step to forget.
New. The assistant, and each engine you can choose, was tested against simulated providers and has not yet been run by a customer against a live account with each vendor. The Test connection button on the AI page checks yours before you rely on it.
More that runs by itself
Quieter alerts
One-click suggestions that wait, raise a threshold, mute a backup window or learn what is normal. Read more.
Checks for the rest of the stack
NAS and backup systems, UPS units, printers, Redfish servers, firewalls, Tailscale and Kubernetes. Read more.
Auto-remediation with one script
Run an approved script automatically when a monitor or a device keeps failing. Set how many failures first and how many attempts; if it is still down after the last one, a separate alert tells you. Success is judged by the problem clearing, never by the script’s exit code. Read more.
Playbooks for Windows sites
Ansible runs from Linux collectors. A site whose collector is on Windows can borrow another Linux collector of the same customer to run its playbooks, as long as that one can reach the site’s devices.
See TechSavve RMM in action
Book a free demo with the DFWTechPros team.