Information Technology.
28 real jobs AI can take on for a information technology team - most of them inside the licence you already pay for. The colour on each card is the kind of AI; the line at its foot is what it takes to get it.
THE COLOUR IS THE KIND OF AI — AUTONOMY RISES LEFT TO RIGHT
The line at the foot of each card is what it takes to get it - a free chatbot, the licence you already pay for, or a build.
If these words are new, the homepage walks the whole spectrum →
Service desk.
Lack of self-service options and slow response times can lead to higher costs and additional downtime for users.
Affected users get a clear, translated outage update with an FAQ fast, cutting confusion and inbound tickets during an incident.
Try it in your own AI
- Context: I'm in IT at [company]. [Service] is down, users are finding out the hard way, and every minute without a clear message becomes another wave of tickets.
Objective: From the incident details below, draft the user-facing update: what's affected, what isn't, what to do in the meantime and when we'll say more. Add a short FAQ for the questions users will really ask. [Paste the incident details.]
Style: Update under 150 words; FAQ of five questions at most.
Audience: Everyone - assume no technical knowledge.
Tone: Calm and straight - no blame, no hedging, no 'we apologise for any inconvenience' boilerplate.
✓ In the licence you already pay for
Staff resolve common connectivity and access issues themselves and track their own tickets, freeing the service desk and lifting satisfaction.
Try it in your own AI
- Context: I can't [the problem, e.g. connect to the wi-fi at the [location] office] and I need it working for [what you're trying to do]. I've already tried [what you've tried].
Objective: Walk me through the likely causes one at a time, starting with the most common, and tell me what to check at each step before we move to the next.
Style: One step at a time - wait for my result before the next.
Tone: Plain English - assume I know my way around a laptop but not a network.
✓ In your licence - assembled once, then it runs. We set these up →
Correlated incidents across tools point to the most probable cause with recommended fixes, shrinking mean time to resolution.
✓ In your licence - assembled once, then it runs. We set these up →
Tickets are created, prioritised and routed to the right team automatically, with status updates issued at every milestone.
✓ In your licence - assembled once, then it runs. We set these up →
A custom assistant answers policy and access questions, checks ticket status, and hands off cleanly to a person when needed.
Try it in your own AI
- Context: I'm [role] at [company]. I requested [access/equipment] on [date] and I'm blocked on [the work waiting on it].
Objective: Tell me where the request sits, who it's waiting on and what I can do to move it - and if you can't see that, tell me who can rather than guessing.
Style: Three lines - status, blocker, next step.
✓ In your licence - assembled once, then it runs. We set these up →
Change management.
Many organizations face underutilization of applications and missed ROI because users don’t know about or aren’t trained to use them.
New tools launch with an approved adoption plan, compliance checklist, FAQs and training, so people reach productivity sooner.
Try it in your own AI
- Context: We're rolling out [tool] at [company] and the risk isn't the technology, it's adoption - tools land, nobody changes how they work, the licence money evaporates. I own the rollout.
Objective: Draft the adoption plan: phases, the compliance checks needed before launch, the training and FAQs for first users, and the proposal note that gets leadership to sign it off. Work from the material below. [Paste tool details and any constraints.]
Style: Plan as phased steps with owners; proposal note under a page; FAQs as Q&A.
Audience: The leadership note is for sign-off - lead with the outcome, not the features.
✓ In the licence you already pay for
Upcoming features, tips and service news reach users on a regular cadence, with feedback captured to track satisfaction.
Try it in your own AI
- Context: I run internal comms for IT at [company]. Features ship, tips exist, nobody hears about any of it - then we wonder why usage is flat.
Objective: From the release notes below, draft this month's user update: the three changes that matter to normal users, one tip worth stealing, and a way for people to tell us what's annoying them. [Paste the release notes.]
Style: Intranet post, under 300 words, scannable.
Tone: Human - a colleague sharing something useful, not a vendor newsletter.
✓ In the licence you already pay for
Requirements, risks, owners and executive updates are pulled together from day one, keeping delivery on track through to handover.
Try it in your own AI
- Context: I'm kicking off [project] at [company] - [one line on what it is and why now]. The scattered starting material is below; the usual failure is that requirements, risks and owners never get pinned down in week one.
Objective: Pull the material into a project one-pager: purpose, goals, deliverables, the requirements we've agreed versus assumed, the risks with owners, and the open questions. [Paste the emails, notes and documents.]
Style: One page, headed sections, risks as a table.
Audience: The exec sponsor reads the top half; the delivery team works from the rest.
✓ In the licence you already pay for
Stakeholder inputs become consistent, well-structured business and product requirement documents that keep everyone aligned.
Try it in your own AI
- Context: I'm gathering requirements for [system/product] at [company]. The stakeholder input below is the usual mix - wishes, constraints and solutions dressed up as needs.
Objective: Sort it into a requirements document: real requirements separated from proposed solutions, must/should/could applied, and conflicts flagged for a decision rather than smoothed over. [Paste the stakeholder input.]
Style: Our standard headings; every requirement testable ('the user can...', not 'the system is intuitive').
✓ In the licence you already pay for
Staff answer a few questions and get matched to approved hardware, with approval, ordering and delivery tracking in one flow.
✓ In your licence - assembled once, then it runs. We set these up →
Device and app management.
Manual processes for device and app management can lead to inefficiencies and downtime for end users, which means overall decreased productivity.
Common IT tasks - research, documentation, training, inventory, backups and code fixes - move faster with ready-to-use prompts.
Try it in your own AI
- Context: I'm in IT at [company] and I need to document [system] properly - the only person who understands it end to end is [person], and that's the problem.
Objective: Interview me about [system] - one question at a time on architecture, dependencies, failure modes and recovery - then turn my answers into a two-page document a competent newcomer could work from.
Style: Non-technical language where possible, the process as numbered steps, key facts as a table.
Audience: The next person who has to fix it at 2am.
Works in a free chatbot
Device details, error codes and a healthy-device comparison reveal the missing policy, so install failures are remediated quickly.
Try it in your own AI
- Context: I manage devices at [company]. [App] is failing to install on [device or group] with error [code], users are blocked and the error text is telling me nothing.
Objective: Work through the likely causes of this error on [platform/management tool], tell me what to compare between a failing device and a healthy one, and give me the order you'd check things in. [Paste the device details and error code.]
Style: A checklist, most likely cause first, with what 'found it' looks like at each step.
✓ In the licence you already pay for
Market research becomes an aligned technology strategy with OKRs and a delivery plan, all communicated clearly to stakeholders.
Try it in your own AI
- Context: I'm [role] at [company] and our technology strategy for [area] is due a refresh - the current one predates [what changed]. Leadership will fund a plan, not a trend deck.
Objective: From the research and current strategy below, draft the refresh: where we are, what's changed in the market, the three moves we should make and the OKRs that would prove they worked. [Paste the research and the current strategy.]
Style: Strategy under four pages, OKRs as a table.
Audience: The exec team - business outcomes first, technology second.
✓ In the licence you already pay for
Opportunities, customer feedback and technical input combine into a persuasive product strategy and matching OKRs the team can rally behind.
Try it in your own AI
- Context: I'm a product manager at [company] in [area]. I've got customer feedback, market research and technical input that all say something slightly different, and I need one strategy the team can rally behind.
Objective: Synthesise the material below into a product strategy: the opportunity, who it's for, what we build first and why, what we're explicitly not doing, and OKRs to match. Flag where my inputs contradict each other rather than smoothing it over. [Paste the feedback, research and technical notes.]
Style: Narrative strategy in paragraphs, one page; OKRs as a table.
Tone: Persuasive but honest - it has to survive the team's scrutiny, not just excite them.
✓ In the licence you already pay for
Requirements, vendor research, build-versus-buy analysis and an RFP come together to support a confident, well-documented purchase.
Try it in your own AI
- Context: We need [capability] at [company]. I have to recommend build versus buy and then run a fair process - and the decision will be audited later, so the working matters as much as the answer.
Objective: From the requirements below, build the build-versus-buy analysis with the key differences as a table, then draft the RFP we'd send if we buy. [Paste requirements and any vendor material.]
Style: Analysis on one page, table included; RFP with our requirements as numbered questions vendors must answer directly.
✓ In the licence you already pay for
Endpoints and cloud connections are scanned continuously, with fixes flagged for performance, security, compliance and cost.
✓ In your licence - assembled once, then it runs. We set these up →
Tests run automatically against generated data, surfacing failure patterns and actionable fixes that speed up debugging.
✓ In your licence - assembled once, then it runs. We set these up →
Licence requests are matched to cost-effective options and idle seats are reclaimed, trimming software spend without disruption.
✓ In your licence - assembled once, then it runs. We set these up →
Critical device alerts are assessed, root causes traced to recent changes, and clear fixes sent to affected users at speed.
Try it in your own AI
- Context: I'm on the device team at [company]. Alerts have spiked in the last 24 hours and I need to separate one root cause from twenty coincidences before users start phoning.
Objective: From the alert data below, group the alerts by likely cause, connect them to any recent changes in settings or updates, and draft the note to affected users with the fix or workaround. [Paste the alert list and the recent-changes log.]
Style: Groups as a table - alert, count, likely cause, suggested fix. User note under 100 words.
Tone: The user note calm and practical.
✓ In your licence - assembled once, then it runs. We set these up →
Data pipelines, queries and reports come together quickly so usage trends and impact are visible to anyone who needs them.
Try it in your own AI
- Context: I'm building usage reporting for [product/tool] at [company] so decisions stop being anecdotes. The raw data below has the usual problems - inconsistent categories, relative dates that break when reused.
Objective: Help me shape it: clean the categories as specified, build the queries for [the questions you need answered, e.g. active users by team, month on month], and tell me which trend in the result deserves attention and why. [Paste the data or query.]
Style: Working shown step by step, results as tables, the 'so what' in two sentences at the end.
Built for your business - the work you'd hire us for →
Day in the life.
See how people can use your AI assistant to perform common tasks throughout their day to save time, generate value, and improve their wellbeing.
A launch infrastructure manager uses AI through the day to stay on top of priorities and run a smoother launch.
Try it in your own AI
- Context: I manage launch infrastructure at [company]; launch is [date]. Today's job is knowing what's off-track before someone asks me.
Objective: From the checklists and threads below, give me green/amber/red by workstream with the evidence, what turned amber since last week, and the chase message for each amber, drafted. [Paste the checklists and threads.]
Style: Status table, then the chasers ready to send.
Tone: Chasers direct and blame-free - the deadline is the villain.
✓ In the licence you already pay for
An IT administrator uses AI across the day to handle alerts, tickets and admin tasks with less effort.
Try it in your own AI
- Context: I'm an IT administrator at [company]. The overnight queue below is alerts, tickets and requests; some of it matters.
Objective: Triage it: what's urgent and why, what's routine, what can be batch-answered - then draft the batch replies and the escalation note for anything above my access. [Paste the queue.]
Style: Triage table, drafts after.
✓ In the licence you already pay for
An IT product manager uses AI throughout the day to support users and keep work moving.
Try it in your own AI
- Context: I own [internal product/tool] at [company]. Users write in daily; the feedback below is this week's, and it needs to become decisions rather than a folder.
Objective: Theme the feedback, split 'broken' from 'wished for', tell me which theme costs users the most time, and draft the update note telling users what we're doing - and honestly what we're not. [Paste the feedback.]
Style: Themes ranked with counts; the note under 200 words.
Tone: The note candid - users trust roadmaps that say no.
✓ In the licence you already pay for
Security and compliance.
Manual processes can lead to an increase in MTTR and additional risk of a breach.
The multi-cloud attack surface, affected assets and key CVEs are surfaced and summarised, so remediation is focused and well prioritised.
Assumes your security tooling is connected to your AI.
Try it in your own AI
- Context: I'm in security at [company] running estates on [cloud providers]. Leadership asks 'are we exposed?' and I need an answer grounded in our own assets, not the headlines.
Objective: From the asset and vulnerability data below, map which CVEs touch which assets by provider, rank by real exposure - internet-facing and unpatched first - and give me the executive summary of where the risk concentrates. [Paste or connect the asset and CVE data.]
Style: Technical detail as tables; the executive summary five sentences with every acronym explained.
Audience: The summary is for people who will never read the tables.
✓ In the licence you already pay for
A reported CVE is checked against your estate to find exposed assets, label them and recommend mitigations for fast response.
Assumes your security tooling is connected to your AI.
Try it in your own AI
- Context: [CVE-ID] has been flagged by [the news / our threat feed] and I need to know within the hour whether [company] is exposed - we run [technology] in places.
Objective: Explain the vulnerability plainly, tell me what to check to find exposed assets in our estate, and rank the mitigations by how fast they cut risk. [Paste the CVE details and what you know of the estate.]
Style: What-it-is in three sentences, then checks and mitigations as ordered lists.
Audience: I'll forward the top of this to a non-technical exec - make the first paragraph carry the whole story.
✓ In the licence you already pay for
A suspect script is reverse-engineered, its intent and indicators checked against threat intelligence, and a response plan produced.
Try it in your own AI
- Context: This script turned up [where, e.g. in a scheduled task nobody recognises] at [company] and I need to know what it does before I decide how loudly to raise the alarm.
Objective: Walk through the script step by step, infer the intent, flag anything destructive or data-stealing, note any IPs or hostnames worth checking against threat intelligence, and recommend what we change to defend against it. [Paste the script.]
Style: Step-by-step first, then the verdict - malicious, suspicious or benign, with your confidence - then recommendations.
Audience: The verdict paragraph goes to people who won't read the analysis.
✓ In the licence you already pay for
An incident is summarised, response steps surfaced, impacted devices identified and an executive-ready report produced quickly.
Assumes your security tooling is connected to your AI.
Try it in your own AI
- Context: Incident [ID] is open at [company] - [one line: what fired and when]. I'm building the picture before the response call.
Objective: Summarise what happened from the data below, identify affected users and devices and whether they're patched and compliant, set out the response steps in order, and draft the executive summary. [Paste the incident data.]
Style: Timeline, then affected assets as a table, then response steps numbered. Executive summary five sentences.
Audience: The summary is for non-technical leadership - impact and action, not mechanism.
✓ In the licence you already pay for
Upgrade requests are understood, stakeholders notified, the upgrade script drafted and a change request raised for approval.
Assumes your security tooling is connected to your AI.
Try it in your own AI
- Context: I'm a network engineer at [company]. [The upgrade, e.g. new code on the core switches] has been requested and the change process needs the script, the comms and the approval trail before anyone touches anything.
Objective: Draft the upgrade script for [device/platform] with rollback steps included, the stakeholder notification, and the change request with risk and back-out plan filled in. [Paste the request details and device specifics.]
Style: Script commented; notification under 100 words; change request in our standard fields.
Tone: The notification neutral and specific - what, when, impact, contact.
✓ In the licence you already pay for
Most of these run on the licence you already pay for.
The gap is that nobody's pointed it at the job yet. That's the work we do, and it starts with a half-hour call - bring the one that looked most like your week and we'll tell you whether it's a quick win or a proper build. No licence yet? We'll help you pick one first.
← All twelve teams