Named in the answer.

People and agents on one shared context

DRIS works in the Marketing room beside MIRA, NIA and the people you add, on one attributed thread. He grounds on the same knowledge hub every agent in the workspace reads, and a content gap he finds goes to the content team as a task on their board, within a daily limit.

He earns room to act on his record

Autonomy is earned on his record, one level at a time. Thirty days at a level, with at least 25 decided approvals and few dismissed, earn a nomination for the next, and an admin approves or declines it. Publishing to a live page, changing index or canonical rules and changing how the queue is weighted stay with a person however far he rises.

Every action, approval and spend in one place

Each of his actions carries its own approval bar, and work above it is held until a person decides. His Autonomy surface shows what he may do alone, per kind of action, with its track record. You set his daily spend cap, and answer engine reads pause at it. Every fix served on your site records who approved it and when.

He works alone inside limits, a person holds the approvals

He runs site audits, answer engine reads and competitor snapshots on his own inside the spend cap, and ranks what he finds as proposals. He grounds on your organisation’s knowledge. When a fix’s measurement window closes he measures what it moved, keeps the lesson in the workspace memory when the evidence is strong enough, and ranks that kind of fix by its record.

An objective, not a dashboard

You declare one number, a target and a date. He captures the baseline first, refuses a target that does not beat it, and measures every proposal against it.

Measurement that will not guess

An engine that could not be reached is never recorded as not naming you. A pillar with no data shows as not measured rather than scored zero. On priority targets, repeated samples tell a real change from one answer’s noise before he raises an alert.

Fixes that go live, and come off

Approved title, description, heading and structured data fixes go live on your site through the site connection, which is in Beta, and Withdraw takes one off within about five minutes. Index rule and canonical changes are never served that way.

Skills

• Citation-worthiness auditScores how citable a page is to an answer engine and lists the fixes.
• Answer-block authoringWrites short answer blocks for a target’s accepted questions, checks each and opens it as a proposal.
• Schema engineeringGenerates structured data from a page’s content, validates it against what the page shows, then proposes it.
• Machine-readability layerRecommends which AI crawlers to allow, proposes llms.txt lines, checks pages render without scripts and reports bot activity.
• Knowledge-consistency enforcementCompares the facts your pages state with your brand facts and flags where they differ or are missing.
• Third-party presence auditRanks the outside domains the engines cite most and checks what each says about you.
• Competitor gap analysisFinds the queries where a competitor is cited and you are not, scores each and routes it to its owner.
• Intent classificationTags each query by intent, and by whether an answer engine or a click is the likelier result.
• AI answer share of voiceHow often each engine mentions and cites you against each tracked competitor, per engine and blended.
• Citation source tracingMaps the sources the engines cite for each question, with your own share of them.

Tools

• Read the current visibility positionThe visibility grade, the score per pillar and the open work, from the room or his surfaces.
• Route a job to the sub-agent who owns itHands a page audit, a page score, a keyword lookup or a competitor scan to the agent whose tool covers it.
• Discover targetsSuggests the pages and topics to track, from your sitemap, your top traffic or a competitor.
• Generate promptsBuilds the buyer questions a target is checked against, for a person to accept.
• Start a full auditAudits the whole site in the background; Run audit checks a single page.
• Discover keywordsFinds new keywords from a seed phrase, your own site or a saved competitor; Upload a CSV adds your own.
• ScanReads what a competitor covers and can open proposals for what you do not.
• Approve and RejectDecides each proposal in the Queue, with a short window to undo.
• WithdrawTakes an approved fix back off your site.
• Hand to content teamSends a content gap to the content team as a task on their board.
• Generate this weekWrites the weekly digest from what was measured, saying so when nothing moved.

Signals

• Search performanceClicks, impressions, click rate and position by page and query, for the baseline, rank tracking and each proposal’s diagnosis.
• Site analyticsSessions and conversions by source, including visits referred by AI assistants, for attribution.
• Answer engine answersWhat each engine says to a buyer’s question, which sources it cites and whether it names you, for the six dimensions.
• AI answers in search resultsThe AI summaries search pages show, kept apart from the chat engines.
• Site crawlPage structure, metadata, robots rules for AI bots and llms.txt, for the site audit and agent readiness.
• Page performanceLoad and interaction timings, for performance regressions.
• AI crawler visitsWhich AI bots read which pages, for whether the engines can see the site at all.
• Keyword dataVolume, difficulty and intent, for ranking the keywords worth winning.
• BacklinksWho links to you and to competitors, for the off page score.
• Competitor contentWhat competitors publish, for the gaps you can close.
• Your knowledge hubThe documents your organisation uploads, so his answers and fixes stay grounded in your facts.

Integrations

• Google Search ConsoleConnected from the workspace, so he reads your search performance.
• Google Analytics 4Connected from the workspace, so he reads traffic and conversions by source.
• Site connection (Beta)Your developers connect your site once with a key; approved fixes then go live on it, and withdrawn ones come off.

A property is connected, targets are discovered from the sitemap, and the first run returns the quadrant plot, the six dimension scores per page, the citation graph naming who the engines reached for instead, and a ranked list of the gaps with the fix that closes each one.

Read how teams use Nagent’s AI teammates in their own growth work.