Execution layer
Seven agents. One person who is accountable.
Automation without judgement produces volume nobody wants. Aeologic pairs a parallel agent fleet with a named human who owns the plan and approves the output.
The fleet
What each agent is responsible for
Each agent owns one job, hands off explicitly, and writes its reasoning into the record so a human can audit any decision after the fact.
- A-01Weekly, plus on demand
Strategy Agent
Sets the plan every other agent executes
Audits your visibility across Google and the AI engines, maps the competitive gap, and turns it into a prioritised quarter of work with named owners and expected outcomes.
- Keyword and prompt research across search and AI surfaces
- Competitive gap analysis against up to twenty rivals
- Topic cluster architecture and internal link planning
- Quarterly roadmap with impact estimates per initiative
Hands off to · Research Agent · Content Agent
- A-02On every brief
Research Agent
Grounds every brief in current evidence
Reads the live SERP, the current AI answers and the primary sources behind both, then compiles the entity list, question set and citations a writer needs before typing a word.
- Live SERP and AI answer capture for every target query
- Entity, statistic and source extraction with attribution
- People-also-ask and prompt-variant harvesting
- Freshness checks that flag briefs built on stale data
Hands off to · Content Agent
- A-03Continuously
Content Agent
Writes the draft in your voice
Produces long-form articles, landing pages and FAQ clusters structured for both ranking and citation, in a brand voice trained on your existing library.
- Long-form drafting against an approved brief
- Brand voice matching from a trained profile
- Answer-first passage structuring for LLM retrieval
- Source attribution on every factual claim
Hands off to · Visibility Agent · Publishing Agent
- A-04Daily
Visibility Agent
Watches how the engines describe you
Runs your prompt set across seven AI engines on a schedule, records every answer, and raises a task the moment a citation is lost or a fact drifts.
- Scheduled prompt runs across seven engines
- Share-of-voice and sentiment trending
- Citation source extraction and clustering
- Alerting on lost citations and factual drift
Hands off to · Strategy Agent · Authority Agent
- A-05Weekly
Authority Agent
Earns the citations models trust
Identifies the third-party domains the engines already lean on in your category and runs the placement programme that gets you onto them.
- Source-graph analysis of AI citation patterns
- Placement targeting and pitch angle development
- Anchor text portfolio planning
- Monthly link integrity verification
Hands off to · Publishing Agent
- A-06Continuously
Community Agent
Finds and joins the conversations that matter
Surfaces the forum threads shaping your category, ranks them by how strongly they influence AI answers, and drafts disclosed contributions for human approval.
- Cross-platform community monitoring
- Citation influence scoring per thread
- Disclosed response drafting
- Escalation routing for support and product issues
Hands off to · Visibility Agent
- A-07On every publish
Publishing Agent
Ships the work and makes it discoverable
Handles the technical last mile — schema, metadata, internal links, sitemap pings and crawler verification — so nothing sits in a draft folder losing value.
- Schema markup generation and validation
- Metadata and canonical hygiene
- Internal link insertion across the existing library
- Sitemap submission and IndexNow pings
Hands off to · Visibility Agent
The human layer
Why a person still owns the outcome
The failure mode of agentic marketing is confident volume in the wrong direction. Every Aeologic account has someone whose job is to prevent that.
- H-01
A named strategist, not a ticket queue
Managed accounts get one senior strategist who owns the roadmap, writes the briefs the agents execute, and joins a monthly call with your team. Same person, month after month.
- H-02
Review before anything is published
No agent output reaches your live site without a human editor approving it. That is a hard rule on managed accounts and a default we strongly recommend on self-serve.
- H-03
Judgement the models do not have
Positioning calls, regulated-industry claims, competitive sensitivity, the thing your CEO said in the last earnings call — these need a person who knows your business.
Questions about the agents
Do the agents publish without a human seeing the output?
No. Human review before publication is a hard rule on managed accounts and the default on self-serve. The agents are built to make that review short — sources attached, brand voice matched, claims screened — not to eliminate it.
What does 'running in parallel' actually mean?
Once the Strategy Agent has produced a plan, the remaining agents work simultaneously against it rather than passing a single task down a chain. Authority placements are being pitched while content is being drafted and technical fixes are being shipped.
Can I turn individual agents off?
Yes. Each agent can be disabled per workspace. Teams in regulated industries commonly run without the Community Agent, and teams with an in-house PR function often disable the Authority Agent.
Which models power the agents?
A mix, selected per task and changed as better options ship. Research and drafting use frontier models; classification and scoring largely use smaller fine-tuned models. Your content is never used to train third-party models.
See the fleet run against your domain.
A strategist will run a live baseline on your site and walk you through what each agent would do first.
7-day trial · No card required · Cancel anytime