FinesselyWorkFinessely Engine

Finessely Engine.

Case study · In house

Ten agents run our sales and marketing. A person approves every word before it goes out.

10Agents across marketing and sales
2Models: Claude writes and analyses, Grok drafts social replies
0Posts, replies or emails sent without a person approving them

The problem

A small team, a big pipeline to feed.

Selling AI services means being present where buyers talk, writing useful content every week and following up with every lead. Done by hand, that is a full time job before any client work happens.

We wanted a system that does the research and the first drafts, keeps us inside every platform's rules, and leaves the judgment to a person.

The system

Ten agents, one approval queue.

Marketing

AgentWhat it does
Site AuditorChecks every page for SEO and quality issues and writes a ranked fix plan.
SEO OptimizerProposes better titles, descriptions and alt text for approval.
Post WriterDrafts LinkedIn posts and X posts and threads on a topic.
Content RepurposerTurns one article into posts, threads, a newsletter and video scripts.

Sales

AgentWhat it does
ListenerFinds posts on X and Reddit that match our keywords.
Intent ScorerScores each post for buying intent, with a reason.
Reply DrafterWrites helpful replies for high intent posts.
Lead ProspectorBuilds lead lists that match our ideal customer.
Outreach WriterDrafts a personal first email for each new lead.
Follow-up AgentSends timed follow ups and stops when someone replies.

Simulated run

Agents draft. People decide.

engine.log

    Approval queue0drafts waiting for a person

    What we learned

    Rules we now build into every client system.

    • Approval is a feature. A single review queue is faster than fixing one bad post in public.
    • Play by each platform's rules. Low daily limits, clear disclosure and no automated LinkedIn comments.
    • Help first. Replies that answer the question get conversations. Pitches get ignored.
    • Stop when they reply. Follow ups end the moment a person answers.

    Read the launch post

    Own your AI.

    Build, customize and deploy AI automations and software that you fully own.