Why are all of these in-house?
We don't yet have published client case studies, and we won't invent them. Instead, here's what our agents actually do on businesses we own and operate. Every story below is real. Where a scenario elsewhere on our site is hypothetical, it's labelled as an example.
Use case 1: how does a 14-bot SEO team run our taxi website?
Fourteen bots (team lead, rank tracking, Search Console, GA4, on-page and schema, internal linking, backlinks, local SEO/GBP, AEO/GEO, technical indexing, page uniqueness, Google Ads, Facebook Ads and GMB) work from a shared task board on a schedule. They edit WordPress only after a backup, write every change to a change log, send a daily Telegram report and keep a live dashboard updated. Bulk changes, new content and anything that affects spend need approval; deletions, redirects and site settings are hard stops. Read more about the SEO bot team.
What we learned: narrow bots with a shared board are far easier to supervise than one general agent; page uniqueness on templated route pages needs ongoing attention; and a daily report is what keeps a person genuinely in charge.Use case 2: how does an ads agent work without spending on its own?
The ads agent audits Google Ads and Meta Ads accounts (structure, search terms, targeting, conversion tracking, landing-page fit), drafts ad assets and sends reports with recommended actions. Every change that affects spend needs the owner's approval, and billing and account access are hard stops. Read more about the ads agent.
What we learned: checking conversion tracking comes first, because without reliable tracking every other recommendation is guesswork.Use case 3: how did the ops agent recover a production 503 outage on GKE?
The production site started returning 503 errors. The ops agent gathered diagnostics and found the cause: the MySQL data disk was full, filled largely by accumulated binary logs, so the database couldn't write and WordPress couldn't serve pages. The recovery:
- Purged old MySQL binary logs to free space immediately and bring the database back.
- Set a binlog expiry so old logs are removed automatically and the same failure doesn't recur.
- Grew the persistent disk to give the database comfortable headroom.
- Confirmed the site was serving normally, and reported what happened and why.
Use case 4: how did research bots build an 829-course pricing workbook?
Parallel research bots, each assigned a slice of providers and categories, gathered public course information into an Excel workbook covering 829 courses, with a source audit of about 6,300 URLs so every figure could be traced to evidence. Rows with weak or missing sources were flagged for human verification before anything was used, and Outlook was used for drafting correspondence, with sending kept with a person. Read more about the research bots.
What we learned: parallel slices make wide research fast, but the audit is what makes it usable.Use case 5: what is our CRM + LMS + labs platform?
We built a custom platform on Supabase (Postgres, authentication) and Netlify (hosting, serverless functions): a CRM for leads, enquiries and enrolments; an LMS for courses, learners and progress; and labs for hands-on learning environments. Access is role-based, and schema changes go through migrations. All three are live: crm.edupromentor.com, lms.edupromentor.com and labs.edupromentor.com. Training operations service.
What we learned: custom builds pay off when the process genuinely doesn't fit standard tools, and are overkill when it does.Use case 6: which WhatsApp, social media and brand visuals agents do we run?
- A WhatsApp Business manager agent handles routine conversations from approved information and routes the rest to a person.
- A social media agent plans and schedules Facebook and Instagram posts.
- A brand visuals agent creates on-brand images for posts and pages.
Posts with claims about prices, offers or results need approval. See social and content and WhatsApp automation.
What isn't on this page?
Client results, testimonials, logos and performance statistics. When we have client stories to share with permission, they'll be clearly marked as such. Voice agents are available to build, but we don't run one live, so there's no voice use case here.
How do these use cases apply to your business?
The same architecture (a team lead, specialists, a task board, stop rules, backups and reports) applies to finance, HR, sales, support, documents and operations. See how we build agent teams and our services, then tell us about one process.