Peligent Book a call

What we run

Case studies with skin in the game.

These systems don't come from a portfolio slide - they run our own companies right now. Every engagement gets the patterns we've already paid to learn.

Internal · autonomous agents

13 agents

running the morning shift, every day

Pouria AI Portal

Claude · systemd timers · Telegram · self-hosted

The problem

A founder operating multiple companies has the same problem as any executive team: mornings disappear into briefings, status checks, content drafts, and site audits - work that is essential but repetitive.

What we built

  • A portal of 13 autonomous AI agents on schedules: morning briefings, blog topic research and drafting, SEO audits, immigration-news tracking, self-improvement reviews of the agents themselves.
  • Every consequential action routes through a human approval step on Telegram - approve, revise, or reject from a phone.
  • A dead-letter dispatcher retries failed jobs; a watchdog reports the health of the whole fleet.

Where it landed

The repetitive shift runs itself before 9am. The human does judgment work: approvals, decisions, exceptions. This is the reference architecture we implement for client organizations.

Recruitment firm · our company

$0.03

to source and rank one candidate

Talione

Claude + GPT · n8n · Postgres · Apify · Apollo

The problem

Executive recruitment lives and dies on two grinds: finding the right candidates fast, and finding the companies about to hire. Both were manual, expensive, and slow.

What we built

  • An AI sourcing module inside the firm's CRM: it harvests candidate profiles, then ranks them against the role profile with reasoning attached - at roughly three cents per ranked candidate.
  • A 26-workflow automation pipeline that watches job postings, identifies the hiring companies, finds their decision-makers, enriches contact data, and syncs everything into the CRM.
  • Automated email identity provisioning: a new salesperson gets their address, send-as, and signature created automatically.

Where it landed

Sourcing that took a researcher's day now takes minutes. The sales team opens the CRM to find decision-makers already discovered, enriched, and staged for outreach.

Our product · now our engine

18 tools

one copilot over live store data

Peligent Platform

Node · Postgres + pgvector · OpenAI + Claude · WordPress

The problem

Store owners drown in operational questions: what's selling, what's broken, which products need better copy, why did speed drop. The data exists - spread across a dozen admin screens.

What we built

  • An AI copilot that connects to a live store and answers with actions, not links: 18 tools covering semantic product search, bulk updates, report queries, image editing, content generation, site-health and speed checks.
  • Monitoring bots for retention, errors, plugin conflicts, and growth opportunities that report on schedule.
  • A full SaaS platform - auth, billing, indexing, vector search over products, orders, and customers.

Where it landed

Built as a product, battle-tested by real stores, and now the internal platform our e-commerce client work runs on. Client engagements get a mature system, not a first draft.

E-commerce · our stores

1,464

products classified by AI

Brewmart + Teslakala

WooCommerce · n8n · Claude · Metabase

The problem

Two real stores - a Toronto coffee-equipment shop and a high-traffic electronics retailer - with the classic catalog problems: thousands of products, inconsistent categories, support questions around the clock.

What we built

  • AI catalog classification that reorganized 1,464 products into a clean category tree - keyword and content-based, with human review on the ambiguous tail.
  • Storefront chatbots handling product questions and sales support, live on a store serving thousands of customers.
  • Sales analytics dashboards, currency-rate APIs, and cache-safe UI update pipelines.

Where it landed

Catalogs customers can actually browse, support that answers at 3am, and dashboards the owners check every morning. Our own money is on these systems working.

Want systems like these inside your organization?

Start with a free 30-minute discovery call. We'll tell you which of these patterns fits your operation - and which don't.

Book a discovery call See what we run

No slides. No hype. 30 minutes.