AI agent systems.
Measurable results.

Wild Bean Cafe, BP's retail food and coffee brand operating across service station sites, evaluates capital decisions site by site — equipment upgrades, new fit-outs, refurbishments. Each one needed a financial model comparing cost, projected returns and break-even. Built by hand in spreadsheets, every model took days, and the same assumptions were re-keyed each time.

A financial modelling agent connected to site-level operational data. On request it builds scenario models, compares investment options side by side, calculates projected returns and payback, and produces a one-page summary the leadership team can decide from directly.

Models that took days now take hours. Assumptions are applied consistently across sites, manual preparation dropped by roughly three quarters, and leadership reviews investment options as a like-for-like comparison instead of a stack of spreadsheets.

Chainspin's team was manually watching hours of live streams from every streamer on their roster to find the best moments, clip them, and post across social platforms. The process consumed the equivalent of three full-time editors and still couldn't keep up with the volume of content coming in.

Pexalo built an end-to-end clipping engine that picks up every stream from Chainspin's roster, downloads the full recording, and runs custom AI detection to identify the most viral moments. The system clips, formats and posts highlights to social channels automatically. A dashboard lets the team monitor progress, review clips, and check streamer performance in real time. The scoring and distribution workflow is publicly documented.

Chainspin no longer needs three clipping staff. Every stream is processed automatically, viral moments are posted within hours instead of days, and the dashboard gives full visibility into streamer output and clip performance across channels.

Jeremy Chew founded Geo One to help Malaysian small businesses get noticed and recommended by AI platforms, but building and running the agent systems needed to deliver this at scale would have required a large technical team. The service needed marketing agents for content and visibility, plus an authority agent that could earn real third-party citations — not just on-site SEO.

Pexalo designed and built Geo One's entire AI agent workforce. Marketing agents handle content creation, social scheduling, FAQ development and website optimisation to improve how AI platforms understand each client's business. The authority agent identifies citation opportunities across external publications and pitches on behalf of clients, making third-party mentions accessible and affordable for small businesses.

Geo One now operates as a fully agent-powered GEO agency serving Malaysian SMEs. Clients get continuous AI visibility improvement, real third-party citations, and a transparent dashboard — all run by the agent systems Pexalo built. The platform launched with pricing from RM 500/month.

Coinpresso is a specialist crypto marketing agency competing for the most contested terms in its own category. Ranking for them meant publishing at a volume the team could not sustain by hand, at a standard that would survive a client reading it. The agency also needed to prove its own results before selling the service, and had no way to measure AI-search visibility, which conventional SEO tools do not report on.

Pexalo built the agent workspace Coinpresso runs on. A Crypto PR agent researches, drafts and reviews wire releases in the campaign's house style, with every citation traced back to a retrieved source. A blog agent plans five to eight posts a day across service pillars, each required to carry something a competitor could not publish. A daily report covers revenue, spend, attribution and anomalies. Alongside it, a GEO visibility tool collects real citation data across ChatGPT, Gemini and Perplexity, so the agency can audit AI presence from evidence rather than spot-checks. Nothing reaches a newswire without a person approving it. The agents were trained jointly: Coinpresso supplied the domain expertise, meaning the crypto-marketing judgement, the house style and the sense of what a client will actually read, plus the historical content and campaign data the models learned from, and reviewed output continuously. Pexalo built the systems and tuned them against that feedback.

Within seven days of the agents running at full output, Coinpresso held position 1 for "crypto seo", "cryptocurrency seo" and "seo for crypto", climbing 65, 37 and 55 places respectively. Site visibility rose to 9.44%, a gain of 7.6 points, and organic traffic more than doubled, up 137.93%. Six keywords improved for a combined visibility gain of 50.27%. Liam described it as the biggest uplift in visibility the agency had seen in over three years. Neither side could have produced this alone: Pexalo does not have a crypto marketer's instincts, and Coinpresso did not have the engineering capacity to run agents at this volume. The result came from the agency's expertise and data being built into systems that could act on it every day. Figures as measured in September 2026; search positions move, so treat them as a point in time rather than a standing claim.

Omega Pets, a Malaysian pet care brand with products across Lazada, Shopee and 61+ stockists, was growing fast but the small team was stretched thin across daily operations — from order processing and inventory tracking to customer enquiries and content creation. The repetitive admin was eating into time better spent on product development and partnerships.

Pexalo built a suite of AI agents tailored to Omega Pets' daily workflows. The agents handle tasks across the business — automating routine operations, streamlining communication, and reducing the manual work that was slowing the team down. The system was designed to grow with the company, adding new agents as new operational bottlenecks emerge.

Omega Pets now operates with significantly less manual overhead on day-to-day tasks. The team spends more time on growth — new product lines, retailer relationships and brand building — while the agents handle the operational work that used to consume their week.

The sales team was receiving high volumes of inbound leads across multiple channels with no consistent prioritisation system. High-value leads were often treated the same as low-intent enquiries, and delays in follow-up were costing closeable opportunities.

Pexalo built a lead qualification agent that scores inbound enquiries in real time based on intent signals, business profile, and product fit. The agent assigns priority, routes the lead to the right team member, and triggers an immediate notification with a pre-filled brief.

Lead response time fell by 90% and qualification accuracy reached 85%. Missed follow-ups dropped by 95%, and the sales team now spends its time on the leads most likely to convert.

The finance team was manually entering supplier invoices, matching them against purchase orders and delivery notes, and chasing approvals over email. The process was slow and error-prone, occasional duplicate or incorrect payments slipped through, late-payment fees added up, and leadership had little real-time visibility of what was owed.

Pexalo built an accounts payable agent that captures invoices from email and supplier portals, extracts line-item data, matches each invoice against the relevant purchase order and goods-received record, and flags discrepancies for review. Clean invoices are routed to the right approver automatically and synced to the accounting system, with every step logged.

Invoice processing time and manual data entry both fell sharply. Discrepancies and duplicate invoices are caught before payment, late-payment fees dropped, and the finance team redirected hours each week from data entry to reconciliation and cash-flow planning.

Hiring managers were buried in applications for every open role, screening CVs manually and losing strong candidates to slow responses. Interview scheduling meant endless back-and-forth, and onboarding new hires relied on a manual checklist that was applied inconsistently across sites.

Pexalo built a recruitment and onboarding agent that screens and scores applicants against defined role criteria, shortlists the strongest candidates, and coordinates interview scheduling across calendars. Once a candidate is hired, the agent triggers a standardised onboarding workflow covering documents, account setup, and a task checklist, with automatic reminders and escalation of any exceptions to the HR lead.

Time to shortlist dropped significantly and scheduling became hands-off. New hires move through a consistent onboarding process at every site, and the HR team recovered substantial time previously lost to admin while improving the candidate experience.