Trellis Workit's trial-less
We use analytics cookies to count visits and see how pages are used, including session replays with form fields masked, and we remember your language. Analytics load only after you accept. The booking calendar on the contact page loads with that page. No ads, and no data sold.

Growth needs
Trellis

We build internal systems with AI for mid-market enterprises.

On the call, in Arabic or English, you show us how the work moves today. We tell you which steps an agent could take over.

We build with
  • Claude Code
  • React
  • AWS
  • Gemini
  • Python
  • LangChain
  • Azure
  • Llama
  • Next.js
  • LlamaIndex
  • Google Cloud
  • Flutter
  • Microsoft Copilot

Partnerships

Certified Services Partner, Claude Partner NetworkOpenAI Select PartnerMember of NVIDIA InceptionClickUp Registered Partner 2026

Trellis Work is Certified in the Claude Partner Network, an OpenAI Select Partner, a ClickUp Registered Partner and a member of NVIDIA Inception.

AI Opportunity Mapping

See how it works.

Video transcript

Where can AI help your company? Your own staff can show you.

AI Opportunity Mapping runs one department at a time. Two weeks, from the first session to a written plan.

On screen: the departments Finance, Procurement, HR, Operations and Sales, and a two-week timeline from the first session in week 1 to the written plan in week 2.

We set up a workstation with the leading AI models, including private models that can run inside your company.

On screen: a workstation with tabs for Claude, ChatGPT, Gemini, Qwen and DeepSeek. Qwen and DeepSeek are marked "Can run privately inside your company". The screen reads "Try your own work on sample data".

It uses sample data, so no real company data is needed.

On screen: the sample files are supplier emails, purchase orders, invoices, an HR policy and job descriptions, each marked "Sample". Beside them: "Sample data only. No real company data needed."

Your staff try their own work on it. Junior staff find everyday tasks. Managers find product ideas.

On screen: someone types "Summarise these supplier emails". Junior staff, everyday tasks: summarise supplier emails by order; reconcile invoices to the ledger; answer HR policy questions; draft an RFQ from a purchase request. Managers, product ideas: spend view by project; supplier scorecards; delivery tracker for every site.

Then we sort every idea with you. Build now, or build later.

On screen, build now: summarise supplier emails by order; reconcile invoices to the ledger; answer HR policy questions; spend view by project. Build later: draft an RFQ from a purchase request; supplier scorecards; delivery tracker for every site.

You get a written plan, with use cases ranked for each department.

On screen: the AI opportunity plan, with use cases ranked by department. Procurement: summarise supplier emails by order (now); draft an RFQ from a purchase request (later); supplier scorecards (later). Finance: reconcile invoices to the ledger (now); spend view by project (now). HR: answer HR policy questions (now); screen CVs against the job description (later). Operations: check delivery notes against orders (now); delivery tracker for every site (later).

From there, every stage is a loop of strategy, build and test. Proof of concept, pilot, then production rollout. Every company's path is different.

On screen: four loops of strategy, build and test that climb a rising curve, one each for AI Opportunity Mapping, proof of concept, pilot and production rollout. In the example, mapping runs once, the proof of concept twice and the pilot three times.

Commercial terms are agreed at the pilot. Everything built before the pilot is yours to keep, with no lock-in.

Growth requires structure. Trellis Work is the structure. It's trial-less.

On screen: the rising path settles onto a lattice that forms the Trellis Work mark, then the Trellis Work logo with the tagline It's Trial-less.

We build internal systems with AI for mid-market enterprises.

On screen, with music and no narration: Partners. OpenAI Select Partner. NVIDIA Inception Program. Certified Services Partner, Claude Partner Network. IBM Business Partner.

Interested? Book a call, or visit us directly in the office.

On screen: aitrellis.com, the partner badges along the top, and a map of Riyadh with the office marked: Trellis Work, Abi Bakr As Siddiq Rd, Exit 6, Al Taawun, Riyadh.

Footage: BaldasaridStock, Little Blue Planet and Freepik, on Freepik. Map: © OpenStreetMap contributors.

A proof of concept runs on your systems before you agree terms for the build.

  1. Yours to keep, even if you stopAI Opportunity MappingTwo weeks
  2. Yours to keep, even if you stopProof of conceptOn your systems
  3. PilotLive, with real usersTerms for the build agreed here
  4. Production rolloutCompany-wide, run by us

Much of your team's week goes to work you did not hire them for.

Re-typing order emails into purchase orders. Chasing signatures. Copying the sign-in sheet into payroll. Answering the same question twice before lunch.

  • What AI models can do
  • What companies use
  • The gap

Three ways we take work off your team.

Smarter Employee

Your people keep the judgement. The agent takes the rest.

Faster Processes

When an approver or a supplier goes quiet, the agent chases.

Transformational Products

We build the product and keep shipping it. A software factory on a monthly retainer.

Case studies

Saudi company, field teams on remote sites

Paper allowance claims, replaced by a portal

Field technicians claimed overtime and food allowance on paper. We built a portal with GPS-verified submissions, an offline path for remote sites, automatic monthly totals and a full audit trail.

In daily use since 15 February 2026

US medical device company, manufacturer in Germany

Orders go from the inbox to a purchase order without anyone re-typing them

Every order arrived by email and was keyed into a purchase order by hand. An agent now reads each order email and produces the purchase order.

Saudi business setup firm

Two internal systems and a Claude rollout

We built the admin dashboard their team uses to follow up with clients, and an employee tracking system. Then we rolled Claude out on role-based tracks, finance included.

We stay after go-live, on a retainer you can cancel in any month.

One improvement a month

Each one goes into production, and the retainer is flat.

Inside the systems you already run

Your staff keep the systems they use today, such as Oracle Fusion or Outlook. Nothing is migrated.

Why not McKinsey, or one good hire?

Trellis Work

  • What you get firstA written plan in two weeks
  • Who builds itWe do, from mapping to go-live
  • Before you commit to the buildA proof of concept on your systems, yours to keep
  • After go-liveWe operate it and ship one improvement a month

Large consultancy

McKinsey, Accenture
  • What you get firstA strategy document
  • Who builds itA separate delivery team or a vendor
  • Before you commit to the buildA proposal for the next phase
  • After go-liveA handover, or a managed-services contract

Hiring in-house

  • What you get firstA job post
  • Who builds itOne hire, once you find them
  • Before you commit to the buildA salary you are already paying
  • After go-liveOne person carries it

What we build, by industry

Tech and AI companies

Ship faster without growing the team.

  • Enterprise Claude rollout and activation
  • Claude Code enablement for engineering
  • AI FinOps and spend control

Contracting and manufacturing

The margin lives in purchasing and on site.

  • AI Procure-to-Pay Agent
  • AI Workforce Controller
  • RFP Agent for tenders

Consultancies

More proposals out, same headcount.

  • RFP Agent for tenders and proposals
  • AI Chief of Staff for partners
  • AI Opportunity Mapping

Private equity

Numbers you can defend in the room.

  • AI Asset Valuation Engine (Coming soon)
  • Document work for due diligence
  • AI Opportunity Mapping across the portfolio

Healthcare

Governance designed in.

  • Appointment and admin workflows
  • AI Workforce Controller for staff attendance
  • Approval gates in front of anything irreversible

Enterprise Claude, from rollout to spend control.

We roll Claude out across the company: seats, workspace governance, Claude Code for the engineering teams, and training that turns licences into daily use. Then we track spend per team and set budget alerts.

What clients ask before mapping starts.

How long does AI Opportunity Mapping take?

Two weeks, from the first session to the written plan. Terms depend on what we go on to build together, and we agree them on the call.

Does our real data leave the company during mapping?

No. Mapping uses sample data, so no real company data is needed.

Do we need our own AI or engineering team?

No. Your IT team grants access and your staff show us how the work moves. We do the rest.

What do we keep if we stop after the proof of concept?

Everything built up to that point, with no lock-in.

What happens after go-live?

We operate it and ship one improvement a month, on a flat retainer you can cancel in any month.

Do you work in Arabic?

Yes. We work in Arabic and English, from Riyadh.

Which models do you use?

Whichever fits the work, including Claude, OpenAI's models, Gemini and Llama.

Ask an AI for a summary of Trellis Work

The company profile

Who we are, and what we have built.

Leave your details and download it.

We use your details to follow up once. No newsletters.

Growth needs structure

it's trial-less