Everything,
asked plainly.

Tend installs and runs a managed AI operating layer inside a small or mid-sized business, and stays responsible for making it useful.

Not software you log into. Not a chatbot. Not a self-serve workflow builder. A done-for-you installation, with the people behind it.

Four stages, sequential. Each stage delivers something the client can see and use.

• Stage 1 - Setup (about 2 weeks). We map the business into a structured ontology - SOPs, services, people, data, decisions - and deploy a small team of named agents on top.
• Stage 2 - Day-one launch. The executive surface goes live. The founder can talk to the business through it on day one.
• Stage 3 - Three months of build. Integrations, automations, specialist subagents. The work that makes the agents actually run the business, not just describe it.
• Stage 4 - Operator handover. At month three the operator surface is handed over. From there we propose a scaled consultancy if it's a fit.

Same TendOS, two modes. Same chrome, same agents, same playbook underneath. The mode determines what the user sees and how loud the interface is.

Why two: the executive surface gives the buyer something tangible on day one. They don't have to wait three months to see value. The operator surface becomes the proof of work and the bridge to the consultancy.

The headline engagement is one number for setup, one number per month for three months, and a real conversation about what comes next.

Tend covers API and model costs through its own accounts. Agent design is cost-aware by default. No credits, no per-seat fees.

Two paths, depending on the client's appetite.

The point of the staged engagement is to land the client somewhere durable - either a deep consultancy or a quiet retainer - rather than a hard cliff at month three.

SMBs with enough operational complexity that an installed AI layer matters, and enough cash flow to invest in one.

• Revenue band: roughly $500K to $5M.
• Buyer: founders, operators, heads of ops.
• Trigger: obvious friction in communication, follow-up, scheduling, reporting, or workflow coordination.
• First geography: DFW. Existing relationship density, ability to support on-site installations, faster first sales.

The opening market stays intentionally broad. The first several engagements teach the niche better than abstract strategy will.

Three layers. The client buys the whole stack as a managed service.

• Ontology / Business Playbook. Markdown-based, inspectable, portable. The structured memory the AI runs on.
• Agents on top. A small named team of specialist agents. Each one owns part of the operation, escalates to humans, reports through the surfaces.
• Two surfaces. Executive on day one, operator at month three.

They keep the Business Playbook. They keep the deployed agents. They become responsible for managing them.

Tend is not selling hostage software. Tend is selling installation, operation, managed improvement, and the consultancy relationship that follows. The asset stays with the business.

Those are self-serve agent tools. They're strong for technical operators who want to configure and maintain the system themselves. The client does the work.

Tend is the opposite stance: done-for-you, with humans behind it. The client gets a structured ontology, a real client surface from day one, and a team that owns the operating rhythm. They don't have to learn a new tool to get value.

Enterprise platforms. Built for larger orgs with longer rollouts, more procurement friction, and budgets to match.

Tend optimizes for the SMB band where enterprise platforms are too heavy and self-serve tools are too thin. Faster to install, lighter to procure, more specific to one business.

MSPs wire tools together and provide ops support, but rarely deliver a living chief-of-staff layer plus a structured business memory plus a managed operating rhythm in one package.

Generic AI agencies are crowded and vague. Broad sales promise, thin operating model, often one-off projects.

Tend's position: done-for-you, staged, with a real client surface from day one and a structured handover at month three. The executive surface is the proof. The operator surface is the depth. The consultancy is the scale.

• Thomas - offer definition, agent setup, discovery and early client work, commercial and product judgment.
• Jony - GTM, pitch deck, outreach, pipeline. Built the operator surface that anchors the demo.
• Jakeh - production hardening, integration execution, technical shipping. Takes the demo into client-ready builds.
• Elie - integration prep, testing support, delivery support.
• David - infrastructure, technical support, systems and security.

One tenant currently powering the demo: Daizo, a sunglasses company. Five named agents (Scout, Chase, Milo, Vera, Tally), an 8-entity ontology, the operator surface in production-feel demo, and the executive surface live in dark mode.

• Operator demo - tend-workspace.vercel.app
• Executive demo - tend-executive-v1.surge.sh
• Demo repo - github.com/iterationlab-ai/tend-demo

• Exact contract wording around API ceilings and special workloads.
• Niche refinement after the first several deployments.
• Operator + executive in one app (route toggle) vs the current two-deploy demo.
• Persona naming logic for the AI surfaces.
• How quickly the executive surface dashboard expands beyond the split-view MVP.