Built by Berry — Operational AI
Menu

Navigation

Built by Berry is the operational AI firm — we ship the systems, the agents, and the training to run them.

Start a Project
Minimal systems-inspired composition

The Stack / Ashlar

Operator-built workflows, not engineering tickets.

The product layer for ops, finance, support, and compliance teams who already know what the workflow should be — and are tired of waiting for a backlog to clear before agentic AI can run it.

Built on the same stack we deliver to clients: laravel-swarm and Swarm Cloud underneath, operators on top.

Join the waitlist

One email when early access opens. No drip sequence. No SDR.

Status

Workflow product

Stage — Pre-launch · Waitlist open

For — Ops, finance, support, compliance

Built on — laravel-swarm + Swarm Cloud

Replaces — Bespoke internal builds for routine work

Trust — Audit + redaction from the framework up

The problem

The workflow is obvious. Getting it built is not.

Most companies do not have an AI problem. They have a delivery problem. The operator who knows exactly what should happen cannot ship it without an engineer, and the engineer cannot ship it without a quarter.

01

Ops blocked on engineering

The person closest to the work cannot change the work. Every adjustment is a ticket. Every ticket waits.

02

Internal builds that rot

Six months later the workflow has drifted, the engineer has moved teams, and the script lives on someone's laptop. Nobody owns it.

03

AI demos that do not get adopted

Polished demos. No operator handle. No supervision surface. No audit trail. The team that has to use it does not, and the leader who bought it cannot tell.

What an operator does

Four moves, no orchestration code.

Ashlar takes the four things operators actually do — design the work, plug it into the business, watch it run, step in when it matters — and makes each one a first-class surface.

01 · Design

Lay out the workflow

Drop in the steps the work actually has. Decide who runs each one — an agent, a person, a service. Set the rules for handoff, for review, for escalation. Edit it later without a code change.

02 · Plug in

Connect the systems already in play

CRM. Ticketing. Inbox. Storage. Internal API. Ashlar wraps each one as a typed tool the workflow can call, with the same auth and audit boundaries your team already trusts.

03 · Supervise

Watch runs live

Every run, every step, every tool call. Filter by status, by owner, by run id. See where time is going. See where humans are waiting. See where the workflow keeps catching the same exception.

04 · Intervene

Pause, approve, redirect, override

When the work needs a human, it asks for one — and the human responds inside Ashlar, not through a Slack message that disappears. Every intervention is recorded.

A run in motion

Ops can read this. Engineering does not have to translate it.

Stylized for clarity. The real canvas shows live steps, owners, and time-in-state. The point is what an operator sees, not the colors.

Run · refund-review · rid_7f1c

Waiting on approval · 14m

Classify request — agent · TriageBot

2s · ok

Pull order history — tool · CRM

0.6s · ok

Check policy — agent · PolicyAgent

3s · ok

Approve refund > $500 — human · finance-team

waiting · 14m

Issue refund — tool · billing

pending

Reply to customer — agent · ResponseAgent

pending

Built on the stack

Three altitudes. One system.

Ashlar is the operator surface for the same engine that runs every agentic system we ship for clients. You are not buying a product that wraps a black box. You are buying a product whose entire foundation is open source and auditable.

Who it is for

Fit

  • Operations leaders with workflows they could draw on a whiteboard today.
  • Companies running the same multi-step process hundreds of times a week.
  • Teams where compliance, audit, or finance has a veto on AI work.
  • Leaders who want operator ownership of agentic work, not engineering ownership.

Not fit

Not fit

  • Teams looking for a chatbot to embed in a marketing site.
  • Companies whose workflows are genuinely one-off and never repeat.
  • Engineering teams who want a code-only orchestration framework — install laravel-swarm directly.

Why we are building this

From Daniel

The pattern repeats on every engagement. The operator already knows what the workflow should do. They have explained it to engineers, to consultants, to themselves at 11 p.m. The build takes a quarter. The workflow changes in week two of production. The build does not.

Ashlar is the version of the operator's whiteboard that actually runs the business. Agents, tools, humans, and audit all on one canvas — owned by the person who knows the work, watched by the leader who has to answer for it. Built on the framework we ship in public, so nothing in here is a black box we get to redefine later.

— Daniel

Two doors

Wait for Ashlar — or have us build the workflow now.

Ashlar will let your team build their own. Until it ships, we build them for you on the same stack.