Work/ASICS
AI Engineering

Eight weeks from signing to the first agent live.

Our assessment showed that 81% of 42 capabilities needed work before ASICS should scale agents. Rather than build all of it themselves, they decided on our Agentic Runway. We deployed it in their own AWS account and moved their agents onto it.

Customer ASICS
Client ASICS
Sector Retail
Service AI Engineering
Implementation 8 weeks
Platform Brighting Agentic Runway
The challenge

The gaps were known. The calendar was the problem.

Our capability assessment scored the platform ASICS had built themselves. Many capabilities needed work before agents could scale globally. The shopping list was clear.

Closing that list themselves meant many engineering hours and learning curves across identity, tooling, data access, knowledge retrieval and observability. Every one of those is a project on its own. Their small team would be busy with plumbing instead of agents.

ASICS has AI objectives to meet before the end of the year. They needed a mature runway in place, not a build plan for one.

ASICS store display
We worked alongside the ASICS platform team for the full eight weeks.
What we did

ONE RUNWAY, ASICS CLOUD, SEVEN CONNECTIONS

The Agentic Runway is a product we build and license. It already carries the capabilities the assessment found missing. The work was deploying it inside ASICS and wiring it to the systems their agents depend on.

01

Deploy in their account

The runway runs in the ASICS AWS account. Their tenancy, their controls, their spend.

02

Give agents an identity

We connected the runway to the ASICS identity and access management, so every agent gets its own identity in Okta.

03

Open the tool layer

Their most important MCP tools now sit behind one gateway. Agents reach tools through it, not around it.

04

Connect the data

Snowflake is queryable through MCP tools, with the semantic layer attached so agents work from defined business terms.

05

Connect the knowledge

SharePoint is wired in, so agents use enterprise knowledge within the permissions that already apply to it.

06

Put it under observation

The runway emits its metrics through OpenTelemetry. We routed them into the ASICS OTEL tooling, so quantitative and qualitative metrics and their alerts sit in one place.

07

Migrate the agents

We moved their existing agents onto the runway and connected reporting and visualisation tooling so output lands where people already work.

The result

A mature platform, running in the ASICS cloud

8wks
from signing to the first agent live
All
existing agents migrated onto the runway and running
1
central place to manage agent budgets, so no overrun arrives as a surprise
1
place for metrics and alerts, inside their own OTEL tooling

The agents that used to run on the self-built platform now run on the runway, with identities, governed data access and enterprise knowledge behind them. Every agent ASICS adds from here starts on the same base.

What the runway closed

The four domains the assessment flagged now sit in the platform

Agent discoverability

Agents and tools live on the runway and behind the gateway, so what exists is visible in one place.

Agent autonomy

What an agent may reach is set by its identity and the gateway, outside the prompt and outside the code.

Agent intelligence

Snowflake with the semantic layer for data, SharePoint for enterprise knowledge. Agents work from defined sources.

Agent trust

Okta identities tied to ASICS IAM, metrics and alerts in their own OTEL tooling, and budgets managed in one place.

Brighting platform — cost & usage and agents dashboards
What we delivered

A platform, three agents, and a way to add more

The runway, deployed

Licensed by Brighting, running in the ASICS AWS account, configured together with their team.

Seven integrations, live

Okta, the MCP gateway, Snowflake, the semantic layer, SharePoint, OTEL tooling and LLM Models.

Their agents, migrated

All agents moved off the self-built platform, and a team that knows how to put the next one on.

TJ Sizemore, ASICS

“What set Brighting apart is they didn’t just get the vision, they’d already built toward it. No ramp-up, no starting from zero, just momentum from day one.”

TJ Sizemore, Director Data Science and AI, ASICS
What we brought

A product, not a project

The Agentic Runway is available via Infrastructure as Code or via the AWS Marketplace. We offer this as a perpetual license.

Brighting Agentic Runway

Our agent platform, licensed and deployed in your cloud. Built once, implemented across several customers.

Capability Framework for Scaling Agentic AI

The 42 capabilities that set the scope here. It told us what to close, and in which order.

AWS Bedrock AgentCore Okta MCP gateway Snowflake Semantic layer SharePoint OpenTelemetry Reporting and visualisation

More of our work

All cases →
ASICS

Full readiness assessment for AI scalability across 42 capabilities

42 capabilities scored · 2-week assessment
HEMA

Real-time pricing on every shelf, in every store

800+ stores live · 4-year partnership
ASICS

Answers that prioritize the roadmap

Semantic layer on Snowflake + dbt

Know your gaps already?

The Agentic Runway carries the capabilities most teams are still building. We deploy it in your cloud and move your agents onto it.

+31 20 210 13 90 info@brighting.nl Grasweg 183, 1031 HX Amsterdam