AI implementation
2690 is an AI implementation agency for growing consumer brands. We pick one recurring job, connect the information it needs, and build AI that drafts, checks or flags the work. A named person on your team approves every consequential step. Founders Matt and Mike Lady do the work directly.
By Matt Lady and Mike Lady, co-founders of 2690 · Updated September 2026
AI with a defined job inside a workflow, not a general chatbot bolted onto the side.
Drafting, checking, summarizing or flagging inside a job your team already does, such as turning customer feedback into a creative brief or checking campaigns before they go live.
A shared record for the job, with owners and sources, so AI reads the same trusted data as the team.
An internal app when the capability does not exist in your stack. See the 2690 Build.
An AI-built app that works in a demo but is not ready for the team. See AI app readiness.
A workspace built around one team job, with boundaries and training. See shared AI workspace.
Most AI implementation pages describe what AI can do. This is where the approval sits in one real plan from our library: checking a launch before it goes live. Illustrative, not a client result.
Reads the launch plan, the live discount codes, shipping rules and stock, and lists every mismatch with the source it came from.
The launch owner sees each flag, opens the source, and marks it fixed, accepted or blocking.
Go-live waits for the launch owner's approval. AI never changes a price, a discount or a live page.
A wrong flag is corrected on the record, so the next check learns from it and the reviewer can see what changed.
See the full plan: check discounts and shipping before launch.
Four steps. You make the business decisions; we handle design, engineering, testing and handover.
We start from a recurring job and a recent example of where it went wrong. We decide together whether AI belongs in it at all.
We identify the sources, owners and access the job needs, and fix the record before adding AI.
We test AI output against examples your team has already judged, including the awkward exceptions.
Your team gets the agreed code, documentation and operating steps, and knows how to spot an output that needs a second look.
We connect Shopify, ERPs, 3PL portals, support desks, review platforms, analytics exports and shared documents where they have reliable interfaces. For the operations side, see ecommerce automation.
AI gets access to just the sources the job needs, with personal details removed where they are not required.
We pick models and vendors for the task and explain the running costs before you commit.
No case studies here: 2690 does not have client results to show yet. You work directly with the two founders.
Co-founder · Agency and brand operations
Matt has ten years in agency, brand and SaaS growth and operations, including advising more than 30 DTC brands in his prior work. He decides with you which job is worth doing and where the approval belongs.
Co-founder · Software engineering
Mike has twelve years building production software at a Fortune 500 company and leads an enterprise AI tools team whose tools support 10,000 engineers. He builds and tests the AI steps and hands them over.
A quick check before you get in touch.
A growing consumer brand with a specific AI idea or a recurring job that needs reading, comparing or checking before a person decides.
You want AI agents that act with no approver, a chatbot for its own sake, or a company-wide AI rollout before one job works.
If you are not sure which AI work is worth doing, see fractional AI officer support.
Short answers to what brands usually ask.
A team that takes AI from an idea to something your staff use in daily work: choosing the job, connecting the data it needs, building and testing it, and handing it over. 2690 does this for growing consumer brands.
Where a recurring job needs reading, comparing or summarizing before a person decides: grouping customer feedback, checking launch details, preparing a weekly review, or flagging order and payout exceptions.
No. AI drafts, checks, summarizes or flags. A named person on your team approves anything consequential, and each output shows its source so the reviewer can check it.
You need the information for one job in a known place with an owner. Often a shared record is the useful starting point, before any AI is added.
Both parts of the work. Like an AI implementation consultant, we advise on which job AI should do and where a person approves. Then we build, test and hand it over, which is the agency part.
Your team owns and runs the result. We hand over the agreed code, documentation and operating steps, and train the people who will use and maintain it.
Tell us what you want AI to do and show us the work it would touch. We will tell you where we would start.
Tell us what you have in mindMay we use Google Analytics to understand visits and improve this site? Analytics cookies are optional. Privacy details