Kijestic AI. Applied AI and implementation.
Applied AI, implemented.
We work out which workflow is worth fixing, connect the right tools to the software you already use, and build and run the system around that workflow.
Who it is for
Operations with a recurring workflow problem.
Multi-location service businesses, appointment businesses, ecommerce and consumer brands, and operations-heavy companies. Client work to date spans med spa, skincare, supplements, and hair, barber and nail salons.
Proof
What we have delivered, and where each capability stands.
-
Agent workflows
Defined work done by AI inside rules you set: follow-ups, replies, hand-offs. Sensitive steps wait for a person you named.
Approval steps and action records are tested in our own staging with sends simulated.
-
AI operations
We run what we build: watching it, correcting it, and changing it when your business changes.
In operation for a skincare brand's content review.
-
Integration and site implementation
Booking, search and site work connected to the software a business already pays for.
Delivered for a med spa: online booking and search work on its existing site.
-
Customer-facing AI
Guidance tools that run where your customers are, inside limits set in advance: guidance, not a diagnosis.
Delivered: an on-device skin scan on a live store.
-
MCP servers and connectors
Tools that let AI assistants work with your systems, inside limits you set.
We built one ourselves: Kijestic Acceptance, in private validation. Not yet delivered for a client.
-
Internal AI systems
Kijestic runs its own operations on an internal system. Review gates sit in front of anything a customer sees.
It is not a product we sell.
-
Acceptance and governance
Agreeing what done means before the work starts, and checking it on evidence.
Used in our own work. Not yet delivered for a client.
How we work
We do not start with AI.
Start with the business.
Bottlenecks, labor, leads that leak, customer friction, revenue and operating risk. The workflow comes first; the tool comes last.
Work out what it is worth.
What is solving the problem worth in time and money? We do not recommend a build just because AI can do it.
Use what you already have.
Your existing software, existing AI products, connectors and APIs come first. We build custom parts only where they add enough value to justify them.
Define done before building.
Acceptance criteria, the output you should be able to see, what failure looks like, and which system of record settles it. Agreed in writing before work starts.
Implement the smallest useful system.
Connect the systems, test it, put it to work, and train the people who will live with it.
Operate and improve.
We watch for failures, integration drift, model changes and workflow performance, and report what the system did.
What we build
Systems with a job, not features.
Lead response
Capture, qualify, route and follow up, so an inquiry is not left waiting for a free hour.
Sales operations
Pipeline updates, research and follow-up that keep deals moving without manual entry.
Customer support
Triage, answer, escalate and record, with a person on anything that should not be automated.
Operations
Repetitive administrative work, reporting and coordination between tools.
Revenue reactivation
Old leads, inactive customers and lifecycle follow-up, designed to stop when someone opts out.
Internal knowledge
Search, synthesis and operating support over your own documents and records.
These are the kinds of systems we scope. What we have delivered, what runs in our own operation and what is offered but not yet delivered for a client is listed under Proof.
Where to start
Before you buy more AI, find out where it belongs in your operation.
The AI Systems Audit works through your workflows with the people who do them, and gives you a written audit with the economics shown as estimates from your own numbers.
How an engagement works
Audit, implementation, then operation.
Each step is separate. You can stop after any of them.
-
Step 1
AI Systems Audit
A written audit of where AI belongs in your operation and what it takes to build it. Scope and fee confirmed in writing first.
Smaller question? The AI Opportunity Assessment is a 30-minute conversation and a short written brief.
-
Step 2
Implementation project
One agreed workflow. We design it, connect your tools, build what the project needs, test it, deploy it and train your team.
Priced per project, in writing, before work starts. The proposal lists what is configured from systems we have already built and what is developed for you.
-
Step 3
Ongoing operation
We watch the system, correct it, change it and support it, and send a regular record of what it did.
A separate agreement. Never added to a project without your say.
Questions
What does the audit cost?
A fee set by the scope and confirmed in writing before anything starts. You are never charged automatically.
Who does the work?
Kelvin Tran, the founder, leads every audit.
What do you need from us?
Time with the people who do the work, read access to the systems involved where you can give it, and the numbers you already track.
What if we do not go ahead?
The written audit is yours to keep and act on, with us or without us.
How is the audit different from the assessment?
The assessment looks at one workflow from a 30-minute conversation. The audit works through the operation with the people who do the work and covers every workflow in scope, with the economics and a roadmap.
How is implementation priced?
Per project, in writing, before work starts: the scope, what is configured from existing systems, what is developed for you, the timeline assumptions and the price. Ongoing software access and operations, if you want them, are a separate agreement.
Do I have to replace my current software?
Usually not. We work with the tools you have where they allow it, and check read and write access for your specific tools during the audit. Some tools only allow reading data, some allow no connection at all; we tell you which before you commit.
Do you build custom features?
Yes, when the project needs them and they are in the agreed scope. Implementation is not unlimited custom development: the proposal lists what will be built, and changes are agreed in writing.
What needs my approval?
Approval before sensitive actions is tested in our own staging with sends simulated. It is not yet running for a client.
What support is included?
Each implementation includes testing, deployment and onboarding for your team, and a defined support period stated in the proposal. Monitoring, changes and support after that are covered by an ongoing operation agreement if you choose one.
What does delivery depend on?
Access to your tools and accounts, the permissions those tools allow, decisions from you on the rules the system follows, and third-party approvals such as phone number registration. The proposal lists each dependency and who owns it.
What happens to the information I send?
It is used to review your request and prepare the work, stored with restricted access, and never sold. Please do not send passwords, payment details, or patient or customer personal information. Privacy notice.
We define how a system will be judged before we build it. That practice became a product of its own, Kijestic Acceptance, now in private validation. You do not need it to work with us.