Services
AI Solutions
I’m all about digital growth - for the past 10 years I’ve been working on digital innovations and building dozens of software products. I know each stage of the cycle like the back of my hand. As the CEO at Appunite I coordinate creating exceptional products with entrepreneurs and companies. The combination of my sales and tech experience makes me capable of understanding our clients needs and challenges, and it has helped companies from a variety of industries grow digital. My main goal is always to enable our client to focus on their industry transformation and the core of the business, while Appunite is responsible for building and delivering the products.
See this graph.
Looks familiar?
Each node is a function in your company. AP, AR, sales ops, recruiting, support, procurement, payroll, content.
Each function is a service: inputs go in, decisions get made, outputs come out.
Each service is governed by a CxO and run by a domain expert. The person whose judgment decides whether the close is clean, the deal closes, the candidate gets hired.
The expert's real value is the gap between the SOP and reality. The seventeen exception types they handle every month. The vendor that always needs a second look. The unwritten threshold that overrides the written one.
Call it the conformance gap. In typical functions it's 30%. In exception-heavy work like AP or supply chain disruption, 70% or more.
Two stories about how to scale that work without the expert. Both fail.
Story one: drop in an agent and replace the expert. This is the pitch in every keynote right now.
It doesn't work. Enterprise AI failure rates sit at 75 to 95% and have been flat across every model generation since GPT-3. BCG, Deloitte, RAND, IBM, McKinsey, MIT NANDA all converge on the same number.
The reason it fails: this work isn't bounded, checkable, structured, or verifiable the way code is. The model is good enough. The substrate is wrong.
Story two: write SOPs, hand them to employees, scale by hiring more of them. When that hits a ceiling, layer a generic platform on top. NetSuite, Salesforce, Workday, ServiceNow.
That made sense when code was expensive. One platform amortized across thousands of customers had to be built for an idealized workflow.
The cost of that tradeoff is exactly the conformance gap. The platform fits the documented workflow, not yours. Your team papers over the gap manually, every month, forever.
Both stories fail the same way: they try to replace the expert's judgment instead of encoding it.
Judgment can't be wholesale replaced by an agent. It can be encoded into a system where the expert sits on top.
Claude era changes the math. Code is cheap now. A small team with a capable model can build the custom thing in weeks, not quarters.
And the timing matters. The pace of change today only pays off if your work compounds. Silos don't compound. Disconnected operations don't compound. A generic platform stitched together with manual fixes doesn't compound. Each month is the same month.
Custom backoffice compounds. Every exception you encode, every workflow you tighten, every model call that gets cheaper or smarter, builds on the last one. The system gets sharper while the team runs on it.
So the right move is neither "swap in an agent" nor "buy generic." It is custom backoffice: software built around how your function actually runs.
That's why we focus on custom backoffice tools and business processes as verticals.
As technical experts we team up with the function's domain expert. Not a sponsor, not a steering committee. The person who owns the outcome.
We do the audit first. We use Domain-Driven Design, event storming, and process mapping to surface the real workflow: every command, every event, every exception. The methodology exists to force the implicit knowledge out of the expert's head and onto a shared model the engineers can build against. Without that step, you build a clean version of the wrong thing.
Then we build. Mostly deterministic code, with model calls only where genuine judgment is required. Roughly 85% code, 15% LLM. Replayable. Auditable.
The deliverable is not an agent. It is a custom backoffice toolthat is hyperpersonalized to your workflow. The expert moves from doing the work to setting the policy.
The bet: in the Claude era the moat is not the model. It is knowing how the AP team at a $2B distributor actually closes the books, and encoding that into the system the team runs on.
