Consulting engagements · Limited availability
Retrieval systems that ship at demo quality get fixed here.
I build and operate EAKC, a live enterprise AI knowledge platform of roughly 455,000 lines of production code. I take a small number of consulting engagements applying the same engineering to your retrieval and agent infrastructure. Every engagement begins the same way: your failing queries, diagnosed on a call, at no cost.
455K
lines of production code in EAKC, built end to end
8% → 72%
code-retrieval accuracy after structural chunking, same corpus
521+
automated tests plus retrieval and answer-level evaluation
32
recorded engineering episodes documenting the work
All figures are verifiable — the platform is live at eakc.adaptivesmartsystems.cc. I show my own system on every call.
Engagements
Fixed-Scope RAG Audit
10 working days
An evaluation set built from your real queries, your accuracy measured, every failure mode diagnosed, and a prioritized fix plan your team can execute.
$2,500 fixed · introductoryDetails →
MCP Server Over Your Codebase
2–4 weeks
Your team’s coding assistants — Cursor, Claude Code, Copilot — answer from your repositories: indexed structurally, cited, and permission-aware.
from $8,000Details →
RAG Rescue / Full Build
4–8 weeks
A demo-grade retrieval system rebuilt for production against a measured accuracy target, with the evaluation harness left in place.
$12,000–25,000+ fixed bidDetails →
Not sure which fits? Start with the free teardown.
A 30-minute call. You bring one to three queries that should work and do not; we trace them through your pipeline together. You leave with the diagnosis and at least one concrete fix, whether or not we ever work together.
Applied solutions
Recurring requests from practices and product teams, delivered as scoped builds on the same engine as the engagements above. Each is priced after the free teardown call.
Grounded auto-reply assistant
For specialists, clinics, and support teams · from $3,000
An assistant that answers your team’s and clients’ recurring questions from your own approved material — protocols, FAQs, past answers — with citations, and drafts replies for your review rather than sending unsupervised. Built for practices where a wrong answer is not acceptable: it says “this needs you” instead of guessing.
Entry price covers non-patient-data scope (general questions, policies, logistics). Handling health records or other regulated data is scoped and priced separately.
Discuss this →Voice appointment booking & tracking
For clinics and service businesses · from $4,000
A voice agent that answers calls, books and reschedules appointments against your real calendar, confirms by message, and logs every interaction for your front desk to review. Scoped as a build: your telephony, your calendar system, and clear human-handoff rules for anything outside the script.
Build price plus an optional management plan from $300/month; per-minute call charges are billed at provider cost, passed through transparently.
Discuss this →RAG inside your existing product
For product teams · from $5,000
Retrieval added to the product you already ship — search over your customers’ data, a grounded assistant in your app, or citations on top of your existing content — delivered as an API your team owns, with the evaluation harness that proves accuracy before launch.
Discuss this →How an engagement runs
- 1 — DiagnoseThe free teardown call. Your failing queries traced live: retrieval, chunking, ranking, or grounding.
- 2 — ScopeA one-page written scope: what is broken, the proposed work, a fixed price, and a timeline. You decide without follow-up pressure.
- 3 — Deliver, measuredWork ships against an agreed, measurable target — evaluation-set accuracy, latency, or cost — so you watch the number move, not just the code change.