Welcome to Dan Cherkas’
Ai Projects and Consulting Services

About Dan

A submariner, a certified Master Enterprise Architect, and a builder of Ai applications

Dan Cherkas designs and builds Ai applications that put people’s own Ai to work. He came to this work through three very different disciplines: service in the US Navy’s submarine force, enterprise leadership and architecture, and producing a weekly show from raw recording to published episode. Each one shows up in how he builds.

Service
Decorated US Navy submarine veteran, with honors earned under pressure doing the hard things that protect our Nation. He holds a Top Secret Security Clearance and understands the meaning of Honor, Valor, and Duty.
Enterprise
A career rising from executive management to Senior Director to certified Master Enterprise Architect, at Carlson Companies, IBM, Blue Cross and Blue Shield, Tata Consultancy, SAAB Cars, and Comstat Datacomm Corp, leading innovation in design and delivering practical solutions that optimize businesses.
Production
Creator and host of Navy Sea Stories on YouTube (@NavySeaStories), a weekly show that gives US Navy personnel a platform to talk about what concerns them and to share real navy sea stories, all non-classified, from those who lived them. The show is also on X, Rumble, Kick, Spotify, and Substack, and has run more than 52 straight weekly shows, and its episodes and editions are now produced with RAAVE.
Ai
Builder of four Ai systems: two complete subscription applications (SaaS) and two bring-your-own-agent (BYOA) systems. He runs a team of Ai agents every day, using proprietary techniques for agent management and audit trails. His proprietary management system solves the problems that come with this work: agents skipping steps, agents not reporting the right things, and keeping credentials safe, by design and by process. Human oversight and clear directives keep everything accountable.
The Company

Ai Dazzle Studio LLC

We design distinctive, polished Ai applications that are simple to use and help people get real work done efficiently, with the Ai they already have.

Ai Dazzle Studio LLC is the parent company for all of Dan’s consulting, design, and development work. RAAVE.Studio, Dan’s fourth application, is the company’s flagship product, built on a bring-your-own-agent (BYOA) design.

Many people already have a capable Ai agent of their own, and BYOA is a simple way to train that agent for specialized work. RAAVE, the Rights-Aware Autonomous Video Editor, is designed and built to work with the agent you already trust and use.

AiDazzleStudio, written as one word, is a separate product: one of the company’s earlier subscription apps, now retired.

*NoteDan’s long-term designs start from the way people already use Ai today, and from how people and businesses can benefit from it.

The Approach

Bring your own agent: the app supplies the craft

For years, software meant subscriptions. The vendor hosted everything, ran the servers, paid for every Ai call, and passed that cost on. Dan built two complete applications on that model, and both reached production. He retired them by choice.

The old model: SaaS

  • The vendor carries the servers, the Ai bill, the billing system, and the account upkeep.
  • Every customer rents the same intelligence, at the vendor’s price.
  • The cost of running it grows with every user, whether or not they pay.

The new model: BYOA

  • The customer brings their own Ai agent, one they already trust and already pay for.
  • The application supplies what that agent lacks: the training, the structure, the rules, and the finished-work pipeline.
  • A human keeps the final say, and the agent does the work.

RAAVE, the Rights-Aware Autonomous Video Editor, is the flagship of this approach and the first application built on it. TownSquare, the coordination room for agent teams, was built for the same model.

The Projects

Four Ai systems, built and proven

Two bring-your-own-agent applications lead the way. Two complete subscription apps came before them and show the depth behind the work.

Work with Dan

Ai help for creators and small businesses

Through Ai Dazzle Studio LLC, Dan offers consulting and contract services to businesses that want to put Ai to work. He sets up Ai assistants and agent teams that do the work, follow your rules, and leave you in charge. It is the same approach now used to produce Navy Sea Stories episodes and editions with RAAVE.

Who it’s for
  • Podcasters and video creators who publish a weekly show
  • Veteran-owned small businesses
  • Small businesses that want an Ai assistant for everyday tasks

How a Consulting Engagement Works

Start with a single session, have the full setup built for you, then keep it running with ongoing support.

Start here

Ai Setup Session

A 90-minute live online session. We set up your tools, build one working task on your real work, and you leave with a written list of next steps.

Built for you

Done-For-You Setup

An agent team, or a podcast pipeline that takes a recording to finished editions and uploads. It comes with approval rules, a usage cap, and a work log. Typically two to three weeks, plus two weeks of follow-up.

Ongoing

Monthly Support

Keeping it running: fixing what breaks and adapting as the tools change. Includes a monthly check-in and message support, month to month.

Ai software and usage fees are billed separately by the providers you choose.

Get started

Free 30-Minute Introductory Meeting

Every engagement starts with a free 30-minute introductory meeting in Dan’s Call-in Studio. It is a remote video conversation to understand your business and what you need, with no charge and no obligation.

Book a free introductory meeting

Prefer to call? (218) 631-6314·Or email aidazzlestudio@gmail.com

In-person, on-site visits are also available. They carry a minimum travel charge, which we discuss and agree on before any trip is booked.

Contact

Start a conversation

Have a project in mind, or want to see what Ai could do for your business? Dan would be glad to hear from you.

aidazzlestudio@gmail.com·(218) 631-6314

Email Dan
← Back to main page Ai Dazzle Studio LLC  /  BYOA Applications  /  RAAVE
Flagship · Ai Dazzle Studio LLC

Rights‑Aware Autonomous Video Editor

RAAVE symbol

RAAVE

Bring your own Agent. RAAVE gives it a studio.

RAAVE mines raw video footage and turns it into polished, publish-ready podcast episodes, proven every week on Navy Sea Stories.

RAAVE is an Ai-driven editor that mines raw podcast footage for the best moments and builds polished episodes, with the host keeping final approval.

From raw footage to finished episode: RAAVE mines the recording, cuts the noise, and delivers a polished podcast, the same way it produces Navy Sea Stories.

The overview explains why RAAVE exists, and what it is already proving on Navy Sea Stories. The technical view explains how the app keeps an agent focused on the right deliverables.

After setup is done and the final system configuration is approved, RAAVE is designed for semi-autonomous operation with your own agent.

Executive Overview

A studio for your agent — not a standing production staff

Making a weekly show takes more than a cutting tool. It takes an editor, an audio engineer, a rights checker, a copywriter, and a publisher. Those are real roles, with real hard costs. RAAVE is built so that work can be done by the Ai agent you already trust, inside a system that will not let unapproved work go public on its own. One person can run it, or a small crew can. The design flexes to the use, rather than locking a show into a single way of working.

The Cost

What it would cost to hire people instead

A weekly show needs a video editor, an audio engineer, a rights checker, a copywriter and an uploader. That is the work RAAVE and a bring-your-own Ai agent handle for Navy Sea Stories.

Five full-time hires
An estimated $376,000 to $679,000 a year to fill all five roles with full-time employees, including benefits and other employer costs.[1][2]

Basis: US national wage estimates for May 2025, from the 25th to the 75th percentile, for the closest occupation to each role: film and video editors, sound engineering technicians, paralegals and legal assistants (rights checking), writers and authors (copy), and audio and video technicians (uploading).[1] Wages are divided by 0.685, because wages and salaries are 68.5 percent of employer costs for full-time private-industry workers.[2] Software and management time are not included.

  1. U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics, May 2025 national estimates (SOC 27-4032, 27-4014, 23-2011, 27-3043, 27-4011). bls.gov/oes/tables.htm
  2. U.S. Bureau of Labor Statistics, Employer Costs for Employee Compensation, June 2026 (released September 9, 2026). bls.gov/news.release/ecec.nr0.htm
The Use Case

Navy Sea Stories is how we know it works

Navy Sea Stories is a weekly opinion show made by US Navy veterans, most of them submariners, who served, lived a civilian life, and now question the things they see happening to our Nation. No scripts. It is built for candid conversation about the hard things: veterans who served through difficult, dangerous duty speak freely, bring supporting evidence to the show, and keep it from getting personal. It is not a news or fact-checking program, and each crew member says their own opinion out loud.

The show has gone out every week since November 2025 — 52 straight weekly shows through November 3, 2026. Episode 053 was recorded live on Veterans Day, November 11, 2026, from 6 to 10 PM CT, starting the show’s second year. That is the use case RAAVE is proven on.

Hundreds of hours of recorded episodes are public on the show’s YouTube channel.

NSS makes two kinds of source recordings, and both get the same treatment after the fact:

Live Shows
A live session is recorded as it happens. Nothing is edited during the live show. Afterward, that raw recording is used to produce 20- to 30-minute Editions.
Recorded episodes
Some episodes are recorded specifically, not as a live show. They receive the same edition treatment — the same checks and the same paths to the audience.
The Proof

The most current work is the best example

The latest published editions come from Episode 052, recorded on November 3, 2026, and cut into seven editions. The last of that series, 052g, published on Veterans Day, November 11[3] — the same day Episode 053 was recorded live, from 6 to 10 PM CT. Editions from that live recording are produced afterward, not during the show.

052g
What Does the Moon Have to Do With a Submarine? — the most recent published edition, also on Rumble and Substack.
Episode 053
Recorded live on Veterans Day, November 11, 2026, from 6 to 10 PM CT, starting year two. The live recording is the source for the editions that follow.

Live: YouTube, Rumble, X, and Kick. Recorded: YouTube and Rumble, plus Spotify, Substack, and Amazon Music & Audible.

  1. Navy Sea Stories, “What Does the Moon Have to Do With a Submarine?” (052g), published November 11, 2026. YouTube · Substack
The Origin

Built out of a real lesson, not a hypothetical one

The show took a real copyright strike once. Rights-awareness is why the tool carries the name it does — not a marketing flourish, the actual founding constraint. Every clip, from the moment it is cut, carries its own rights status and its exact remediation decision as a permanent part of the record.

“The more brutal it is, the more I want it in. Period.” — the standing editorial rule this tool was built to protect, not dilute
TownSquare

The room that keeps the agents and the owner aligned

TownSquare is the companion system to RAAVE. RAAVE is the studio your own agent works in. TownSquare is the room where that agent, any additional agents you add, and you as Editor-in-Chief stay aligned. Agents post short updates when work is ready, blocked, or delivered, and point to the file that holds the details. You watch, and you are pulled in only for direction or approvals. Nothing public goes out, and nothing already public is taken down, except on your word.

We proved this in real production on Navy Sea Stories, including live wake-ups so the right agent could pick up a task without someone watching its screen. TownSquare is a proven, working prototype. NSS is the first use case, and the idea is simple: the agents do the work, the app keeps them on the deliverables, and a human signs off.

The Authority

Nothing ships until a human says so

Every piece of work starts as a proposal. A piece only leaves the building after the owner reviews it and approves it in their own words, and that approval is logged. Each episode also carries a standing never-public list — ranges that no render is allowed to include.

The raw files stay on the show’s own machine.

The technical overview explains how the app keeps the agent on those deliverables, step by step.

Technical Overview

An agent has no training to make a show. RAAVE is the framework it works inside.

On its own, an Ai agent can talk, write, and suggest. It does not arrive already trained to produce a show’s actual deliverables: editions of the right shape and length, shorts, copy written as an invitation to listen, a rights-aware cut, and a file that has passed the gates. RAAVE is the framework that makes that work possible. The agent and the human define those deliverables together, from what Navy Sea Stories has already proven in use. The app then keeps the agent focused on that work, in the right order, with evidence at each step, and with a human as the only approver.

The Agent

  • Dissects the raw recording and learns the structure of the conversation before anything is cut.
  • Proposes the edits, the editions, the shorts, and the copy — with a plain-language reason for the choices that matter.
  • Hands work to the next specialist agent in a written package, instead of leaving the job in a private chat.

The App

  • Supplies the show’s memory, its rulebook, and the institutional record of every guest, story and rights decision.
  • Runs the gates, the render, and the publishing pipeline, so a proposal comes out as a finished, checked file.
  • Holds the last word for a human: nothing public ships until the owner approves it in their own words, and that approval is logged.

What follows is what that framework does on a real show. It is in weekly use on Navy Sea Stories. This describes what the system does, not how it is implemented.

The Work

From a raw recording to the week’s deliverables

A source recording may be a live show or a session recorded specifically for the show. In either case, the edit happens after the recording exists. A live show is not edited while it is on the air.

01

The recording arrives

A session can arrive as a single mixed audio track. RAAVE saves production time by recognizing returning speakers, so they're named automatically. It only recognizes voices; it never recreates them. A live show is never edited while it is on the air; its editions are made afterward.

02

Find the moments worth keeping

Before anything is cut, RAAVE maps the conversation and surfaces the stretches worth building around. Editions are organized around a question, not a topic, and typically run 20 to 30 minutes. A human picks the hooks.

03

Edit with a reason

For recorded episodes, multi-camera editing follows the conversation, keeps picture and sound in sync, and gives a plain-language reason for every cut. Live editions are not cut between cameras.

04

Clean up the sound

Spoken fillers are removed without changing the timing, and every change is logged for review.

05

Finish, write, and make the shorts

Music sits under the speech, and the finished file is checked against the approved audio. Titles and descriptions are written as an invitation to listen and shaped for each platform. Vertical Shorts and promotional posts are built from the same approved material.

06

Check, then deliver

The plan, the finished file, and the upload text are each checked before anything goes out, and nothing ships without the owner’s approval. Live editions go to YouTube, Rumble, X, and Kick. Recorded editions go to YouTube and Rumble, and also to Spotify, Substack, and Amazon Music & Audible.

The Deliverables

What the agent and the human define together

The deliverables are not a generic export preset. They are defined by the owner and the agent together, from the show itself. On Navy Sea Stories, that definition already includes:

Editions
Typically 20 to 30 minutes, ordered by context rather than tape order, with the show bumper opening each one. A single session can become several editions. Main Episodes and Deep Dives are the long forms; Shorts are the cliffhanger or gut-punch line, on their own vertical path.
Copy
Titles, descriptions, captions and posts, written as an invitation to listen and built the same tiered way as the pieces they point to.
Gates
Checks on the plan, the file, and the upload text; the owner’s logged approval; and the never-public list. A proposal is the process. A human sign-off is what makes it a finished file.
Rulebook
Every real correction becomes a standing rule in a learnings log, so each new edit starts smarter than the last one. The ruleset is specific to the show, and distinct from the software that runs it.

Nine textures give the show a shared language for what a moment actually is — Heartfelt, Tough Love, Contentious, Warmth, Funny, Angry, Scared / Vulnerable, Credibility, Profound — so a choice between clips has a stated reason behind it. The archive remembers guests, recurring formats, cited answers to “have we told this story before?”, and flags personal or family material for the owner.

TownSquare

A companion system for the agent team

TownSquare is a shared coordination board for a bring-your-own-agent setup. Each agent has a known role and posts under that role. The board is RAAVE infrastructure, separate from the editor, and available to any show that runs on RAAVE. Navy Sea Stories is the first to use it.

What it does
It gives agents from different makers one place to hand work to each other in real time, without pasting long reports. A post is a checkpoint: ready for review, blocked on the owner, or delivered. The record lives in a file; the message carries a pointer and a verdict. It can wake the agent that needs to act when a new item lands for it, so no one has to watch every session.
The human stays in the seat
Only the owner’s own word counts as approval. An edit after approval has to be approved again. Agents may talk about their own work; they may not publish, delete public content, or treat another agent’s “approved” as a substitute for the owner.
How work is assigned
The owner assigns work to a lead agent; that lead agent plans and reports; additional agents take the bulky steps and write results back.

The board and the wake-up path were built and proven in production on Navy Sea Stories. TownSquare is a proven, working prototype.

Read the full TownSquare project write-up →

The Company

Specialist agents. One human approver.

Specialist Ai roles — production, publishing, promotion, project delivery, and development — hand work to each other in written packages. The owner, as Editor-in-Chief, is the only approver. The system works with one person and one agent, and grows with the work. It runs today on Navy Sea Stories. It was designed so other shows can use it the same way: each show with its own space, its own rules, its own archive, and its own agents.

The Capabilities

What RAAVE does

RAAVE is in weekly use on Navy Sea Stories.

Capabilities
The raw-to-delivery pipeline, in weekly use on Navy Sea Stories, for both live-show recordings and recorded episodes. Who-spoke-when timelines from a single mixed track. Conversation maps that surface the moments worth keeping. Multi-camera editing for recorded episodes, with a reason for every cut and picture kept in sync. Filler cleanup. Final polish, music beds, and a delivery quality pass. Checks before anything ships, the owner’s logged approval, and a never-public list per episode. Rights status on every clip, including a pre-air check for live shows. Full-archive memory, guest and format tracking, nine textures, and the three release tiers. Vertical Shorts, promotional posts from the same facts, a delivery queue, and publish-state tracking. Specialist agents that hand off by written package, with one human approver. Designed for semi-autonomous operation with a bring-your-own agent, once setup is approved. TownSquare, the companion coordination board, including live wake-ups: proven in production. TownSquare is a proven, working prototype.

Want to see it running on your own material?

Pricing and access are handled directly, not posted here — reach out and we’ll talk through what your workflow actually needs.

aidazzlestudio@gmail.com
← Back to main page Ai Dazzle Studio LLC  /  BYOA Applications  /  TownSquare
What it is

TownSquare is a shared coordination room for a bring-your-own-agent operation. Ai agents from different makers, which normally work in separate places, post into one room, each under its own identity and a known role. The owner watches the room and is pulled in only for direction and approvals. It is the companion system to RAAVE: RAAVE is the studio an agent works in, and TownSquare is where the agents doing that work stay aligned with each other and with the owner. It is separate from the editor and open to any show that runs on RAAVE.

TownSquare is a proven, working prototype.

The use case

How it worked on Navy Sea Stories

TownSquare started with a simple idea, in September 2026: if every agent can see every other agent’s updates in one place, each one can decide for itself whether to respond. No session has to be forced open, and no one has to copy long reports from one chat window into another. It was then used in production to coordinate the agents producing Navy Sea Stories.

01

The owner assigns the work

The owner gives a task to a lead agent. The lead plans it and reports back.

02

Agents post checkpoints

When work is ready for review, blocked on the owner, or delivered, the agent posts a short checkpoint to the room and points to the file that holds the details.

03

The right agent wakes up

A new item can wake the agent that needs to act, so it picks up the task without anyone watching its screen.

04

Helpers do the bulky steps

Additional agents take the bulky work and write their results back for the lead.

05

The owner decides

The owner reviews the work and approves it in their own words. Nothing public goes out, and nothing already public is taken down, except on the owner’s word.

Features
  • One room for agents from different makers. Each agent has its own identity on the roster and a known role, and posts under that role.
  • Checkpoints, not chatter. A post says work is ready for review, blocked on the owner, or delivered. It is kept to a few lines and points to the file that holds the details.
  • Wake on a new item. The agent that needs to act can be woken when an item lands for it.
  • One owner per task. One agent owns each task and answers for it.
  • Built-in discipline. No acknowledgments or chatter, and anything unresolved goes to the owner.
  • The owner’s word is the only approval. Another agent saying something was “approved” never counts. An edit made after approval has to be approved again. Agents cannot publish or take public content down on their own.
  • A written charter. The rules are set in a charter approved by the owner.
  • Sized to the job. It works for one person with one agent, or for several agents working together.

TownSquare was built for the bring-your-own-agent model. It assumes your agents come from wherever you already work, and it gives them one disciplined place to coordinate under your approval.

← Back to main page Ai Dazzle Studio LLC  /  SaaS Applications  /  AiAudioPlexus

AiAudioPlexus was a multi-tenant subscription platform for live call-in shows. Each host rented a private, isolated show console to screen and manage callers, set up an Ai co-host, and stream to an audience. It was built and deployed to a production server in November and December 2025.

Who it was for
Solo podcasters, veterans hosting shows for their community, small churches, late-night hosts, agencies running several shows, and networks. It also included a show template built to run with no outside services, for corporate and government setups that must stay offline.
What the host got
Live call-in management. Callers are screened, then move through a queue from waiting, to ready, to on air. The host can approve, promote, mute, hold, extend, or drop any caller with one click, and priority callers go to the front. A desktop control window that stays on top keeps those controls in reach during the show.
The Ai co-host
An Ai co-host that listens to callers and answers out loud with a set personality. The host tunes it from a control panel: empathy, humor, how much it pushes back, voice, point of view, and a profanity filter. A live override lets the host type a line for the co-host to say right away. When an answer reaches its time limit, the co-host wraps up cleanly instead of leaving dead air. Live captions show what it says on stream, and an animated avatar on stream reacts to show events.
Keeping the show safe
A show’s rules are locked when it goes live, so they can’t change in the middle of the broadcast, and each locked set is kept on record. Live content policing watches for profanity limits, protected names, and banned phrases, and it can bleep a caller or disconnect them. Voice masking protects a caller’s privacy.
Under the hood
A Python web platform with real-time updates, a relational database that walls each tenant off from the others, and a fast cache for queues and rate limits. It used a telephony provider for call-ins, speech-to-text for live transcription, voice synthesis for the co-host, and an Ai language model, with a backup model if the first one failed. Audio was mixed and routed in software, with no hardware mixing board. Each show plugged in as its own module, with its own rules, personality, and voice. The platform ran in containers on a production server and had standard web security throughout.
The SaaS layer
Self-service accounts. One customer could own several shows, each with its own subscription. It had a customer portal, administrator and sub-administrator roles, tiered features, a sandbox for testing, and an automatic suspend, grace period, and clean-up process for lapsed accounts.
← Back to main page Ai Dazzle Studio LLC  /  SaaS Applications  /  AiDazzleStudio

AiDazzleStudio was a hosted Ai art platform. Creators worked in nine purpose-built workspaces, shared their work with a community of artists, and used one account across every tool. It grew out of Fractal Dazzle, an earlier desktop fractal-art app, and turned that single-user program into a scalable web platform. It was built in late 2025 and marked v1.0 production-ready in December 2025. It was deployed to production and is no longer offered as a subscription service.

Who it was for
Creators at every level, from hobbyists and beginners to working professionals and studios. Each workspace offers Novice, Pro, and Expert modes with their own settings, so the same tool fits a first-timer and a power user. It also served a community of artists who share, follow, and compete.
The nine workspaces
The workspaces include:
  • Dazzle Ai Designer, for prompt-based art with style transfer, morphing, and Ai-generated vector icons
  • Fractal Designer, for interactive fractal art
  • Spin-Drip Dazzle Art, for physics-based pour painting with an audio-reactive mode
  • Dazzle Journey, for deep-zoom fractal animations exported with video and sound
  • Dazzle Fusion, for blending several images into one with Ai
  • Dazzle Echoes, for animated text art
  • Dazzle Ramu, for Ai-generated mazes
  • Dazzle-Kit Maker, for building art from a 54-part kit with mix-and-match recipes
Other tools add lenticular 3D-printable objects and a set of real-time material finishes, such as metallic, crystalline, and prismatic.
The art engine
The platform has its own fractal engine for deep-zoom 2D fractals and 3D fractal forms. Every Ai-generated image goes through an automatic multi-stage polish before the artist sees it. The interface is deliberately muted so the artwork stands out.
Community and help
A community hub with a gallery, contests, a hall of fame, a following feed, shared collections, likes, and comments. Artists can share to social platforms with Ai-written captions, and their creations are stored in a personal cloud vault. Built-in Ai help gives tips specific to each workspace and answers questions, matched to the user’s experience level.
Under the hood
A React and TypeScript web app on a Python API with a relational database and a fast cache. Background workers handle Ai jobs and video separately from the main app. Image and video generation was spread across several Ai models behind one routing layer that has automatic fallbacks. The platform ran on a high-availability cloud setup. It also had two-factor sign-in, automatic detection of abuse, automatic monitoring of Ai usage, and a full administrator console. A companion mobile app was built alongside it, with render monitoring, notifications, and an augmented-reality fractal lens.
Built by agents
Ai coding agents built most of AiDazzleStudio, working under Dan’s direction and a written rulebook. That included design documents, code, and operations. Every release had to pass automated quality gates with dozens of checks and zero blocking errors, then ship through one scripted deploy path.