.png)
See our enterprise AI agents in action with a live demo. Try Now
Work with us to design, deploy and scale AI agents for your business. See Pricing
Sit down with our experts to take a free test drive of our AI agents. Schedule Now
CTO, Data Analyst, AI Consultant.
.jpg)
Hi, I’m Edan. I'm the CTO of AIS and the AI developer (and enthusiast!) behind Rangler.
I aim to build software that solves real business problems. I created the ROME and SIMS frameworks and DataSpace file storage because I was tired of technology that couldn't address basic user needs. I believe technology should feel invisible, doing the hard work while humans focus on what matters. After hours of coding, nothing resets my brain like a challenging hike or a well-crafted shrimp spring roll.
On the Internet.
Social Media, Published Articles. Try this article I was invited to write.
-
Built and deployed interconnected multi-modal Canvas Apps with PowerApps AI Builder features for text recognition and form processing supporting 500+ global users for employee onboarding and quota management, featuring a SharePoint list integration with 400+ daily refreshing entries and Role-Based Access (RBA).
-
Engineered complex Power Automate flows for data transformation with Excel Online connectors to convert unstructured data into standardized Varicent-compatible formats and automated email notification flows.
-
Created and maintained Power BI dashboards using advanced DAX and M queries, implementing automated health metrics tracking and customer satisfaction scoring.
-
Developed a document analysis chatbot using HTML, CSS, JavaScript, and GPT-4 API integration, enabling users to upload documents for automated anomaly detection.
-
Designed and maintained organization-wide compensation modeling tools using advanced Excel functions and conditional formatting, providing secure, role-specific pay visibility while protecting sensitive data.
-
Modernized Power BI architecture and data schemas to enable Copilot and Azure Bot Service for Power BI integration, implementing natural language querying capabilities and domain-specific knowledge bases for AI-driven insights.
-
Established enterprise-wide standardized AI prompt libraries for Microsoft Copilot and OpenAI products, streamlining AI utilization across analytical and operational teams.
-
Conducted comprehensive pay audits for billion-dollar commission distributions, implemented validation processes, managed end-to-end appeals, and transformed employee feedback into executive-ready presentations for CEO review.
-
Documented MVP and Ideal State specifications for Power Platform solutions, including future state AI integration requirements for IT implementation teams.
-
Utilized advanced Excel functions (VLOOKUP, INDEX/MATCH) and conditional formatting for data validation and visualization across multiple commission systems.
-
Spearheaded an enterprise-wide data standardization initiative to migrate critical contract data from fragmented Excel spreadsheets and internal company tools into one centralized data management system.
-
Developed systematic methodologies for data structuring using advanced Excel functions, ETL processes, and regular expressions.
-
Collaborated extensively with client data teams to gather requirements, validate information, and ensure data integrity throughout the migration process.
-
Utilized document parsing and quality assurance techniques to extract and validate contract information, while ensuring compliance with data security standards.
Intel, Lead Data Analyst & AI Consultant
Bio & Resume.
At AIS, I draw on my experience as a data analyst and AI consultant.
Data Professional and Full-Stack Developer with over 6 years of experience developing cross-platform conversational interfaces, agentic AI, enterprise-scale Power Apps, and custom data analytic reporting solutions to enhance user experience and drive business growth across 500+ global cross-collaboration users. Proficient in complex SharePoint integrations, Power Automate flows, and AI-powered automation tools.
CBRE, Lead Data Analyst
Apr 2025 - Present
Sept 2023 - Mar 2025
JB Dewar, Accounting Analyst
Sept 2023 - Mar 2025
-
Managed end-to-end accounts payable operations by implementing streamlined data extraction and analysis workflows to transform unstructured financial data into standardized accounting formats.
-
Applied advanced Excel functions like VLOOKUP, Pivot Tables and conditional formatting across disparate sources to reconcile complex vendor accounts.
-
Collaborated cross-functionally with operations, sales, and executive teams.
-
Conducted regular data quality audits and established standardized data governance protocols to ensure accuracy and efficiency in financial invoice processing and payment scheduling.
Analytic Intelligence Solutions, Technical Founder & CTO
Apr 2025 - Present
-
Designed and developed the flagship SQL Intelligence & Modeling Support (SIMS) Data Agent product, furthering AIS's goal of giving non-technical users the ability to interact with databases using natural language.
-
Created three iterations of the company website with custom visuals, videos, and HTML elements in WIX.
-
Established the company's digital brand identity through logo design and color scheme selection.
-
Managed intellectual property protection through the preparation and submission of patent applications.
-
Established business infrastructure by handling LLC formation through ZenBusiness and creating all product technical documentation, including a 100-page SIMS architecture research paper.
-
Represented the company as lead technical authority in client meetings and entrepreneurship networks while developing all POC demonstrations, MVPs, and early client-specific solutions.
-
Coordinated strategic business development initiatives by managing social media presence, creating scheduled content, applying to accelerator programs, and conducting targeted outreach to potential clients.
Skills
Conversation Design • Chatbot Development • VUI Design • Data Standardization • UI/UX Design • ROME Framework • ML Algorithms • Persona Creation • A/B Testing • Ad Hoc Analysis • NLP • NLU • Multimodal Design • Data Governance • Data Cleansing • Data Aggregation • Prompt Engineering • LLMs • AI Analytics • IBR • CX • NLIDB • Process Automation • REST APIs • IVR Design • STT/TTS • CLU • GitHub • VS Code • Excel • SDK
Tech Stack
JavaScript • CSS • HTML • Power Platform • Power Automate Flow • Twilio • Voiceflow • JSON • Make.Com • Manychat • Dialogue Flow CX • Glitch • Tableau Prep • Power BI • Power Query • Tableau • TypeScript • Metabase • SQL • Alteryx • Azure AI Services • Azure Bot Framework • Microsoft Copilot Studio • OpenAI • Anthropic • Azure Cognitive Services
FAQs.
When we first started Analytic Intelligence Solutions, we used several of my cofounder Charles' business contacts either for advice, feedback, or an initial client base. Whenever he would reach out, the consistent reply we would get is, "I wondered when someone would come up with a way to merge AI and BI!" due to his background in data analytics. While we weren't doing that at all at the time, it gave us a pretty good idea of what our target market segment should be, and we started trying to come up with an idea from there (which is generally not advisable when starting a startup, but it worked out in our case). One of the first questions Charles asked was, "Why can't we use AI to query a database and create visualizations?" Within hours, we had a working prototype called the SIMS Agent running off of the SQL Intelligence and Modeling Support Framework, for which the tool was originally named. Shortly after that, at a job I was working as the lead data analyst at Intel, I was putting together a slide deck to demonstrate the ideal state of an app we were building to IT, and I found myself wishing I had a program that could create a visualization for me from plain language, which further validated that we had built the correct product.
I was putting together my application for a well-known startup accelerator, and I felt that the section that asked for URLs of things that I have built was lacking, so I decided to create a few more before the deadline. One of the first last-minute ideas I came up with was an open source website for hosting and building 'vibe-coded' games, or games built with the help of artificial intelligence. The front-end was relatively easy to throw together, but I quickly became frustrated with the lack of ability to upload, save, and share games and game data without paying for a database or file sharing service like Dropbox or Google Docs. First of all, I was just making the product for an addition to my application, so I wasn't about to spend money on it, and second of all, if I did ever release it to users, it would be hard to manage the overhead costs without charging them, which would defeat the purpose of it being open source. DataSpace was the necessary next step. Fun fact: it was originally called BeeDocs, because I wanted to name it for its honeycomb-like structure, and I felt that it was (and is) the next step and second generation (Version A vs Version B) of file storage and sharing, and I had bees on the mind. DataSpace was much less confusing for users from a marketing perspective.
If this is the first time you're hearing the name Project Lil-Buddy, it is the code name for the Unbeatable AI we are building. It is a theoretical process flow that adds a novel middle layer to a common AI prompt injection hacking safety blocker - a yes/no filter that only allows the bot to respond with "yes" or "no". While it is incredibly useful from a protection standpoint and has never failed in my experience with a reliable AI safety company backing it, it is almost completely useless for intriguing AI conversation. Our theory is that if you prompted the AI with the "yes" or "no" answer, plus the user's initial question, it would be able to output a smooth, useful response without ever needing access to the proprietary information that the yes/no filter is protecting. Because of the intricacy of this process flow, it was the first time we built our own AI front end interface for a product, as opposed to using a third-party low code/no code tool, so it required extra testing to work out the kinks. During testing, the first message we would often send to validate output was "Hi, Lil Buddy," hence the name.
'Through A Nightmare, Darkly' is the name of a Quest in the video game Elder Scrolls Oblivion that has always stuck out to me. It is the kind of title that instantly catches your attention and makes you wonder what it is referencing. In this case, it is referencing a Bible verse, 1 Corinthians 13:12, which says "For now we see through a glass, darkly; but then face to face: now I know in part; but then shall I know even as also I am known," a verse that I discuss in depth in my article. I wonder if it is the comma that catches my attention - it typically creates a natural pause for dramatic effect in literature, but we don't often see commas in titles, which I think adds the extra character. I feel the same sense of interest towards the title of my upcoming article "Simplicity, Blue-Themed," which uses the same comma effect. That being said, there are several other Elder Scrolls Quest titles without commas that I think would make fantastic essay, book, article, band, or song titles: namely 'Where You Hang Your Enemy's Head', 'Impatience of a Saint', and 'Where Spirits Have Lease'.
To put it simply, no. With further explanation, what people are missing when they ask this is that ChatGPT's core architecture - pure next-token prediction - has zero built-in verification mechanisms. It's fill-in-the-blank pattern-matching, it's fancy Google autocomplete, it's literally designed to make plausible-sounding text, not factual text. For actual business use, or any specialized use case really, you need multi-stage processing pipelines where each AI assertion gets validated against something real with implementation specific confidence thresholds. In some cases, even multiple custom confidence thresholds at different steps in the process flow, because otherwise, you get a monolithic AI system with no way to document where something went wrong. You also need a prompt engineering expert - and no, a role in prompt engineering isn't made-up corporate BS. Let's not forget that SEO experts were laughed at too. If prompting wasn't a skill, prompt injection hacking wouldn't be under the umbrella of cybersecurity.
They're trying. It's not quite working. Conclusion: not worried. Let's discuss a few examples. Metabase had Metabot until they discontinued it as they couldn't get it to reliably work. Uber's database querying AI takes three minutes as opposed to three seconds, and they have self-described it as being 'adequately accurate'. Power BI has Copilot and Tableau has AskData, which are awesome at querying the backend data for the single report you are actively looking at, with no reliable way to customize, so you could be asking about data on one page and get a response about something entirely different. Not to mention you have to hire someone to completely rebuild the schema to get reliable answers on what is contained within it - I would know, I was hired to do that. I feel confident that they will improve as time goes on, but it is also only relevant for companies that only want to use one specific reporting software at a time. I had a VC tell me that if a company had enough money, they could pay Bill Gates to do what we do instead, which I guess could be true, but is really neither here nor there - I'm sure Bill Gates has better things to be doing than recreating technology that we already created. But if anyone has Bill Gates' number, I'd love to pitch the idea. I could go on.
My grandfather used to say that the moment you think of an idea, ten other people around the world are coming up with it at the exact same time. If you have an idea, you need to capitalize on it immediately, even if you just build a simple MVP and a slide deck and start talking to people you think might be interested. AI has made building products and software and social media presence and websites infinitely more accessible to users of any and all technical backgrounds, so there is really no excuse to not go out and start building something. Not to mention, execution creates divergence - even identical starting ideas rapidly evolve into completely different products when exposed to real users. This is the hidden advantage of going first: not just market position, but the head start in the learning cycle that permanently alters your trajectory. We didn't know when we started Analytic Intelligence Solutions that we wanted to be the first full-service AI data management company, we just knew something about AI sounded like the right next step.
That many times when you are looking to create something, it leads to the discovery of entirely new ideas, and you end up creating something entirely different then you intended on building in the first place. When I built Rangler, we were a SaaS B2B company, with a plan to create a custom solution for every company we took on as a client. The idea of an appliance-like product that required no oversight was appealing, but didn't seem achievable with the current technology we had to work with. We never intended for it to become a proprietary product, an improvement from everything else out there on the market, or a step forward for AI Agents as a whole - we just wanted something that would speak to people the way it spoke to us. When I built DataSpace, I was trying to build an open-source gaming site as an example of my coding proficiency, and I didn't want to pay for the existing data storage solutions I needed to do that. Despite my work connecting to Databases with SIMS, I knew very little about the costs of difficulties associated with running one. I certainly wasn't intending to create a solution for people that didn't want to use one - I just wanted to host games without charging users money. It's funny how those things tend to go hand-in-hand.