Vibe Excelling: Testing Elkar’s AI Add-On for Excel

A rep from Elkar — a small startup bringing AI into Excel — reached out with a review request. For transparency: I only used their regular free trial and nothing beyond that.

I’ve already done a couple of pieces about trying to do day to day finance tasks with AI in which I got mixed to disappointing results – but this time the experience was very different. I was very impressed with Elkar and it was successful where my other experiments were not.

Elkar integrates directly into Excel as an add-on that sits right alongside Microsoft’s Co-Pilot. Click it, and a chat window opens up where you can type your request, see how the AI approaches the task, and watch it populate your spreadsheet. You can choose the AI backbone (Sonnet, Gemini and GPT). I went with the default which was Sonnet 4 Thinking.

First Impressions

I started simple, using one of their demo prompts: a loan amortization table. Elkar handled it smoothly — formulas, charts, and all. I then asked it to build a fixed-principal amortization schedule. Again, it performed well.

Consolidation Test (Where ChatGPT Stumbled)

If you’ve read my earlier post on testing AI consolidations, you’ll know it failed to handle balance sheet eliminations properly.

Elkar, however, nailed it. On the P&L, it produced cleaner results than ChatGPT, and when I asked for a consolidated balance sheet, it correctly eliminated both the investment in subsidiary and the subsidiary’s equity. See in the image below how it correctly understood the need to these consolidation tasks without mentioning in the prompt the need to do that. And the output has full formulas!

That alone is a big step up. It even did a consolidation work paper for me!

Consolidation work paper created by Elkar - see on the right side how it understands it needs to eliminate "investment in company d" and shareholder equity of the subsidiary. And in E11 you can see the formula used.

Tackling a Messy P&L

Next, I gave Elkar a 200+ row detailed P&L and asked it to convert it into a clean, classic-style format. Within minutes, it was done.

For the most past the detailed P&L had summary lines that are easy to connect to a summary but there was one “curve ball”  in which the detailed P&L had different sections for other income and other expenses but on the summary we had a net other income (expense presentation). Elkar picked up on that and correctly offset the lines. That’s the kind of accounting nuance I didn’t see in my trials with the regular AI chat.

Elkar made P&L after I asked to add change % column

Projections, Pivots, and Beyond

After testing consolidations and P&Ls, I moved on to other common finance tasks:

Basic Projections I fed Elkar a few assumptions and asked it to build out a projection. It handled this cleanly and quickly — no real surprises, but a useful time-saver.

Pivot Table on 1,000+ Inventory Items Elkar generated a pivot table without issue. The output was accurate, though the process seemed more complicated than necessary. Still, it got the job done.

Internet research and Stock Prices & Charting This was the most ambitious test as it’s based on outside info and not information I’ve provided. I asked for a last 12-month stock price history with a chart. Here, Elkar stumbled with wrong date ranges and miscalculated returns. I’ve see this internet research issues with the main chats as well so perhaps it’s part of the same limitations I’ve seen with AI: formatting quirks, date logic, hallucinations and occasional math inconsistencies.

My Take

Overall, I walked away very impressed. Elkar outperformed my expectations, especially compared to the experience I had with Gemini, Perplexity and Chat GPT. The balance sheet consolidation logic was the most impressive win. It’s also great to work inside Excel and see all the formulas so you can track what the AI did.

The Takeaway for Finance Teams

AI in Excel isn’t just about novelty — it’s about cutting down the time spent on repetitive modeling, formatting, and reconciliations. Elkar shows what’s possible when AI is embedded directly into the finance team’s most familiar tool. At $40/month I think it’s a winner.