AI-Powered Accounting for the Mid-Market and Above. The last #ERP you'll ever need. Built in the #AI era
New York, NY
- Tweets 210
- Following 135
- Followers 813
- Likes 69
Pinned Tweet
ERP is like having chronic back pain.
It grows with time. Surgery is risky. Expensive. Could make it worse.
We've fixed that, with our NextDay Migration™
No more nine-month rollouts, 50%+ failure rates, or implementation fees 2–3× the subscription cost.
You're live with your data in 24 hours.
It’s a different way to think about ERP.
And it’s working.
▶️ Catch the full TBPN conversation with DualEntry Co-Founder Santiago Nestares on how we’re rewriting the ERP playbook: piped.video/rdIEDcqvI1E
3
9
4,170
Here's where Terra stands against other leading models:
→ Gemini 4 Argon: 91.1%
→ GLM-5.3 Flash: 88.1%
→ Kimi K3.1 Pro: 86.1%
→ GPT-6.1 Sol: 81.2%
→ GPT-6 Astra: 81.2%
→ GPT-6.1 Terra: 78.2%
2
84
See how the models compare across accuracy, cost, and consistency 👇
dualentry.com/accounting-ai-…
24
OpenAI's new GPT-6.1 Terra flopped on our Accounting AI Benchmark.
It scored just 78.2% accuracy, trailing GPT-6.1 Sol (81.2%), GPT-6 Astra (81.2%), and Gemini 4 Argon (91.1%).
We ran it through 101 accounting tasks, from journal entries and bank reconciliation to accounts payable and month-end close.
Here's what we found 🧵👇
1
1
1
119
Terra passed 79 of 101 accounting tasks, but the results reveal some significant weaknesses.
It scored a perfect 100% on transaction classification, but managed just:
→ 53.8% on accounts payable
→ 58.3% on accounts receivable
→ 58.3% on month-end close
It also passed only 65.3% of tasks consistently across all three runs.
1
37
Moonshot's open Kimi K3.1 Pro just debuted at #4 on our Accounting AI Benchmark with 86.1% accuracy!
That's ahead of Claude Opus 5.5 (82.2%), Claude Fable 5.1 (85.1%) and GPT-6 Astra (81.2%)🤯
5
1
8
404
See how all the models compare on accuracy, cost, and speed 👇
dualentry.com/accounting-ai-…
89
Google just took the #1 spot on the DualEntry Accounting AI Benchmark, with Gemini 4 Argon scoring 91.1%!
That’s 7.9pts higher than Gemini 3.8 Flash and pushes Meta’s Muse Wonder, which ranked #2 when we tested it this week, further down the leaderboard.
Full benchmark: dualentry.com/benchmark
1
3
185
DualEntry retweeted
We just ran Meta’s new Muse Wonder through the DualEntry Accounting AI Benchmark.
It shoots straight to #2 overall, ahead of every OpenAI, Anthropic, and Google model we’ve tested, and 21.8 points higher than Muse Spark 1.1.
1
1
8
2,126
One of @ironselfstorage's favorite features: MCP.
Their team connects DualEntry to Claude, then asks questions about their 30-property portfolio, like:
Where are expenses creeping up?
Which facilities are seeing worrying revenue trends?
Where should the team be paying attention?
@faraz_tx explains how DualEntry has sped up their close:
1
1
1
270
Anthropic’s IPO prospectus reportedly shows a nearly $42B net loss in 2025, alongside an $8.06B operating loss and $7.33B in compute and infrastructure spend.
We broke down what the filing says about the economics of frontier AI.
1
1
2
343
10-day close → 2-day close.
55,000 records migrated in a week.
Two months of parallel close planned. They only needed one.
@getwellrewarded is a healthcare navigation company with contracts complex enough that every amortization schedule used to be built by hand in a spreadsheet. They evaluated NetSuite and Sage, then chose @DualEntry.
The best part isn't the speed. It's what their team does with the time they got back: actually interpreting the numbers instead of hunting for errors in them.
1
3
197
Honored to be featured in @harmonic_ai Hot 25 for Q4 2026 🙌
That makes four consecutive quarters on the list, with DualEntry ranking among Harmonic’s top 15 startups throughout 2026.
We’ve grown the team, shipped a lot of product, and helped more finance teams move off legacy systems and onto a frictionless AI-native ERP.
Proud of how much we’ve built, and excited about what comes next!
1
1
8
256
SF, we’re coming back for Poker Night!
We’re bringing it back for #SFTechWeek, this time with our friends @SlashApp
Join us for an invite-only Poker Night for finance leaders and founders.
📍 Slash HQ, San Francisco
📅 Thursday, October 8
⏰ 6-11 PM
Seats are limited - Register to join us at the table.
Link in comments.
2
212
On Lightwork, we broke down why we started with the general ledger, our reverse-wedge approach to ERP, how we think about faster migrations, and what AI changes for finance teams.
We also got into 'Project Helios', which has since become DualEntry Tasks.
Full episode below! 👇
Before @santinestea started @dualentry, he built his previous company to $100M in revenue across 20 countries.
Now, he's trying to solve the very problem that held his business hostage.
Santiago had prior issues reporting numbers during a nine-month Enterprise Resource Planning (ERP) legacy migration. He knew there could be a better, easier way to migrate, so he started building DualEntry.
This week, Santiago joined Lightwork to recap the moment next-gen ERPs became technically buildable, their fundraising story, and the latest results from DualEntry's "Project Helios".
Full video link in the comments below.
CC: @clairejyz @seisler1
0:00 Welcome to Lightwork
1:41 What Is DualEntry and What Is an ERP
2:20 Why Build a New ERP
3:50 How AI Made Next-Generation ERP Possible
5:44 Leaving a Successful Company
7:58 Finding the Right Way to Tackle ERP
12:23 Which DualEntry Customer Stories Stand Out?
14:03 How Lightseed and DualEntry First Connected
18:12 How DualEntry Has Evolved Since the Partnership
20:40 Inside Project Helios
24:59 Building Fast-Moving Teams at DualEntry
26:09 What Makes a Great DualEntry Hire
30:02 The Future of AI, Fintech, and Organizations
1
9
380