Developer Update #4: Infrastructure upgrades

Over the past few weeks, we’ve been deep in the trenches, rolling out one of the biggest infrastructure upgrades to date for Nexgent. It’s been an intense period of development, testing, and plenty of trial and error - but the outcome sets a strong foundation for the next phase of the platform.

Why We Changed Data Providers

Until recently, Nexgent was relying on high-value trade streams only - focusing on large swaps to reduce infrastructure overhead. It worked for an MVP, but it wasn’t the long-term solution. We’ve now switched to a new data provider that gives us full access to every DEX trade on Solana in real-time - and it’s a total game-changer.

Yes, the new provider is actually cheaper (a win for our bootstrapped budget), but the real reason for the change was data access. We can now see it all: every trade, every second, with near-zero latency.

That means instead of watching ~10 large trades per second, we’re now processing 500 to 1,200 trades per second - over 50 million trades per day. That level of visibility allows us to move faster, analyze deeper, and signal with far more confidence than ever before.

What That Enables

Previously, our signals were strong - but missing key context. Now, we’re capturing the full trading picture. We’re aggregating activity into one-minute buckets and analyzing things like buy/sell ratios, volume surges, and anomaly patterns.

This feeds directly into our surge detection engine and proprietary trading strategies, helping us catch trends early and act with conviction. It’s already showing incredible promise, and it’ll only get sharper as our machine learning models train on this full data context.

How We’re Scaling It

With this much data flying in, we had to rethink the entire backend. Storing every trade in a Postgres database just isn’t viable anymore - not at this scale.

So we moved to a bucket-based aggregation system. Instead of saving every trade individually, we now store time-sliced summaries of trading activity, broken down into relevant data points. This massively improves processing efficiency and unlocks a modular system that’s far more scalable.

It also opens the door for a future we’re really excited about: user-defined trading strategies. With our modular structure, you’ll eventually be able to plug in custom strategies - whether you’re looking for surge plays, volume shifts, reversals, or even arbitrage - all powered by the same core data engine.

We’re not there yet, but that’s the vision. A truly programmable AI trading layer.

Early Results Are Promising

We’ve already spun up a few new strategies using the updated infrastructure, and the early results are looking great:

• Much higher signal accuracy

• Stronger win rates

• More context-aware decisions

We’re still experimenting with stop-loss models and refining our approach based on the type of trade (quick snipe vs. longer hold vs. opportunistic arb), but the toolkit we’re building is getting sharper by the day.

What's Next

We’re heads-down continuing to build - but once we lock in more strategy modules and finish stress testing the new infra, we’ll reopen access for early testers. We can’t wait to get feedback on how this performs in the real world.

Final Thoughts

Yes, this upgrade delayed our timeline a bit. But we’re not here to rush something out the door. We’re here to build a reliable, high-performance platform that lasts - and sometimes that means investing in the hard stuff early.

We’re incredibly thankful to the community that’s stuck with us while we’ve been heads down. We believe this new phase will set Nexgent apart - and we can’t wait to show you what’s next.

— The Nexgent Team