Precisely's agent library ships now, ahead of the no-migration data platform due in early 2027

A developer who signs up for the AI Studio free trial on Monday can be prototyping within minutes, according to Precisely ^[16][11]. Several of the starter assets show address geocoding and data replication pipelines. Others build on that data, including a property analyzer and a real estate intelligence agent ^[12]. They work inside Claude, Microsoft Copilot and ChatGPT ^[11].

She cannot yet put into production the part of the launch that deals with mainframe and SAP data. The Precisely Platform manages that data alongside cloud lakehouses without migrating it, and it is in preview, with general availability due in early 2027 ^[3][4][15]. Stewart Bond, a vice president at IDC, said many organizations "continue to struggle with fragmented data environments" ^[17].

Matt Waxman, Precisely's chief product officer, made the case for the platform half. "Agents don't stop to question the data they're given," he said. "They act on it at scale, far faster than people can catch and correct errors." ^[8] SiliconANGLE described the risk the same way: when agents act on data with no person in between, an error a human reviewer could have spotted may turn into an automated decision before anyone intervenes ^[19]. Precisely's protections against that are the shared rule layer and the per-asset scores. Both belong to the preview platform ^[5][7][15].

Here's what teams tell themselves: a pilot of AI Studio is a pilot of Precisely's governed, no-migration approach. Here's what users actually do in that pilot. They run ready-made agents against data, which in the published examples means geocoding and property analysis, while the rule layer Waxman describes is still in preview ^[12][15]. Waxman himself said a convincing demo is the easy part compared with "building AI you can trust at scale" ^[13].

The one outcome figure in the launch comes from a partner. At Korem, a Canadian geospatial firm, time to value on customer projects "used to be calculated in months," said Jimmy Duchesne, its director of solutions innovation and presales. He said it can now be measured in days ^[14]. Time to value is the right figure to ask direct customers for once the platform ships. The report does not include pricing for either product, or say which of the two Korem's projects used ^[14][16].

To decide what to run before 2027, sort each use case on two axes. One is where the data sits: a cloud lakehouse, or mainframe and SAP systems. The other is who acts on the output: a person who reviews it, or the agent alone ^[3][19]. If a person reviews the output, the AI Studio trial is a cheap way to learn on either kind of data, because someone can still catch the errors Waxman describes ^[8][16]. An agent acting alone on lakehouse data belongs in the preview, if your team can get access, so you can see what quality and governance scores the platform gives your own assets before you trust them ^[7][15]. For agents acting alone on mainframe or SAP records, I'd wait for general availability and keep a person in the loop until then. Waiting costs you the months-to-days gain Korem reports, assuming it carries over to your data ^[14].