DEV Community: Sid Probstein

The AI agent stack that’s quietly taking over enterprise workflows

Sid Probstein — Sat, 03 May 2025 02:58:27 +0000

Accenture, IBM, and AWS are all placing bets on Crew AI. Why? Because it makes building and deploying real AI agents possible.

With Crew AI, teams are spinning up agents that:

Launch predictive marketing campaigns
Automate financial back-office ops
Optimize inventory and logistics
And tackle 100+ other enterprise use cases

But here’s the catch: agents are only as good as the data they can reach. That’s where SWIRL comes in.

By pairing Crew AI with SWIRL, you get more than just agents—you get enterprise-ready, data-rich workflows that scale. No custom plumbing. No brittle integrations.

With Crew AI + SWIRL, your agents can:

Connect to 100+ enterprise data sources out-of-the-box
Fetch the most relevant structured/unstructured data across silos
Respect row-level permissions with real enterprise auth
Summarize and answer with your LLM of choice
Plug in easily via zero-code connectors

Want to see this in action?

Message me for a demo or check the open source edition here: https://github.com/swirlai/swirl-search

It was already indexed!

Sid Probstein — Sun, 13 Aug 2023 16:48:52 +0000

I recently had the pleasure of chatting with @dmitrykan on his Vector Podcast. Check it out: https://dmitry-kan.medium.com/vector-podcast-with-sid-probstein-search-in-siloed-data-with-swirl-f2b9595a2715

We talked about quite a few things, including:

The challenges of enterprise search in the post-cloud era
How cross-silo search is particularlyt tricky because of entitlements (aka permissions) across silos
Zero-code configuration of connectors in Swirl, where JSON path and developer API doc get the job done
How large language models contextually re-rank disparate search results

There was an interesting twist at the end of the call. Dmitry uses a service called Clearword to transcribe recordings. Dmitry asked: “how quickly can you index the transcript and search it with Swirl?”

Here is my answer: It was already indexed!

Since there is no audio and it goes by quickly, let me explain ... Clearword emailed the transcript to both of us shortly after recording ended. It was indexed by Microsoft Outlook within seconds of arriving in my inbox.

To verify this, I copied some text from the middle of the transcript and pasted it into Swirl, which returned the link to the email message with the transcript and the phrase I searched for.

That simple truth - that the average enterprise is awash in search forms - is the entire reason metasearch is such a game-changing approach. Instead of making yet another repository, Swirl sends queries to existing search APIs and re-ranks the results from everything. It saves users a huge amount of time without a major IT project.

Want to see for yourself? git it going with 2 commands via Docker here: https://github.com/swirlai/swirl-search

Swirl 2.5 released

Sid Probstein — Wed, 09 Aug 2023 14:37:03 +0000

I am delighted to announce availability of Swirl 2.5!

This version focused on performance. Configured with 12 SearchProviders, Swirl 2.5 supports ~15 queries/second on a Standard F16s v2 server (16 vcpus, 32 GiB memory) with a median response time of ~3 seconds.

Version 2.5 also includes SearchProviders for HubSpot contact, company, and deal records, plus improvements to the Galaxy search UI (shown above).

Check out the Release notes for full details: https://github.com/swirlai/swirl-search/releases/tag/v2.5.0

What is Swirl? A new open source metasearch engine; it queries anything with an API then uses spaCy to re-rank the unified results without copying any data! Includes zero-code configs for Apache Solr, ChatGPT, Elastic Search, OpenSearch, PostgreSQL, Google BigQuery, RequestsGet, Google PSE, NLResearch.com, Miro, Microsoft 365, HubSpot, Atlassian, YouTrack, GitHub & more!

I wrote a metasearch engine called Swirl

Sid Probstein — Thu, 03 Aug 2023 18:48:40 +0000

Hi all! Swirl sends queries to existing search engines, unifies the results and re-ranks them all using large language models. It solves cross-silo information access and search problems in a fraction of the time and effort required to copy, ingest and index data.

Here's a brief video intro: https://youtu.be/sfsBYyu6qDQ

Swirl was written in python atop the django/celery/redis stack with a choice of Sqlite3 or PostgreSQL back-ends. The source code is available under the Apache 2.0 license. The distribution includes zero-code configs for Apache Solr, ChatGPT, Elastic Search, OpenSearch, PostgreSQL, Google BigQuery, RequestsGet, Google PSE, NLResearch.com, Miro, Microsoft 365, Atlassian, YouTrack, GitHub, HubSpot & more. Plug-in your access tokens or use Microsoft 365 to login and users can stop searching and start Swirling!

Links:

We are seeking contributions and feedback from developers working on all kinds of search solutions... thanks!